Development and validation of a machine learning model for MASLD risk prediction in lowlanders with long-term high-altitude exposure

Development and validation of a machine learning model for MASLD risk prediction in lowlanders with long-term high-altitude exposure | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Development and validation of a machine learning model for MASLD risk prediction in lowlanders with long-term high-altitude exposure Yongjiang Zhou, Youqing Huang, Jian Feng, Qingqing Wang, Hanyu Ding, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9109584/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 9 You are reading this latest preprint version Abstract Background The impact of long-term high-altitude exposure on metabolic dysfunction–associated steatotic liver disease (MASLD) risk in lowlanders remains poorly defined. Given the scarcity of medical resources in high-altitude regions, a practical tool for early identification of MASLD is critical. This study aimed to develop and validate a MASLD risk prediction model tailored for lowlanders with long-term high-altitude exposure using routine clinical indicators. Methods This retrospective study analyzed 663 lowlanders with long-term high-altitude exposure from July 2022 to June 2023. Missing data were imputed using the MissForest algorithm. A rigorous feature selection strategy comprising univariate analysis, LASSO regression, and the Boruta algorithm was employed. Six machine learning models—Logistic Regression (LR), Random Forest (RF), Decision Tree (DT), XGBoost, Support Vector Machine (SVM), and LightGBM—were trained and evaluated using 10-fold cross-validation. Performance was assessed via AUC, accuracy, sensitivity, specificity, F1 score, calibration curves, and Decision Curve Analysis (DCA). Model interpretability was quantified using SHapley Additive exPlanations (SHAP). A nomogram and an online dynamic calculator were developed based on the optimal model. Results Nine predictors were identified: altitude, systolic blood pressure (SBP), body mass index (BMI), white blood cell count (WBC), red blood cell count (RBC), hemoglobin (HGB), hepatic steatosis index (HSI), triglycerides (TG), and high-density lipoprotein cholesterol (HDL-C). Among the six models, Logistic Regression. SHAP analysis further elucidated the model's interpretability, ranking BMI as the primary risk driver followed by TG and SBP. The nomogram and web-based calculator demonstrated good calibration and provided net clinical benefit across clinically relevant thresholds. Conclusions We developed and internally validated a machine learning–based model to predict MASLD risk in lowlanders with long-term high-altitude exposure. The online calculator ( https://altitude-masld-predictor.shinyapps.io/dynnomapp ) may facilitate early screening and individualized risk stratification in resource-limited high-altitude settings. Metabolic dysfunction–associated steatotic liver disease MASLD High-altitude exposure Machine learning Risk prediction model Nomogram Military health Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1. Introduction Metabolic dysfunction–associated steatotic liver disease (MASLD) is a recently adopted term that has replaced non-alcoholic fatty liver disease (NAFLD), emphasizing the central role of metabolic dysregulation in disease initiation and progression[ 1 , 2 ]. MASLD is now the most prevalent chronic liver disease worldwide, affecting more than 30% of adults, and its prevalence continues to increase [ 3 , 4 ]. The disease spectrum ranges from simple hepatic steatosis to metabolic dysfunction–associated steatohepatitis (MASH), progressive fibrosis, and ultimately cirrhosis and hepatocellular carcinoma [ 5 , 6 ]. As a multisystem disorder, MASLD is also associated with elevated risks of cardiovascular disease, chronic kidney disease, and all-cause mortality [ 7 ]. Because MASLD often develops silently and lacks specific symptoms in the early stage, it poses an escalating burden on public health systems. Therefore, establishing effective strategies for early identification and risk prediction has substantial clinical and societal importance. In recent years, the expansion of mountaineering, infrastructure development, mineral resource exploitation, military operations, and renewable energy projects has markedly increased the number of lowlanders who reside or work in hypoxic high-altitude environments for prolonged periods [ 8 – 10 ]. During the construction of the Qinghai–Tibet Railway alone, more than 100,000 workers were stationed long-term at altitudes of 2,800–5,200 m [ 11 ]. Similarly, routine military training on the western plateau and the rapid expansion of wind and photovoltaic industries in Qinghai, Tibet, and Sichuan have exposed large populations of non-indigenous individuals to chronic hypoxia and acclimatization-related metabolic stress [ 12 ]. High-altitude tourism further contributes; approximately three million tourists from lowland regions enter Tibet annually, and more than 70% of itineraries exceed 10 days, suggesting that even short- to medium-term exposure may perturb metabolic homeostasis [ 13 , 14 ]. Although epidemiological data indicate that approximately 160 million people worldwide permanently reside at altitudes above 1,500 m [ 15 ], evidence regarding MASLD risk among lowlanders who migrate to or are stationed at high altitudes remains limited. Moreover, the generalizability of existing lowland-derived risk models to this population has not been established, underscoring the need for targeted risk assessment in long-term high-altitude lowlanders. The distinctive high-altitude environment—characterized by hypobaric hypoxia and cold exposure—can profoundly reshape metabolic homeostasis [ 16 ]. Prior studies suggest that chronic hypoxia may promote hepatic lipid accumulation and fibrosis by modulating oxygen-sensing pathways (e.g., the HIF signaling axis) and triggering inflammation and oxidative stress [ 17 , 18 ]. However, the impact of hypoxia on hepatic metabolism remains controversial. Some animal studies indicate that hypoxic exposure may improve mitochondrial function and alleviate steatosis [ 19 ], whereas others suggest that it exacerbates metabolic disorders [ 20 ]. The limited and inconsistent evidence linking altitude exposure to MASLD risk constrains our understanding of susceptibility mechanisms and the geographical epidemiology of MASLD in this unique setting, highlighting a critical gap in risk stratification and early screening for high-altitude-exposed lowlanders. Liver biopsy remains the gold standard for diagnosing and staging MASLD and provides direct histological evidence of inflammation and fibrosis [ 21 , 22 ]; however, its invasiveness and associated risks limit its suitability for large-scale screening and longitudinal follow-up [ 23 ]. Accordingly, non-invasive prediction models have become a major focus. Traditional statistical approaches are often constrained when handling high-dimensional clinical variables and complex non-linear relationships. In contrast, machine learning (ML) offers advantages in feature selection, pattern recognition, and risk stratification and has been increasingly applied to metabolic disease prediction. Recent studies have shown that ML-based MASLD prediction models can improve screening efficiency and diagnostic accuracy. For example, Deng et al. developed a model incorporating age, BMI, triglycerides, fasting blood glucose, waist circumference, and HDL-C with an AUC of 0.862 [ 24 ]; Masaebi et al. proposed a model combining questionnaire data, blood tests, and liver ultrasound to inform prevention and management [ 25 ]; and an additional study in a diabetic population reported an 8-parameter ML model with an AUC of 0.84 [ 26 ]. Despite the growing body of MASLD prediction research, models tailored to lowlanders with long-term high-altitude exposure are still lacking. Because high-altitude acclimatization induces substantial physiological adaptations, such as compensatory erythrocytosis and blood pressure fluctuations, directly applying models derived from lowland populations may introduce predictive bias. Therefore, this study aimed to develop and validate a machine learning model to predict MASLD risk in lowlanders with long-term high-altitude exposure using demographic characteristics, lifestyle factors, and routine clinical indicators. By identifying key predictors, we sought to clarify how the high-altitude environment may influence metabolic homeostasis and to provide a low-cost, clinically practical tool to support early screening and individualized health management in resource-limited high-altitude regions, thereby addressing an important gap in high-altitude environmental medicine. 2 Materials and Methods 2.1 Study Population This retrospective study analyzed demographic and clinical data from lowlanders with long-term high-altitude exposure who attended the General Hospital of Western Theater Command between July 2022 and June 2023. A total of 834 individuals were initially screened. According to predefined inclusion and exclusion criteria ( Fig. 1 ) , 663 participants were included in the final analysis. The inclusion criteria were: (a) age ≥ 18 years; (b) exposure to altitudes > 2,500 m; and (c) duration of high-altitude exposure > 1 year. Participants were excluded for the following reasons: incomplete clinical data (n = 43), exposure to altitudes < 2,500 m (n = 21), high-altitude exposure < 1 year (n = 15), other liver diseases (n = 24), and excessive alcohol consumption (n = 68). Given the retrospective design, written informed consent was waived by the institutional ethics committee. All personal identifiers were anonymized prior to analysis to ensure confidentiality and compliance with research ethics. 2.2 Data Collection Baseline information was collected prior to physical examination by trained medical personnel using standardized questionnaires and interviews, including demographic characteristics (age, gender, education level), lifestyle factors (smoking, alcohol consumption, tea drinking, dietary patterns, and physical activity type and intensity), and environmental exposure history (residence location, altitude, and duration of exposure). Physical examinations included measurements of height, weight, systolic blood pressure (SBP), and diastolic blood pressure (DBP). Altitude was determined based on the specific residence/duty location recorded in the medical examination data and corresponding geographic altitude information. For participants with multiple postings, altitude and exposure duration were defined according to the primary location (longest duration). Blood pressure was assessed using standardized automated upper-arm sphygmomanometers to ensure consistency and reproducibility. Venous blood samples were collected immediately upon admission for a comprehensive laboratory analysis, categorized as follows: (1) hematology, including white blood cell (WBC) count, red blood cell (RBC) count, neutrophil count(NEUT), lymphocyte count(LYMPH), monocyte count(MONO), hemoglobin (HGB), and platelet count(PLT); (2) liver function, comprising alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin (TBIL), direct bilirubin (DBIL), indirect bilirubin (IBIL), total protein (TP), albumin (ALB), globulin (GLB), and the albumin/globulin (A/G) ratio; (3) lipid profile, consisting of total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C); (4) renal function, assessed via urea and creatinine (Cr); (5) coagulation profile, including prothrombin time (PT), prothrombin activity (PTA), international normalized ratio (INR), fibrinogen (FIB), activated partial thromboplastin time (APTT), thrombin time (TT), and D-dimer; and (6) glycemic control, measured by glycated hemoglobin (HbA1c). Additionally, abdominal ultrasonography was performed by the same medical team using standardized equipment. 2.3 Patient Grouping In accordance with the MASLD definition proposed by the 2023 international multi-society Delphi consensus, hepatic steatosis was diagnosed by senior physicians using abdominal ultrasonography. Patients were diagnosed with MASLD if they exhibited hepatic steatosis and met at least one of the following cardiometabolic criteria [ 27 ]: (a) overweight/obesity (BMI ≥ 23 kg/m² for Asians); (b) type 2 diabetes; (c) blood pressure ≥ 130/85 mmHg or receiving antihypertensive treatment; (d) fasting TG ≥ 150 mg/dL (1.7 mmol/L); (e) HDL-C < 40 mg/dL (1.0 mmol/L) for men or < 50 mg/dL (1.3 mmol/L) for women; or (f) HbA1c ≥ 5.7%. Fasting plasma glucose was not measured in this study; therefore, glycemic status was assessed using documented type 2 diabetes and/or HbA1c. Exclusion criteria included excessive alcohol consumption (> 30 g/day for men, > 20 g/day for women) and other liver diseases (e.g., viral hepatitis, drug-induced liver injury, autoimmune liver disease). Participants without hepatic steatosis on abdominal ultrasonography served as controls. 2.4 Machine Learning 2.4.1 Data Processing Variables with > 20% missingness were excluded to reduce potential bias [ 28 ]. To assess generalizability and minimize information leakage, the dataset was randomly split into a training set (70%) and a validation set (30%) using a fixed random seed (123). Missing values were then imputed after splitting using the random forest–based MissForest algorithm, yielding complete datasets for subsequent modeling [ 29 ]. 2.4.2 Feature Selection Feature selection was performed using the training set only. Variables were screened sequentially using univariate analysis, least absolute shrinkage and selection operator (LASSO) regression, and the Boruta algorithm. In univariate analyses, variables with P < 0.05 were considered candidate predictors. LASSO regression was implemented using the glmnet package (α = 1), and 10-fold cross-validation was used to select the optimal regularization parameter (lambda.1se). Predictors with non-zero coefficients at the selected λ were retained [ 30 ]. Ultimately, nine variables consistently supported across these procedures were selected for model development. 2.4.3 Model Construction and Evaluation Model development was conducted in Python 3.10.4 using the nine selected features. Six algorithms were evaluated: logistic regression (LR), random forest (RF), decision tree (DT), support vector machine (SVM), extreme gradient boosting (XGBoost), and light gradient boosting machine (LightGBM) [ 31 ]. All models were implemented using scikit-learn. Hyperparameters were optimized by grid search with 10-fold stratified cross-validation on the training set (random seed = 123), targeting the area under the receiver operating characteristic curve (AUC). The final tuned models were evaluated in the held-out validation set. Performance metrics included AUC, accuracy, sensitivity, specificity, F1 score, Cohen’s kappa, positive predictive value (PPV), and negative predictive value (NPV). Sensitivity, specificity, and F1 score were calculated using a default probability threshold of 0.5. Calibration and clinical utility were assessed using calibration curves and decision curve analysis (DCA). The optimal model was selected based on overall discrimination, calibration, and clinical utility. 2.4.4 Model Interpretation Interpretability remains a challenge in clinical machine learning. To address the "black box" issue, we utilized Shapley Additive exPlanations (SHAP) based on game theory. SHAP quantifies the marginal contribution of each variable, enabling both global assessment of variable importance and local explanation of individual predictions [ 32 ]. Although primarily designed for tree-based models, the LinearExplainer method extends SHAP to linear models like LR [ 33 ]. This approach provided variable importance rankings and elucidated the direction and magnitude of each feature's effect, enhancing the transparency and clinical trust in the model. 2.4.5 Nomogram Construction To enhance clinical applicability, a nomogram was constructed based on the final multivariate LR model. This tool, widely used in predictive modeling [ 34 ], visualizes individual risk. Performance was validated using calibration curves and 1,000 bootstrap resamples to minimize overfitting [ 35 ]. The nomogram assigns scores to variables to calculate MASLD probability, facilitating rapid risk stratification. This low-cost, easy-to-use tool is particularly valuable for resource-limited high-altitude regions, aiding frontline medical personnel in optimizing resource allocation [ 36 ]. 2.5 Statistical Analysis Statistical analyses were performed using R software (version 4.4.3). Normally distributed continuous variables were expressed as mean ± standard deviation (SD), while non-normally distributed variables were reported as median (interquartile range) [M (IQR)]. Categorical variables were presented as frequencies and percentages [n (%)]. Between-group comparisons employed Student's t-test, Mann-Whitney U test, or χ² test as appropriate. All hypothesis tests were two-tailed with statistical significance defined as α = 0.05. 3.Results 3.1 Baseline Characteristics This study screened 834 lowlanders with long-term high-altitude exposure who attended the General Hospital of Western Theater Command. According to predefined exclusion criteria (Fig. 1 ), 171 individuals were excluded, including incomplete clinical data (n = 43), exposure to altitudes < 2,500 m (n = 21), high-altitude exposure < 1 year (n = 15), other liver diseases (n = 24), and excessive alcohol consumption (n = 68), leaving 663 eligible participants for analysis. The median age was 50.0 years (IQR: 45.0–54.0), and 634 participants were male (95.6%). Long-term smoking was reported by 312 subjects (47.06%); 330 (49.77%) had hypertension, and 128 (19.31%) had diabetes. The mean residence altitude was 3,700 m, with a mean exposure duration of 204 months. Overall, 248 participants (37.41%) were stationed at 2,500–3,500 m and 415 (62.59%) at ≥ 3,500 m. Exposure duration was 2–5 years in 93 individuals (14.03%) and > 5 years in 570 (85.97%). Physical activity was classified as low, moderate, and high in 216 (32.58%), 160 (24.13%), and 287 (43.29%) participants, respectively, and 356 (53.70%) reported habitual tea consumption. Compared with the non-MASLD group, the MASLD group differed significantly across multiple variables (P < 0.05; Table 1 ), including general characteristics (BMI, SBP, DBP, and altitude), hematological indices (WBC, RBC, HGB, and leukocyte subsets), and metabolic/liver-related markers (TG, HDL-C, HSI, ALT, and AST). Table 1 Baseline characteristics of participants in this study. Continuous values were presented as median [interquartile range]. Categorical valueswere presented as numbers (percentages). Variables Total(n = 663) Non-MAFLD(n = 373) MAFLD(n = 290) p Gender 0.018 Female 29 ( 4.37%) 23 ( 6.17%) 6 ( 2.07%) Male 634 (95.63%) 350 (93.83%) 284 (97.93%) Smoking 0.527 No 351 (52.94%) 202 (54.16%) 149 (51.38%) Yes 312 (47.06%) 171 (45.84%) 141 (48.62%) HTN 0.335 No 333 (50.23%) 194 (52.01%) 139 (47.93%) Yes 330 (49.77%) 179 (47.99%) 151 (52.07%) DM 0.969 No 532 (80.24%) 300 (80.43%) 232 (80.00%) Yes 131 (19.76%) 73 (19.57%) 58 (20.00%) AC 0.001 HA 248 (37.41%) 160 (42.90%) 88 (30.34%) VHA 415 (62.59%) 213 (57.10%) 202 (69.66%) ED 0.021 STE 3 ( 0.45) 2 ( 0.54) 1 ( 0.34) MTE 87 (13.12%) 37 ( 9.92) 50 (17.24%) LTE 573 (86.43%) 334 (89.54%) 239 (82.41%) PA 0.094 LPA 215 (32.43%) 132 (35.39%) 83 (28.62%) MPA 167 (25.19%) 84 (22.52%) 83 (28.62%) HPA 281 (42.38%) 157 (42.09%) 124 (42.76%) Dietary 0.871 LFD 254 (38.31%) 141 (37.80%) 113 (38.97%) HFD 134 (20.21%) 74 (19.84%) 60 (20.69%) Other 275 (41.48%) 158 (42.36%) 117 (40.34%) Tea 0.009 No 311 (46.91%) 192 (51.47%) 119 (41.03%) Yes 352 (53.09%) 181 (48.53%) 171 (58.97%) Altitude 3700.00 [3000.00,3700.00] 3600.00 [3000.00,3700.00] 3700.00 [3200.00,3800.00] < 0.001 M 204.00 [144.00,240.00] 206.00 [160.00,240.00] 202.50 [120.00,240.00] 0.162 Age 50.00 [45.00,54.00] 50.00 [45.00,54.00] 50.00 [46.00,55.00] 0.687 SBP 123.00 [111.00,134.50] 116.00 [108.00,126.00] 131.00 [120.00,140.00] < 0.001 DBP 74.00 [67.00,82.00] 72.00 [66.00,79.00] 77.00 [70.00,84.00] < 0.001 BMI 24.83 (2.66) 23.45 (2.21) 26.61 (2.08) < 0.001 WBC 5.77 [4.90,6.69] 5.29 [4.52,6.22] 6.35 [5.56,7.40] < 0.001 RBC 5.13 (0.39) 4.99 (0.36) 5.30 (0.36) < 0.001 NEUT 3.32 [2.71,4.03] 3.15 [2.58,3.88] 3.54 [2.93,4.20] < 0.001 LYMPH 1.66 [1.35,2.02] 1.61 [1.32,1.93] 1.72 [1.42,2.12] < 0.001 MONO 0.33 [0.27,0.40] 0.32 [0.26,0.39] 0.34 [0.28,0.43] 0.001 HGB 155.82 (10.65) 152.89 (10.06) 159.59 (10.21) < 0.001 PLT 183.00 [152.00,219.00] 183.00 [152.00,215.00] 184.00 [152.00,222.00] 0.466 Urea 5.25 [4.50,6.07] 5.25 [4.48,6.04] 5.26 [4.52,6.14] 0.554 Cr 78.30 (10.53) 79.14 (10.31) 77.23 (10.72) 0.02 ALT 24.40 [17.70,34.65] 20.70 [15.70,27.40] 30.30 [22.33,41.68] < 0.001 AST 22.50 [19.05,27.60] 21.50 [18.80,25.60] 24.70 [20.00,29.90] < 0.001 ALT/AST 1.08 [0.85,1.37] 0.97 [0.78,1.19] 1.29 [1.03,1.55] < 0.001 HSI 34.11 [30.54,37.02] 31.81 [29.17,34.68] 36.72 [34.24,39.25] < 0.001 TC 4.81 (0.81) 4.78 (0.80) 4.85 (0.84) 0.267 TG 1.62 [1.15,2.32] 1.30 [0.92,1.75] 2.22 [1.59,2.91] < 0.001 HDL-C 1.32 [1.13,1.54] 1.48 [1.28,1.65] 1.16 [1.04,1.30] < 0.001 LDL-C 3.05 (0.66) 2.97 (0.66) 3.15 (0.64) < 0.001 TP 73.62 (3.76) 73.25 (3.79) 74.11 (3.67) 0.003 ALB 47.27 (2.12) 47.02 (2.12) 47.59 (2.09) 0.001 GLOB 26.20 [24.30,28.35] 26.20 [24.20,28.20] 26.30 [24.30,28.58] 0.421 A/G 1.81 [1.66,1.96] 1.81 [1.66,1.96] 1.81 [1.66,1.97] 0.593 TBIL 18.60 [15.05,23.15] 19.10 [15.20,23.50] 17.90 [14.83,22.50] 0.105 DBIL 4.50 [3.70,5.40] 4.60 [3.70,5.50] 4.30 [3.60,5.10] 0.015 IBIL 14.00 [11.40,18.00] 14.40 [11.50,18.10] 13.65 [11.30,17.60] 0.161 PT 10.70 [10.40,10.95] 10.80 [10.50,11.00] 10.60 [10.40,10.80] < 0.001 PTA 105.30 [102.35,109.00] 104.10 [101.80,107.70] 106.50 [104.10,109.00] < 0.001 INR 0.92 [0.89,0.94] 0.93 [0.90,0.95] 0.91 [0.89,0.93] < 0.001 FIB 2.59 [2.30,2.98] 2.55 [2.27,2.93] 2.59 [2.33,2.98] 0.123 APTT 26.88 (1.61) 27.08 (1.62) 26.64 (1.56) < 0.001 TT 16.52 (0.77) 16.53 (0.78) 16.51 (0.75) 0.696 DD 0.11 [0.08,0.19] 0.11 [0.08,0.19] 0.11 [0.08,0.21] 0.337 HbA1c 5.80 [5.50,6.00] 5.70 [5.50,6.00] 5.80 [5.60,6.10] 0.001 3.2 Data Partitioning and Feature Extraction The dataset was randomly split into a training set (n = 465) and a validation set (n = 198) at a 7:3 ratio.To identify predictors closely associated with MASLD, a three-stage screening strategy comprising univariate analysis, LASSO regression, and the Boruta algorithm was employed in the training set. In univariate analyses, 27 variables were significantly associated with MASLD (P < 0.05), including gender, physical activity, tea consumption, altitude, SBP, DBP, BMI, WBC, RBC, and NEUT ( Supplementary Table 1 ). Boruta further evaluated feature relevance and identified 17 “confirmed important” variables (Fig. 2 A–B). LASSO regression selected 12 predictors with non-zero coefficients at the optimal penalty parameter (λ = 0.0067) (Fig. 2 C–D), including altitude, SBP, BMI, WBC, RBC, HGB, HSI, TG, HDL-C, ALB, APTT, and tea consumption. Intersecting the three selection methods yielded nine stable predictors: altitude, SBP, BMI, WBC, RBC, HGB, HSI, TG, and HDL-C (Fig. 2 E), which were used for subsequent model development. 3.3 Model Construction and Evaluation Using the training set, we developed and compared six machine learning models (LR, XGBoost, LightGBM, RF, SVM, and DT). After hyperparameter optimization by grid search with 10-fold stratified cross-validation, model performance metrics were calculated (Tables 2 – 3 ). LR achieved the highest AUC in the validation set (AUC = 0.898) with an accuracy of 0.828, sensitivity of 0.816, specificity of 0.838, and F1 score of 0.807 (probability threshold = 0.5). ROC analysis showed good discrimination (Fig. 3 A–B). Calibration curves indicated acceptable agreement between predicted and observed risks (Fig. 3 C), and decision curve analysis demonstrated net clinical benefit across a range of threshold probabilities (Fig. 3 D). Given its competitive performance and interpretability, LR was selected as the final model. Table 2 Performance of each algorithm in the training set cohort. Model AUC Accuracy Precision Sensitivity Specificity F1 Score Kappa Youden's J PPV NPV LR 0.920(0.895–0.943) 0.830 0.795 0.823 0.836 0.809 0.656 0.659 0.795 0.859 DT 0.894(0.865–0.992) 0.815 0.745 0.877 0.767 0.805 0.631 0.644 0.745 0.889 RF 0.986(0.977–0.993) 0.933 0.922 0.926 0.939 0.924 0.865 0.865 0.922 0.943 XGBoost 0.972(0.959–0.983) 0.899 0.868 0.906 0.893 0.887 0.796 0.800 0.868 0.925 LightGBM 0.976(0.965–0.986) 0.918 0.915 0.897 0.935 0.905 0.834 0.832 0.915 0.921 SVM 0.912(0.885–0.936) 0.839 0.811 0.823 0.851 0.817 0.673 0.674 0.811 0.861 Table 3 Performance of each algorithm in the validation set cohort. Model AUC Accuracy Precision Sensitivity Specificity F1 Score Kappa Youden's J PPV NPV LR 0.898(0.856–0.939) 0.828 0.798 0.816 0.838 0.807 0.652 0.654 0.798 0.853 DT 0.815(0.759–0.876) 0.758 0.701 0.782 0.739 0.739 0.514 0.520 0.701 0.812 RF 0.878(0.832–0.923) 0.788 0.737 0.805 0.775 0.769 0.574 0.579 0.737 0.835 XGBoost 0.878(0.832–0.922) 0.798 0.783 0.747 0.838 0.765 0.588 0.585 0.783 0.809 LightGBM 0.870(0.821.0.917) 0.773 0.744 0.736 0.802 0.740 0.538 0.537 0.744 0.795 SVM 0.896(0.854–0.936) 0.808 0.769 0.805 0.811 0.787 0.612 0.615 0.769 0.841 3.4 Model Interpretation SHAP was applied to interpret the final LR model. Global SHAP importance ranked the nine predictors (highest to lowest) as BMI, TG, SBP, HDL-C, HSI, RBC, WBC, HGB, and altitude ( Fig. 4A ). The SHAP beeswarm plot illustrated the direction of association with predicted risk ( Fig. 4B ): higher BMI, TG, SBP, HSI, RBC, WBC, HGB, and altitude were associated with higher predicted MASLD risk, whereas higher HDL-C values were associated with lower predicted risk. Overall, SHAP provided transparent, patient-level and population-level explanations of the model output. 3.5 Nomogram and Online Prediction Tool To facilitate clinical translation and visualization, a nomogram was constructed based on the multivariate LR model ( Fig. 5A ). This tool converts the regression equation into an intuitive scoring system, allowing clinicians to quantify an individual’s probability of MASLD by summing the points assigned to each predictor. To overcome the limitations of static paper charts and enhance accessibility, we developed a web-based dynamic risk calculator ( https://altitude-masld-predictor.shinyapps.io/dynnomapp , Fig. 5B ). This platform allows clinicians and end-users to access the tool via smartphones or computers. By inputting routine clinical parameters, users can obtain real-time, personalized MASLD risk assessments. This easy-to-use, low-cost, and efficient tool provides a scalable decision support solution, particularly valuable for high-altitude regions with limited medical resources. Calibration curves demonstrated good agreement between predicted and observed probabilities. In the training set, the apparent and bootstrap bias-corrected curves (1,000 resamples) closely approximated the ideal line (Fig. 6 A). In the validation set, calibration also remained acceptable (Fig. 6 B), supporting the robustness of the model. 4. Discussion In this study, we developed and validated a machine learning–based model to predict MASLD risk in lowlanders with long-term high-altitude exposure. Among 834 screened individuals, 663 met the eligibility criteria and were included in the analysis. All six algorithms showed good discrimination, and LR achieved the highest AUC in the validation set (AUC = 0.898). Given its competitive performance and straightforward interpretability, the LR model may support early risk stratification and individualized clinical decision-making in resource-limited high-altitude settings. Using a three-step feature selection strategy (univariate analysis, LASSO regression, and the Boruta algorithm), we identified nine predictors: BMI, TG, SBP, HDL-C, HSI, WBC, RBC, HGB, and altitude. SHAP further supported the relative contributions and directional associations of these variables. Overall, metabolic indicators (BMI, TG, HDL-C), a steatosis-related index (HSI), blood pressure (SBP), hematological parameters related to hypoxic adaptation (RBC, HGB, altitude), and an inflammatory marker (WBC) were associated with MASLD risk in this cohort. These findings suggest that metabolic dysregulation, altitude-related physiological adaptation, and low-grade inflammation may jointly contribute to MASLD susceptibility under prolonged high-altitude exposure. This interpretation is consistent with evidence that chronic hypoxia and cold stress can induce metabolic reprogramming, oxidative stress, and sympathetic activation, thereby promoting hepatic lipid accumulation and inflammation through pathways including HIF-1α signaling, impaired fatty acid oxidation, insulin resistance, and mitochondrial dysfunction [ 37 ]. Notably, the prominent contributions of metabolic and inflammatory markers in our SHAP analysis align with these biological processes. 4.1 The pivotal role of metabolic indicators (BMI, TG, and HDL-C) in MASLD risk Obesity is a well-established risk factor for MASLD. In our cohort, BMI showed stable predictive value and ranked as the most influential feature in the final model. Although high-altitude hypoxia has been proposed to promote weight loss by increasing energy expenditure, our results indicate that adiposity remains a dominant driver of MASLD risk in contemporary high-altitude-exposed lowlanders. This is in line with evidence identifying BMI as an important determinant of MASLD progression and remission [ 38 ]. Furthermore, epidemiological studies suggest that cardiometabolic disorders—including obesity, hypertension, and diabetes—may remain common or even increase in hypoxic high-altitude environments, all of which are closely linked to MASLD [ 39 ]. Our univariate results in the training set are consistent with these observations. Regarding lipid metabolism, elevated TG and reduced HDL-C were among the strongest predictors of MASLD, consistent with prior population-based studies. Flores et al. reported that higher TG and lower HDL-C were associated with NAFLD risk in adults [ 40 ]. Hou et al. showed that increased TG, including postprandial TG assessed by an oral fat tolerance test, was closely related to NAFLD [ 41 ]. Peng et al. also observed an association between dyslipidemia and NAFLD in adult men [ 42 ]. Mechanistically, chronic hypoxia may alter hepatic lipid handling by activating HIF signaling, suppressing mitochondrial fatty acid β-oxidation, impairing VLDL assembly/export, and upregulating lipogenic programs, which collectively favor intrahepatic lipid accumulation [ 43 ]. In animal models, improving oxygenation has been reported to attenuate hepatic steatosis, supporting a link between hypoxia and lipid deposition [ 44 ]. Consistent with these findings, SHAP ranked TG and HDL-C among the top contributors, with TG positively and HDL-C inversely associated with predicted MASLD risk. Taken together, these results underscore dyslipidemia as a core pathway captured by the model in high-altitude-exposed lowlanders. Because BMI, TG, and HDL-C are inexpensive and routinely available, they may serve as pragmatic indicators for MASLD screening and risk assessment in high-altitude regions. 4.2 The critical role of HSI in predicting MASLD in high-altitude populations In our study, the hepatic steatosis index (HSI) showed strong predictive value for MASLD. As a composite score derived from BMI, the ALT/AST ratio, and sex, HSI integrates information related to adiposity and hepatocellular injury. Lee et al. originally developed HSI in a health-screening cohort of over 10,000 individuals and validated it in independent cohorts, demonstrating good sensitivity and specificity for identifying NAFLD as a simple and non-invasive screening tool [ 45 ]. Subsequent studies further supported associations between HSI and liver fat content, insulin resistance, and cardiovascular risk factors, with diagnostic performance comparable to other established indices such as the Fatty Liver Index (FLI) and the NAFLD Liver Fat Score [ 46 ]. With the updated disease nomenclature, accumulating evidence indicates that HSI also performs well in MASLD/MAFLD-related phenotyping and risk stratification. Elevated HSI has been linked to insulin resistance, visceral adiposity, and atherosclerotic burden, supporting its value for stratifying cardiometabolic risk [ 47 ]. Large-scale epidemiological studies have further incorporated HSI with cardiometabolic factors to define MASLD-related phenotypes and have reported positive associations between higher HSI and risks of cardiovascular disease, type 2 diabetes, and all-cause mortality [ 48 ]. Collectively, these findings suggest that HSI reflects not only hepatic steatosis but also broader metabolic dysfunction. In the context of chronic high-altitude hypoxia, the components captured by HSI may be particularly relevant. Hypoxia has been reported to alter lipid handling through HIF-related pathways and impaired oxidative metabolism, potentially favoring intrahepatic lipid accumulation [ 37 ]. In addition, sympathetic activation and systemic inflammation at high altitude may exacerbate insulin resistance and hepatocellular stress, which can influence transaminase levels [ 49 ]. These altitude-related physiological stressors may amplify the clinical signal captured by HSI in predicting MASLD risk. Consistent with this interpretation, SHAP ranked HSI among the leading predictors, indicating a substantial contribution to the model output. Given that HSI relies on routinely available clinical variables and does not require imaging, it may serve as a pragmatic indicator for MASLD screening and risk stratification in resource-limited high-altitude settings, including remote military and field engineering sites. 4.3 The role of high-altitude hematological adaptation markers (RBC, HGB, and altitude) in MASLD risk prediction In the present model, altitude, RBC count, and hemoglobin emerged as important predictors, suggesting that altitude-related hematological adaptation may be informative for MASLD risk stratification in lowlanders with long-term high-altitude exposure. Chronic exposure to high altitude (typically ≥ 2,500 m) stimulates erythropoietin-driven erythropoiesis, resulting in sustained increases in RBC count and hemoglobin as a compensatory response to maintain oxygen delivery [ 50 ]. However, excessive erythrocytosis may increase blood viscosity and vascular resistance, potentially compromising microcirculatory perfusion, including within hepatic sinusoids, and thereby aggravating metabolic stress. These findings are consistent with epidemiological evidence suggesting that erythrocyte-related indices may be associated with fatty liver disease. Analyses including NHANES-derived cohorts have reported associations between higher RBC/hemoglobin levels and NAFLD risk after adjustment for conventional metabolic factors [ 51 , 52 ]. Potential explanations include hemoconcentration-related iron dysregulation and oxidative stress [ 53 ], which may contribute to hepatic injury and metabolic perturbations. In high-altitude settings, maladaptive responses to chronic hypoxia can progress toward syndromes such as chronic mountain sickness, which are characterized by pronounced erythrocytosis, hyperviscosity, and thrombotic risk [ 54 ]. In our model, altitude served as an exposure-related variable and showed a positive contribution to predicted risk, potentially reflecting the cumulative physiological burden of chronic hypoxia. Together, these observations support the inclusion of altitude and erythrocyte-related indices as accessible markers that may improve MASLD risk assessment in high-altitude-exposed lowlanders [ 55 ]. Overall, RBC count, hemoglobin, and altitude may capture clinically relevant aspects of acclimatization and exposure intensity. Incorporating these indicators into prediction frameworks may help identify individuals who warrant closer metabolic evaluation and follow-up in high-altitude environments. 4.4 The role of cardiometabolic indicators (SBP) in MASLD and their amplification in high-altitude environments In our cohort, SBP was positively associated with MASLD risk and contributed substantially to model prediction. This finding is consistent with evidence showing a close relationship between steatotic liver disease and hypertension/cardiovascular disease. Meta-analyses and cohort studies have reported that individuals with NAFLD have a higher risk of incident hypertension; for example, a meta-analysis including more than 46,000 participants estimated an approximately 1.5–1.6-fold increased risk [ 56 ]. Conversely, hypertension has also been associated with an increased risk of NAFLD in population-based analyses [ 57 ]. In addition, NAFLD/MASLD has been recognized as a risk factor for atherosclerotic cardiovascular disease (ASCVD), and its prognostic relevance for major adverse cardiovascular events and mortality has been documented in prospective studies and reviews [ 58 , 59 ]. Several biological pathways may underlie the co-occurrence of elevated blood pressure and MASLD, including insulin resistance, chronic low-grade inflammation, endothelial dysfunction, activation of the renin–angiotensin–aldosterone system (RAAS), and sympathetic nervous system activity [ 60 ]. Hepatic steatosis and metabolic dysfunction may contribute to vascular stiffness and blood pressure elevation through inflammatory signaling and oxidative stress, while sustained hypertension may impair hepatic microcirculation and promote endothelial injury, potentially accelerating the transition from steatosis to steatohepatitis and fibrosis [ 61 , 62 ]. Together, these observations support a clinically relevant liver–cardiovascular interaction. High-altitude hypoxia may further accentuate this interplay. Hypoxia is a potent trigger of sympathetic activation and may increase vasomotor tone and blood pressure. Prior studies have reported higher SBP and peripheral vascular resistance in lowlanders relocating to high altitude as well as in high-altitude residents [ 63 , 64 ]. In addition, altitude-related erythrocytosis and increased blood viscosity may increase vascular resistance and cardiac workload, potentially compounding blood pressure–related metabolic stress. In our study, the identification of SBP as a key predictor suggests that cardiometabolic burden is an important component of MASLD risk stratification in high-altitude-exposed lowlanders. These findings support the practical value of incorporating blood pressure assessment into screening and follow-up strategies for MASLD in high-altitude settings, alongside broader cardiovascular risk evaluation. 4.5 The role of inflammatory markers (WBC) in MASLD risk prediction and their amplification in high-altitude environments Chronic low-grade inflammation is widely recognized as a key feature of MASLD pathophysiology. WBC count is an accessible peripheral marker of systemic inflammation and has been associated with NAFLD/MASLD risk in multiple studies. In a prospective cohort of more than 26,000 Han Chinese adults, higher baseline WBC predicted incident NAFLD after adjustment for adiposity, lipid profiles, and lifestyle factors [ 65 ]. Consistently, WBC levels have been reported to be higher in individuals with hepatic steatosis and to increase with steatosis severity [ 66 ]. Recent genetic analyses further support a potential contributory role of inflammatory and hematological traits. A Mendelian randomization study published in 2024 reported genetic evidence linking hematological traits, including WBC and hemoglobin, with NAFLD risk [ 67 ]. In addition, immunometabolic studies highlight that innate immune activation—such as neutrophil and monocyte recruitment and Kupffer cell activation—contributes to progression from steatosis to MASH and fibrosis [ 68 ], processes that may be partially captured by peripheral leukocyte indices. In our cohort, WBC was retained among the final predictors and contributed to the model output in SHAP analysis, underscoring the relevance of systemic inflammation in risk stratification. High-altitude hypoxia may further influence inflammatory tone: hypoxic stress can activate HIF-1α–related pathways and promote pro-inflammatory cytokine signaling, potentially enhancing leukocyte recruitment and sterile inflammation [ 69 ]. Moreover, oxidative stress and sympathetic activation during hypoxic exposure may modulate immune activity and metabolic homeostasis [ 70 ]. Together, these data suggest that inflammatory burden—reflected in part by WBC—may be particularly pertinent when assessing MASLD risk in high-altitude-exposed lowlanders. Given its low cost and routine availability, WBC may serve as a pragmatic marker to complement metabolic indicators in early screening and risk stratification in resource-limited high-altitude settings. However, WBC is non-specific and may be influenced by infection, smoking, hydration status, and other stressors; therefore, residual confounding cannot be excluded. Future studies, particularly prospective cohorts and external validations, are needed to clarify causal pathways and to determine whether targeting hypoxia-related stress or inflammation improves MASLD-related outcomes. 4.6 Strengths and clinical utility of the machine learning model Compared with prior MASLD-related prediction studies, our model offers several practical advantages for use in resource-limited high-altitude settings. Although a growing number of machine learning (ML) models for MASLD/NAFLD have been reported, many rely on inputs that are difficult to implement at scale in remote regions. For example, Chen et al. developed ML models in individuals with hypertension and prehypertension and reported an AUC of 0.889, with ALT, BMI, waist circumference, and HDL-C among the key predictors [ 71 ]. Other studies have incorporated transcriptomic, metabolomic, lipidomic, or other novel biomarkers to improve predictive performance [ 72 ]; however, these approaches are often costly and may not be feasible for population-level screening in high-altitude areas with limited infrastructure. Similarly, Rohit et al. proposed a non-invasive score combining liver enzymes and MRI-derived parameters (AUC = 0.81) [ 73 ], but MRI availability and cost constrain routine deployment in field settings. Moreover, although some models have focused on specific comorbid populations such as diabetes or cardiovascular disease [ 26 ], few have addressed early, non-invasive risk stratification in lowlanders with long-term high-altitude exposure, including military personnel, construction workers, and migrants. In our cohort, the final model achieved good discrimination in the validation set (AUC = 0.898), consistent with the performance range commonly reported in general or metabolically enriched populations. This performance may be attributable to: (1) a structured, three-stage feature selection strategy that improved predictor robustness; (2) inclusion of HSI, which integrates adiposity and liver injury signals and may enhance clinical interpretability; and (3) the distinctive physiological and metabolic profile associated with prolonged high-altitude exposure, which may be captured by routinely measured indicators. From a clinical and public health perspective, the model uses only routinely collected examination and laboratory variables, supporting scalable and non-invasive screening in high-altitude regions where imaging and specialist resources are scarce. In addition, SHAP-based explanations provide transparent, individualized contributions of predictors to model output, which may facilitate risk communication and guide follow-up intensity. Finally, the accompanying web-based dynamic calculator improves accessibility and usability, potentially enabling integration into health management programs for high-altitude-exposed lowlanders, while acknowledging that external validation is needed before broad implementation. 4.7 Limitations and future perspectives Several limitations should be considered when interpreting our findings. First, this was a single-center retrospective study, which may introduce selection bias and limit generalizability. Future multicenter, prospective studies are needed to validate model performance across diverse settings and populations. Second, MASLD was ascertained by abdominal ultrasonography rather than liver biopsy. Although biopsy remains the reference standard, ultrasonography is non-invasive, widely available, and suitable for large-scale screening; however, some degree of misclassification, especially for mild steatosis, cannot be excluded. Third, although altitude and exposure duration were captured, more granular indicators of hypoxic burden (e.g., ambient partial pressure of oxygen, oxygen saturation [SpO₂], and acclimatization status) were not available and should be incorporated in future studies to refine exposure characterization and improve model transportability. Fourth, the cohort was predominantly male (95.6%), reflecting the occupational composition of high-altitude deployment and construction work in our setting. Consequently, model applicability to females remains uncertain and warrants dedicated validation in female and more sex-balanced cohorts. Finally, we did not include an independent external validation cohort. External validation using geographically and demographically distinct datasets is essential before broader clinical implementation, and future work should also evaluate prospective clinical impact and calibration drift over time. 5. Conclusions This study developed and internally validated a machine learning–based model to predict MASLD risk in lowlanders with long-term high-altitude exposure using routinely available clinical variables. The final LR model demonstrated good discrimination and calibration, and SHAP-based interpretation highlighted the relative contributions of metabolic factors (e.g., BMI, TG, HDL-C, and HSI), blood pressure, and altitude/hematological indices to predicted risk. Given its reliance on low-cost, routinely collected data and its interpretability, the model may support scalable screening and risk stratification in resource-limited high-altitude settings. The accompanying web-based dynamic calculator further facilitates real-world use among high-altitude-exposed populations such as military personnel, construction workers, and long-term residents. External validation in independent and more diverse cohorts is warranted before broader implementation. Abbreviations A/G, albumin/globulin ratio ALB, albumin ALT, alanine aminotransferase APTT, activated partial thromboplastin time AST, aspartate aminotransferase ASCVD, atherosclerotic cardiovascular disease AUC, area under the receiver operating characteristic curve BMI, body mass index Boruta, Boruta feature selection algorithm Cr, creatinine DCA, decision curve analysis DBIL, direct bilirubin DBP, diastolic blood pressure DT, decision tree FIB, fibrinogen FLI, Fatty Liver Index GLB, globulin HbA1c, glycated hemoglobin HDL-C, high-density lipoprotein cholesterol HGB, hemoglobin HIF, hypoxia-inducible factor HIF-1α, hypoxia-inducible factor-1 alpha HSI, hepatic steatosis index IBIL, indirect bilirubin INR, international normalized ratio IQR, interquartile range LASSO, least absolute shrinkage and selection operator LightGBM, Light Gradient Boosting Machine LR, logistic regression LYMPH, lymphocyte count MAFLD, metabolic dysfunction–associated fatty liver disease MASH, metabolic dysfunction–associated steatohepatitis MASLD, metabolic dysfunction–associated steatotic liver disease MissForest, random forest–based imputation algorithm (MissForest) ML, machine learning MONO, monocyte count NAFLD, non-alcoholic fatty liver disease NEUT, neutrophil count NPV, negative predictive value PLT, platelet count PPV, positive predictive value PT, prothrombin time PTA, prothrombin activity RAAS, renin–angiotensin–aldosterone system RBC, red blood cell count RF, random forest ROC, receiver operating characteristic SBP, systolic blood pressure SD, standard deviation SHAP, Shapley Additive exPlanations SVM, support vector machine TBIL, total bilirubin TC, total cholesterol TG, triglycerides TP, total protein TT, thrombin time VLDL, very-low-density lipoprotein WBC, white blood cell count XGBoost, Extreme Gradient Boosting Declarations Ethics approval and consent to participate. This study received approval from the Ethics Committee of the General Hospital of Western Theater Command and was conducted in accordance with the principles of the Declaration of Helsinki.The ethical approval number is 2024EC4-ky024. Due to the retrospective design of the study and the analysis of anonymized data, the Ethics Committee granted a waiver for informed consent. Consent for publication Not applicable. Availability of data and materials The datasets analysed during the current study are available from the corresponding author on reasonable request. Competing interests The authors declare that they have no competing interests. Funding The study was supported by Hospital Management of the General Hospital of Western Theater Command (2024-YGLC-A01). Authors' contributions All authors above appropriately contributed to the development of this manuscript. The conceptualization of the aims of the article was made by YZ and RD. The formal acquisition and analysis of the data were carried out by JF, QW, and HD. YZ, YH and XL were involved in interpreting the data, drafting, revising, and approving the final version for submission. RD was responsible for funding acquisition and supervision. All authors read and approved the final manuscript. Acknowledgements The authors acknowledge all the clinical and research staff from the research centers. References Boldys A, Buldak L. Metabolic dysfunction-associated steatotic liver disease. Navigating terminological evolution, diagnostic frontiers and therapeutic horizon-an editorial exploration. World J Gastroenterol. 2024; 30 : 2387-90. Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multisociety Delphi consensus statement on new fatty liver disease nomenclature. J Hepatol. 2023; 79 : 1542-56. El-Kassas M, Awad A, Elbadry M, Arab JP. Tailored Model of Care for Patients with Metabolic Dysfunction-Associated Steatotic Liver Disease. Semin Liver Dis. 2024; 44 : 54-68. Deprince A, Hennuyer N, Kooijman S, Pronk ACM, Baugé E, Lienard V, et al. Apolipoprotein F is reduced in humans with steatosis and controls plasma triglyceride-rich lipoprotein metabolism. Hepatology. 2023; 77 : 1287-302. Friedman SL, Neuschwander-Tetri BA, Rinella M, Sanyal AJ. Mechanisms of NAFLD development and therapeutic strategies. Nat Med. 2018; 24 : 908-22. Powell EE, Wong VW, Rinella M. Non-alcoholic fatty liver disease. Lancet. 2021; 397 : 2212-24. Cariou B. The metabolic triad of non-alcoholic fatty liver disease, visceral adiposity and type 2 diabetes. Implications for treatment. Diabetes Obes Metab. 2022; 24 Suppl 2 : 15-27. Tymko MM, Lawley JS, Ainslie PN, Hansen AB, Hofstaetter F, Rainer S, et al. Global Reach 2018 Heightened α-Adrenergic Signaling Impairs Endothelial Function During Chronic Exposure to Hypobaric Hypoxia. Circ Res. 2020; 127 : e1-13. West JB. High-altitude medicine. Am J Respir Crit Care Med. 2012; 186 : 1229-37. Huang L. High altitude medicine in China in the 21st century. opportunities and challenges. Mil Med Res. 2014; 1 : 17. Wu TY, Ding SQ, Liu JL, Yu MT, Jia JH, Chai ZC, et al. Who should not go high: chronic disease and work at altitude during construction of the Qinghai-Tibet railroad. High Alt Med Biol. 2007; 8 : 88-107. Zhang W, Jiao L, Liu R, Zhang Y, Ji Q, Zhang H, et al. The effect of exposure to high altitude and low oxygen on intestinal microbial communities in mice. PLoS One. 2018; 13 : e0203701. Basnyat B, Murdoch DR. High-altitude illness. Lancet. 2003; 361 : 1967-74. Siebenmann C, Rasmussen P, Hug M, Keiser S, Flück D, Fisher JP, et al. Parasympathetic withdrawal increases heart rate after 2 weeks at 3454 m altitude. J Physiol. 2017; 595 : 1619-26. Savla JJ, Levine BD, Sadek HA. The Effect of Hypoxia on Cardiovascular Disease. Friend or Foe? High Alt Med Biol. 2018; 19 : 124-30. Murray AJ, Horscroft JA. Mitochondrial function at extreme high altitude. J Physiol. 2016; 594 : 1137-49. Mesarwi OA, Loomba R, Malhotra A. Obstructive Sleep Apnea, Hypoxia, and Nonalcoholic Fatty Liver Disease. Am J Respir Crit Care Med. 2019; 199 : 830-41. Lefere S, Van Steenkiste C, Verhelst X, Van Vlierberghe H, Devisscher L, Geerts A. Hypoxia-regulated mechanisms in the pathogenesis of obesity and non-alcoholic fatty liver disease. Cell Mol Life Sci. 2016; 73 : 3419-31. Sakurai A, Sakurai T, Ho HJ, Chiba H, Hui SP. Kaempferol Improves Cardiolipin and ATP in Hepatic Cells. A Cellular Model Perspective in the Context of Metabolic Dysfunction-Associated Steatotic Liver Disease. Nutrients. 2024; 16(4):508. Savransky V, Nanayakkara A, Li J, Bevans S, Smith PL, Rodriguez A, et al. Chronic intermittent hypoxia induces atherosclerosis. Am J Respir Crit Care Med. 2007; 175 : 1290-7. Chalasani N, Younossi Z, Lavine JE, Charlton M, Cusi K, Rinella M, et al. The diagnosis and management of nonalcoholic fatty liver disease. Practice guidance from the American Association for the Study of Liver Diseases. Hepatology. 2018; 67 : 328-57. Iwaki M, Fujii H, Hayashi H, Toyoda H, Oeda S, Hyogo H, et al. Prognosis of biopsy-confirmed metabolic dysfunction- associated steatotic liver disease. A sub-analysis of the CLIONE study. Clin Mol Hepatol. 2024; 30 : 225-34. Abdelhameed F, Kite C, Lagojda L, Dallaway A, Chatha KK, Chaggar SS, et al. Non-invasive Scores and Serum Biomarkers for Fatty Liver in the Era of Metabolic Dysfunction-associated Steatotic Liver Disease (MASLD). A Comprehensive Review From NAFLD to MAFLD and MASLD. Curr Obes Rep. 2024; 13 : 510-31. Deng J, Ji W, Liu H, Li L, Wang Z, Hu Y, et al. Development and validation of a machine learning-based framework for assessing metabolic-associated fatty liver disease risk. BMC Public Health. 2024; 24 : 2545. Masaebi F, Azizmohammad Looha M, Mohammadzadeh M, Pahlevani V, Farjam M, Zayeri F, et al. Machine-Learning Application for Predicting Metabolic Dysfunction-Associated Steatotic Liver Disease Using Laboratory and Body Composition Indicators. Arch Iran Med. 2024; 27 : 551-62. Nabrdalik K, Kwiendacz H, Irlik K, Hendel M, Drożdż K, Wijata AM, et al. Machine Learning Identifies Metabolic Dysfunction-Associated Steatotic Liver Disease in Patients With Diabetes Mellitus. J Clin Endocrinol Metab. 2024; 109 : 2029-38. Canivet CM, Boursier J, Loomba R. New Nomenclature for Nonalcoholic Fatty Liver Disease. Understanding Metabolic Dysfunction-Associated Steatotic Liver Disease, Metabolic Dysfunction- and Alcohol-Associated Liver Disease, and Their Implications in Clinical Practice. Semin Liver Dis. 2024; 44 : 35-42. Hong S, Lynn HS. Accuracy of random-forest-based imputation of missing data in the presence of non-normality, non-linearity, and interaction. BMC Med Res Methodol. 2020; 20 : 199. Lee Y, Leite WL. A comparison of random forest-based missing imputation methods for covariates in propensity score analysis. Psychol Methods. 2024; doi: 10.1037/met0000676. Pfeifer B, Holzinger A, Schimek MG. Robust Random Forest-Based All-Relevant Feature Ranks for Trustworthy AI. Stud Health Technol Inform. 2022; 294 : 137-8. Li W, Zhou Q, Liu W, Xu C, Tang ZR, Dong S, et al. A Machine Learning-Based Predictive Model for Predicting Lymph Node Metastasis in Patients With Ewing's Sarcoma. Front Med (Lausanne). 2022; 9 : 832108. Ejiyi CJ, Qin Z, Ukwuoma CC, Nneji GU, Monday HN, Ejiyi MB, et al. Comparative performance analysis of Boruta, SHAP, and Borutashap for disease diagnosis. A study with multiple machine learning algorithms. Network. 2025; 36 : 507-44. Ning Y, Ong MEH, Chakraborty B, Goldstein BA, Ting DSW, Vaughan R, et al. Shapley variable importance cloud for interpretable machine learning. Patterns (N Y). 2022; 3 : 100452. He W, Xu C, Wang X, Lei J, Qiu Q, Hu Y, et al. Development and validation of a risk prediction nomogram for in-stent restenosis in patients undergoing percutaneous coronary intervention. BMC Cardiovasc Disord. 2021; 21 : 435. Zhu C, You Y, Liu S, Ji Y, Yu J. A Nomogram to Predict Distant Metastasis for Patients with Esophageal Cancer. Oncol Res Treat. 2020; 43 : 2-9. Liu P, Kuang Y, Ye L, Peng C, Chen W, Shen M, et al. Predicting the Risk of Psoriatic Arthritis in Plaque Psoriasis Patients. Development and Assessment of a New Predictive Nomogram. Front Immunol. 2021; 12 : 740968. Fu X, Fu Y, Liu Y, Guo J, Liu Y, Guo Y, et al. Genetic polymorphisms of 26 Y-STR loci in the Mongolian minority from Horqin district, China. Int J Legal Med. 2016; 130 : 941-6. Huang CF, Dai CY, Lin YH, Wang CW, Jang TY, Liang PC, et al. Dynamic change of metabolic dysfunction-associated steatotic liver disease in chronic hepatitis C patients after viral eradication. A nationwide registry study in Taiwan. Clin Mol Hepatol. 2024; 30 : 883-94. Zhang W, Song WJ, Chen W, Pan Z, Zhang J, Fan L, et al. Metabolic dysfunction-associated steatotic liver disease-related hepatic fibrosis increases risk of insulin resistance, type 2 diabetes, and chronic kidney disease. Eur J Gastroenterol Hepatol. 2024; 36 : 802-10. Flores YN, Amoon AT, Su B, Velazquez-Cruz R, Ramírez-Palacios P, Salmerón J, et al. Serum lipids are associated with nonalcoholic fatty liver disease. a pilot case-control study in Mexico. Lipids Health Dis. 2021; 20 : 136. Hou X, Guan Y, Tang Y, Song A, Zhao J, Ren L, et al. A correlation study of the relationships between nonalcoholic fatty liver disease and serum triglyceride concentration after an oral fat tolerance test. Lipids Health Dis. 2021; 20 : 54. Peng K, Mo Z, Tian G. Serum Lipid Abnormalities and Nonalcoholic Fatty Liver Disease in Adult Males. Am J Med Sci. 2017; 353 : 236-41. Isaza SC, Del Pozo-Maroto E, Domínguez-Alcón L, Elbouayadi L, González-Rodríguez Á, García-Monzón C. Hypoxia and Non-alcoholic Fatty Liver Disease. Front Med (Lausanne). 2020; 7 : 578001. Yu L, Wang H, Han X, Liu H, Zhu D, Feng W, et al. Oxygen therapy alleviates hepatic steatosis by inhibiting hypoxia-inducible factor-2α. J Endocrinol. 2020; 246 : 57-67. Lee JH, Kim D, Kim HJ, Lee CH, Yang JI, Kim W, et al. Hepatic steatosis index. a simple screening tool reflecting nonalcoholic fatty liver disease. Dig Liver Dis. 2010; 42 : 503-8. Castera L, Friedrich-Rust M, Loomba R. Noninvasive Assessment of Liver Disease in Patients With Nonalcoholic Fatty Liver Disease. Gastroenterology. 2019; 156 : 1264-81. Sacks D, Baxter B, Campbell BCV, Carpenter JS, Cognard C, Dippel D, et al. Multisociety Consensus Quality Improvement Revised Consensus Statement for Endovascular Therapy of Acute Ischemic Stroke. Int J Stroke. 2018; 13 : 612-32. Chen X, Du L, Peng J. Association between estimated glucose disposal rate and cardiovascular disease prevalence and mortality outcomes in metabolic dysfunction-associated steatotic liver disease. a comparative analysis of insulin resistance markers. J Glob Health. 2025; 15 : 04249. Tian M, Zhang Y, Liu Z, Zhao N, Huang D, Hou J, et al. A new investigation of nonalcoholic fatty liver disease. Effects of hypoxia on mitochondrial function and lipid droplet autophagy. PLoS One. 2025; 20 : e0331305. Villafuerte FC, Simonson TS, Bermudez D, León-Velarde F. High-Altitude Erythrocytosis. Mechanisms of Adaptive and Maladaptive Responses. Physiology (Bethesda). 2022; 37 : 0. Zhong F, Guan L, Lin H, Zhao M, Qin Y, Li Q, et al. Red Blood Cell Count. An Unrecognized Risk Factor for Nonalcoholic Fatty Liver Disease. Front Endocrinol (Lausanne). 2021; 12 : 760981. Dai X, Zhou G, Xu L. Associations between red blood cell count and metabolic dysfunction-associated fatty liver disease(MAFLD). PLoS One. 2022; 17 : e0279274. Chung GE, Yim JY, Kim D, Kwak MS, Yang JI, Chung SJ, et al. Associations between hemoglobin concentrations and the development of incidental metabolic syndrome or nonalcoholic fatty liver disease. Dig Liver Dis. 2017; 49 : 57-62. Swenson ER. Chronic Mountain Sickness Evolving Over Time. New Data From on High. Chest. 2022; 161 : 1136-7. Zhu N, Wang X, Zhu H, Zheng Y. Blood cell parameters and risk of nonalcoholic fatty liver disease. a comprehensive Mendelian randomization study. BMC Med Genomics. 2024; 17 : 102. Ciardullo S, Grassi G, Mancia G, Perseghin G. Nonalcoholic fatty liver disease and risk of incident hypertension. a systematic review and meta-analysis. Eur J Gastroenterol Hepatol. 2022; 34 : 365-71. Li G, Peng Y, Chen Z, Li H, Liu D, Ye X. Bidirectional Association between Hypertension and NAFLD. A Systematic Review and Meta-Analysis of Observational Studies. Int J Endocrinol. 2022; 2022 : 8463640. Mantovani A, Csermely A, Petracca G, Beatrice G, Corey KE, Simon TG, et al. Non-alcoholic fatty liver disease and risk of fatal and non-fatal cardiovascular events. an updated systematic review and meta-analysis. Lancet Gastroenterol Hepatol. 2021; 6 : 903-13. Duell PB, Welty FK, Miller M, Chait A, Hammond G, Ahmad Z, et al. Nonalcoholic Fatty Liver Disease and Cardiovascular Risk. A Scientific Statement From the American Heart Association. Arterioscler Thromb Vasc Biol. 2022; 42 : e168-85. Pelusi S, Macchi C, Malvestiti F, Margarita S, De Matteis I, Periti G, et al. Interplay among lipoprotein(a), hepatic and vascular damage in individuals with metabolic dysfunction. Cardiovasc Diabetol. 2025; 24 : 447. Doumas SA, Tripathi S, Kashikar A, Khuttan A, Kumar A, Singh H, et al. Nonalcoholic Fatty Liver Disease (NAFLD) and Cardiovascular Risk. Is Imaging Helpful? Curr Probl Cardiol. 2024; 49 : 102065. Ktenopoulos N, Sagris M, Gerogianni M, Pamporis K, Apostolos A, Balampanis K, et al. Non-Alcoholic Fatty Liver Disease and Coronary Artery Disease. A Bidirectional Association Based on Endothelial Dysfunction. Int J Mol Sci. 2024; 25. Simpson LL, Steinback CD, Stembridge M, Moore JP. A sympathetic view of blood pressure control at high altitude. new insights from microneurographic studies. Exp Physiol. 2021; 106 : 377-84. Bilo G, Caravita S, Torlasco C, Parati G. Blood pressure at high altitude. physiology and clinical implications. Kardiol Pol. 2019; 77 : 596-603. Wang S, Zhang C, Zhang G, Yuan Z, Liu Y, Ding L, et al. Association between white blood cell count and non-alcoholic fatty liver disease in urban Han Chinese. a prospective cohort study. BMJ Open. 2016; 6 : e010342. Chao YL, Wu PY, Huang JC, Chiu YW, Lee JJ, Chen SC, et al. Hepatic Steatosis Is Associated with High White Blood Cell and Platelet Counts. Biomedicines. 2022; 10:892. Hu B, Wan AH, Xiang XQ, Wei YH, Chen Y, Tang Z, et al. Blood cell counts and nonalcoholic fatty liver disease. Evidence from Mendelian randomization analysis. World J Hepatol. 2024; 16 : 1145-55. Paquissi FC. Immune Imbalances in Non-Alcoholic Fatty Liver Disease. From General Biomarkers and Neutrophils to Interleukin-17 Axis Activation and New Therapeutic Targets. Front Immunol. 2016; 7 : 490. Hernández A, Geng Y, Sepúlveda R, Solís N, Torres J, Arab JP, et al. Chemical hypoxia induces pro-inflammatory signals in fat-laden hepatocytes and contributes to cellular crosstalk with Kupffer cells through extracellular vesicles. Biochim Biophys Acta Mol Basis Dis. 2020; 1866 : 165753. Drager LF, Polotsky VY, O'Donnell CP, Cravo SL, Lorenzi-Filho G, Machado BH. Translational approaches to understanding metabolic dysfunction and cardiovascular consequences of obstructive sleep apnea. Am J Physiol Heart Circ Physiol. 2015; 309 : H1101-11. Chen C, Zhang W, Yan G, Tang C. Identifying metabolic dysfunction-associated steatotic liver disease in patients with hypertension and pre-hypertension. An interpretable machine learning approach. Digit Health. 2024; 10 : 20552076241233135. Han N, He J, Shi L, Zhang M, Zheng J, Fan Y. Identification of biomarkers in nonalcoholic fatty liver disease. A machine learning method and experimental study. Front Genet. 2022; 13 : 1020899. Loomba R, Amangurbanova M, Bettencourt R, Madamba E, Siddiqi H, Richards L, et al. MASH Resolution Index. development and validation of a non-invasive score to detect histological resolution of MASH. Gut. 2024; 73 : 1343-9. Additional Declarations No competing interests reported. Supplementary Files AdditionalFile1.docx Supplementary Information Additional File 1: Supplementary tables 1 for development and validation of a machine learning model for MASLD risk prediction in lowlanders with long-term high-altitude exposure. This file contains supplementary table that provide additional details on the study results. Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 29 Apr, 2026 Reviews received at journal 25 Apr, 2026 Reviewers agreed at journal 23 Apr, 2026 Reviewers agreed at journal 23 Apr, 2026 Reviewers invited by journal 16 Apr, 2026 Editor invited by journal 20 Mar, 2026 Editor assigned by journal 19 Mar, 2026 Submission checks completed at journal 19 Mar, 2026 First submitted to journal 12 Mar, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9109584","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":628438025,"identity":"1445b7e8-713b-4d76-970a-0c7dd5f6b05c","order_by":0,"name":"Yongjiang Zhou","email":"","orcid":"","institution":"General Hospital of Western Theater Command","correspondingAuthor":false,"prefix":"","firstName":"Yongjiang","middleName":"","lastName":"Zhou","suffix":""},{"id":628438027,"identity":"93e27063-9170-4d10-a81b-8018ee2ec807","order_by":1,"name":"Youqing Huang","email":"","orcid":"","institution":"General Hospital of Western Theater Command","correspondingAuthor":false,"prefix":"","firstName":"Youqing","middleName":"","lastName":"Huang","suffix":""},{"id":628438030,"identity":"9e2b83c3-e555-4185-8433-ed50cd7986ac","order_by":2,"name":"Jian Feng","email":"","orcid":"","institution":"General Hospital of Western Theater Command","correspondingAuthor":false,"prefix":"","firstName":"Jian","middleName":"","lastName":"Feng","suffix":""},{"id":628438032,"identity":"8f716e03-4c4d-4710-a159-c55d7030e951","order_by":3,"name":"Qingqing Wang","email":"","orcid":"","institution":"General Hospital of Western Theater Command","correspondingAuthor":false,"prefix":"","firstName":"Qingqing","middleName":"","lastName":"Wang","suffix":""},{"id":628438033,"identity":"2ec6faa3-1613-4941-b464-7d4e33db695c","order_by":4,"name":"Hanyu Ding","email":"","orcid":"","institution":"General Hospital of Western Theater Command","correspondingAuthor":false,"prefix":"","firstName":"Hanyu","middleName":"","lastName":"Ding","suffix":""},{"id":628438036,"identity":"c7cda8ec-f6fe-44e7-a650-94252f3ec913","order_by":5,"name":"Xinyu Li","email":"","orcid":"","institution":"General Hospital of Western Theater Command","correspondingAuthor":false,"prefix":"","firstName":"Xinyu","middleName":"","lastName":"Li","suffix":""},{"id":628438037,"identity":"c442fa60-7487-4873-84b6-a2a9b2ebefe4","order_by":6,"name":"Ruiwu Dai","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA4UlEQVRIiWNgGAWjYDCCA4wNDAwGDAz8DGAKJEKsFkkQdYA4LVAarJwoLXzHD7d95ik4nLj5/BmD4o9tDHJ8NxIYPxfg0SJ5JrF5No/B4cRtN3IMDA62MRhL3khglp6BR4vBgcRmZh6D20AtvBtAWhI33EhgY+bBp+X8Q4iWzf1nwVrqCWu5AbVlA0MuWEuCASEtkjceNjPOMfhvPONG/geDM+ckDGeeedgsjU8L3/n0xwxv/qTJ9vcfSzOoKLOR5zuefPAzPi3IgA0YoxJAGhS5RALmB0QrHQWjYBSMghEFABhwVIdZQ4w8AAAAAElFTkSuQmCC","orcid":"","institution":"General Hospital of Western Theater Command","correspondingAuthor":true,"prefix":"","firstName":"Ruiwu","middleName":"","lastName":"Dai","suffix":""}],"badges":[],"createdAt":"2026-03-13 03:08:23","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9109584/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9109584/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107742333,"identity":"5c2d74d9-6036-4dbf-bc48-89a6dd863458","added_by":"auto","created_at":"2026-04-24 15:10:49","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":142466,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFlowchart of participant selection, dataset splitting, and model development.\u003c/strong\u003e Of 834 individuals screened, 171 were excluded and 663 were included for analysis. The dataset was randomly split into a training set (70%, n = 465) and a validation set (30%, n = 198). Feature selection was performed using univariate analysis, least absolute shrinkage and selection operator (LASSO), and the Boruta algorithm, followed by model construction, evaluation, and interpretation.\u003c/p\u003e","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-9109584/v1/466b9bb4b67ccd258e342ed5.png"},{"id":107742334,"identity":"cde0158a-9616-423e-8de3-a52324baa26b","added_by":"auto","created_at":"2026-04-24 15:10:49","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":528401,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFeature selection using Boruta, LASSO, and univariate analysis.\u003c/strong\u003e (A–B) Boruta feature importance results; shadow feature importance is shown in blue, with confirmed important, tentative, and rejected features highlighted in green, yellow, and red, respectively. (C) LASSO coefficient profiles of candidate predictors. (D) Selection of the optimal penalty parameter (λ) by 10-fold cross-validation; dashed lines indicate λ_min and λ_1se. (E) Venn diagram showing overlap among features selected by univariate analysis, Boruta, and LASSO; the intersection yielded nine predictors used for model development.\u003c/p\u003e","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-9109584/v1/0108678d788841d3dcc96e7e.png"},{"id":107742336,"identity":"85103c88-35fc-4581-8787-c975741942e1","added_by":"auto","created_at":"2026-04-24 15:10:49","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":510031,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePerformance of machine learning models for MASLD prediction in lowlanders with long-term high-altitude exposure.\u003c/strong\u003e (A) Receiver operating characteristic (ROC) curves in the training set. (B) ROC curves in the validation set. (C) Calibration curves in the validation set. (D) Decision curve analysis (DCA) in the validation set.\u003c/p\u003e","description":"","filename":"Fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-9109584/v1/cb98d38a6a195c9416b5fe05.png"},{"id":107742352,"identity":"74a8344d-0cb5-4dc1-8baf-59a43a7e4120","added_by":"auto","created_at":"2026-04-24 15:10:52","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":210892,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSHAP analysis of the final model for MASLD prediction in lowlanders with long-term high-altitude exposure.\u003c/strong\u003e (A) Global feature importance ranked by mean absolute SHAP value. (B) SHAP beeswarm plot showing the distribution and direction of feature effects on model output (color indicates feature value from low to high).\u003c/p\u003e","description":"","filename":"Fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-9109584/v1/ba61d693da8f9cc83c7b0ac9.png"},{"id":107742337,"identity":"30673b82-9553-48cc-9419-8971040b9e10","added_by":"auto","created_at":"2026-04-24 15:10:49","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":213623,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eNomogram for predicting MASLD in lowlanders with long-term high-altitude exposure.\u003c/strong\u003e (A) Static nomogram derived from the final logistic regression model. (B) Web-based dynamic nomogram interface for individualized risk estimation.\u003c/p\u003e","description":"","filename":"Fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-9109584/v1/8489e9f98b92d6f21c4af3bb.png"},{"id":107742351,"identity":"b84e7812-0c75-49e3-863b-baa1e021def6","added_by":"auto","created_at":"2026-04-24 15:10:51","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":189335,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCalibration of the nomogram for MASLD prediction.\u003c/strong\u003e (A) Calibration curve in the training set. (B) Calibration curve in the validation set.\u003c/p\u003e","description":"","filename":"Fig6.png","url":"https://assets-eu.researchsquare.com/files/rs-9109584/v1/fd49fda174f5ae7103cdc534.png"},{"id":107868862,"identity":"07d2607c-2505-474c-b87e-d90fc59b22c6","added_by":"auto","created_at":"2026-04-27 07:34:35","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2156416,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9109584/v1/802ab7c8-6733-4664-a132-8311c91d663a.pdf"},{"id":107742359,"identity":"6e4c6263-568a-4bf9-8c36-e867672b7778","added_by":"auto","created_at":"2026-04-24 15:10:57","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":23248,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSupplementary Information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAdditional File 1\u003c/strong\u003e: \u003cstrong\u003eSupplementary tables 1\u003c/strong\u003e for development and validation of a machine learning model for MASLD risk prediction in lowlanders with long-term high-altitude exposure. This file contains supplementary table that provide additional details on the study results.\u003c/p\u003e","description":"","filename":"AdditionalFile1.docx","url":"https://assets-eu.researchsquare.com/files/rs-9109584/v1/b729cc9c65eb9a70f8d64ea5.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Development and validation of a machine learning model for MASLD risk prediction in lowlanders with long-term high-altitude exposure","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eMetabolic dysfunction\u0026ndash;associated steatotic liver disease (MASLD) is a recently adopted term that has replaced non-alcoholic fatty liver disease (NAFLD), emphasizing the central role of metabolic dysregulation in disease initiation and progression[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. MASLD is now the most prevalent chronic liver disease worldwide, affecting more than 30% of adults, and its prevalence continues to increase [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. The disease spectrum ranges from simple hepatic steatosis to metabolic dysfunction\u0026ndash;associated steatohepatitis (MASH), progressive fibrosis, and ultimately cirrhosis and hepatocellular carcinoma [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. As a multisystem disorder, MASLD is also associated with elevated risks of cardiovascular disease, chronic kidney disease, and all-cause mortality [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Because MASLD often develops silently and lacks specific symptoms in the early stage, it poses an escalating burden on public health systems. Therefore, establishing effective strategies for early identification and risk prediction has substantial clinical and societal importance.\u003c/p\u003e \u003cp\u003eIn recent years, the expansion of mountaineering, infrastructure development, mineral resource exploitation, military operations, and renewable energy projects has markedly increased the number of lowlanders who reside or work in hypoxic high-altitude environments for prolonged periods [\u003cspan additionalcitationids=\"CR9\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. During the construction of the Qinghai\u0026ndash;Tibet Railway alone, more than 100,000 workers were stationed long-term at altitudes of 2,800\u0026ndash;5,200 m [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Similarly, routine military training on the western plateau and the rapid expansion of wind and photovoltaic industries in Qinghai, Tibet, and Sichuan have exposed large populations of non-indigenous individuals to chronic hypoxia and acclimatization-related metabolic stress [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. High-altitude tourism further contributes; approximately three million tourists from lowland regions enter Tibet annually, and more than 70% of itineraries exceed 10 days, suggesting that even short- to medium-term exposure may perturb metabolic homeostasis [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Although epidemiological data indicate that approximately 160\u0026nbsp;million people worldwide permanently reside at altitudes above 1,500 m [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e], evidence regarding MASLD risk among lowlanders who migrate to or are stationed at high altitudes remains limited. Moreover, the generalizability of existing lowland-derived risk models to this population has not been established, underscoring the need for targeted risk assessment in long-term high-altitude lowlanders.\u003c/p\u003e \u003cp\u003eThe distinctive high-altitude environment\u0026mdash;characterized by hypobaric hypoxia and cold exposure\u0026mdash;can profoundly reshape metabolic homeostasis [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Prior studies suggest that chronic hypoxia may promote hepatic lipid accumulation and fibrosis by modulating oxygen-sensing pathways (e.g., the HIF signaling axis) and triggering inflammation and oxidative stress [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. However, the impact of hypoxia on hepatic metabolism remains controversial. Some animal studies indicate that hypoxic exposure may improve mitochondrial function and alleviate steatosis [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e], whereas others suggest that it exacerbates metabolic disorders [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. The limited and inconsistent evidence linking altitude exposure to MASLD risk constrains our understanding of susceptibility mechanisms and the geographical epidemiology of MASLD in this unique setting, highlighting a critical gap in risk stratification and early screening for high-altitude-exposed lowlanders.\u003c/p\u003e \u003cp\u003eLiver biopsy remains the gold standard for diagnosing and staging MASLD and provides direct histological evidence of inflammation and fibrosis [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]; however, its invasiveness and associated risks limit its suitability for large-scale screening and longitudinal follow-up [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Accordingly, non-invasive prediction models have become a major focus. Traditional statistical approaches are often constrained when handling high-dimensional clinical variables and complex non-linear relationships. In contrast, machine learning (ML) offers advantages in feature selection, pattern recognition, and risk stratification and has been increasingly applied to metabolic disease prediction. Recent studies have shown that ML-based MASLD prediction models can improve screening efficiency and diagnostic accuracy. For example, Deng et al. developed a model incorporating age, BMI, triglycerides, fasting blood glucose, waist circumference, and HDL-C with an AUC of 0.862 [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]; Masaebi et al. proposed a model combining questionnaire data, blood tests, and liver ultrasound to inform prevention and management [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]; and an additional study in a diabetic population reported an 8-parameter ML model with an AUC of 0.84 [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eDespite the growing body of MASLD prediction research, models tailored to lowlanders with long-term high-altitude exposure are still lacking. Because high-altitude acclimatization induces substantial physiological adaptations, such as compensatory erythrocytosis and blood pressure fluctuations, directly applying models derived from lowland populations may introduce predictive bias. Therefore, this study aimed to develop and validate a machine learning model to predict MASLD risk in lowlanders with long-term high-altitude exposure using demographic characteristics, lifestyle factors, and routine clinical indicators. By identifying key predictors, we sought to clarify how the high-altitude environment may influence metabolic homeostasis and to provide a low-cost, clinically practical tool to support early screening and individualized health management in resource-limited high-altitude regions, thereby addressing an important gap in high-altitude environmental medicine.\u003c/p\u003e"},{"header":"2 Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Study Population\u003c/h2\u003e \u003cp\u003eThis retrospective study analyzed demographic and clinical data from lowlanders with long-term high-altitude exposure who attended the General Hospital of Western Theater Command between July 2022 and June 2023. A total of 834 individuals were initially screened. According to predefined inclusion and exclusion criteria \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e\u003cb\u003e)\u003c/b\u003e, 663 participants were included in the final analysis. The inclusion criteria were: (a) age\u0026thinsp;\u0026ge;\u0026thinsp;18 years; (b) exposure to altitudes\u0026thinsp;\u0026gt;\u0026thinsp;2,500 m; and (c) duration of high-altitude exposure\u0026thinsp;\u0026gt;\u0026thinsp;1 year. Participants were excluded for the following reasons: incomplete clinical data (n\u0026thinsp;=\u0026thinsp;43), exposure to altitudes\u0026thinsp;\u0026lt;\u0026thinsp;2,500 m (n\u0026thinsp;=\u0026thinsp;21), high-altitude exposure\u0026thinsp;\u0026lt;\u0026thinsp;1 year (n\u0026thinsp;=\u0026thinsp;15), other liver diseases (n\u0026thinsp;=\u0026thinsp;24), and excessive alcohol consumption (n\u0026thinsp;=\u0026thinsp;68). Given the retrospective design, written informed consent was waived by the institutional ethics committee. All personal identifiers were anonymized prior to analysis to ensure confidentiality and compliance with research ethics.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Data Collection\u003c/h2\u003e \u003cp\u003eBaseline information was collected prior to physical examination by trained medical personnel using standardized questionnaires and interviews, including demographic characteristics (age, gender, education level), lifestyle factors (smoking, alcohol consumption, tea drinking, dietary patterns, and physical activity type and intensity), and environmental exposure history (residence location, altitude, and duration of exposure). Physical examinations included measurements of height, weight, systolic blood pressure (SBP), and diastolic blood pressure (DBP). Altitude was determined based on the specific residence/duty location recorded in the medical examination data and corresponding geographic altitude information. For participants with multiple postings, altitude and exposure duration were defined according to the primary location (longest duration). Blood pressure was assessed using standardized automated upper-arm sphygmomanometers to ensure consistency and reproducibility.\u003c/p\u003e \u003cp\u003eVenous blood samples were collected immediately upon admission for a comprehensive laboratory analysis, categorized as follows: (1) hematology, including white blood cell (WBC) count, red blood cell (RBC) count, neutrophil count(NEUT), lymphocyte count(LYMPH), monocyte count(MONO), hemoglobin (HGB), and platelet count(PLT); (2) liver function, comprising alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin (TBIL), direct bilirubin (DBIL), indirect bilirubin (IBIL), total protein (TP), albumin (ALB), globulin (GLB), and the albumin/globulin (A/G) ratio; (3) lipid profile, consisting of total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C); (4) renal function, assessed via urea and creatinine (Cr); (5) coagulation profile, including prothrombin time (PT), prothrombin activity (PTA), international normalized ratio (INR), fibrinogen (FIB), activated partial thromboplastin time (APTT), thrombin time (TT), and D-dimer; and (6) glycemic control, measured by glycated hemoglobin (HbA1c). Additionally, abdominal ultrasonography was performed by the same medical team using standardized equipment.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Patient Grouping\u003c/h2\u003e \u003cp\u003e In accordance with the MASLD definition proposed by the 2023 international multi-society Delphi consensus, hepatic steatosis was diagnosed by senior physicians using abdominal ultrasonography. Patients were diagnosed with MASLD if they exhibited hepatic steatosis and met at least one of the following cardiometabolic criteria [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]: (a) overweight/obesity (BMI\u0026thinsp;\u0026ge;\u0026thinsp;23 kg/m\u0026sup2; for Asians); (b) type 2 diabetes; (c) blood pressure\u0026thinsp;\u0026ge;\u0026thinsp;130/85 mmHg or receiving antihypertensive treatment; (d) fasting TG\u0026thinsp;\u0026ge;\u0026thinsp;150 mg/dL (1.7 mmol/L); (e) HDL-C\u0026thinsp;\u0026lt;\u0026thinsp;40 mg/dL (1.0 mmol/L) for men or \u0026lt;\u0026thinsp;50 mg/dL (1.3 mmol/L) for women; or (f) HbA1c\u0026thinsp;\u0026ge;\u0026thinsp;5.7%. Fasting plasma glucose was not measured in this study; therefore, glycemic status was assessed using documented type 2 diabetes and/or HbA1c. Exclusion criteria included excessive alcohol consumption (\u0026gt;\u0026thinsp;30 g/day for men, \u0026gt; 20 g/day for women) and other liver diseases (e.g., viral hepatitis, drug-induced liver injury, autoimmune liver disease). Participants without hepatic steatosis on abdominal ultrasonography served as controls.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Machine Learning\u003c/h2\u003e \u003cdiv id=\"Sec7\" class=\"Section3\"\u003e \u003ch2\u003e2.4.1 Data Processing\u003c/h2\u003e \u003cp\u003eVariables with \u0026gt;\u0026thinsp;20% missingness were excluded to reduce potential bias [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. To assess generalizability and minimize information leakage, the dataset was randomly split into a training set (70%) and a validation set (30%) using a fixed random seed (123). Missing values were then imputed after splitting using the random forest\u0026ndash;based MissForest algorithm, yielding complete datasets for subsequent modeling [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section3\"\u003e \u003ch2\u003e2.4.2 Feature Selection\u003c/h2\u003e \u003cp\u003eFeature selection was performed using the training set only. Variables were screened sequentially using univariate analysis, least absolute shrinkage and selection operator (LASSO) regression, and the Boruta algorithm. In univariate analyses, variables with P\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were considered candidate predictors. LASSO regression was implemented using the glmnet package (α\u0026thinsp;=\u0026thinsp;1), and 10-fold cross-validation was used to select the optimal regularization parameter (lambda.1se). Predictors with non-zero coefficients at the selected λ were retained [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. Ultimately, nine variables consistently supported across these procedures were selected for model development.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section3\"\u003e \u003ch2\u003e2.4.3 Model Construction and Evaluation\u003c/h2\u003e \u003cp\u003eModel development was conducted in Python 3.10.4 using the nine selected features. Six algorithms were evaluated: logistic regression (LR), random forest (RF), decision tree (DT), support vector machine (SVM), extreme gradient boosting (XGBoost), and light gradient boosting machine (LightGBM) [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. All models were implemented using scikit-learn. Hyperparameters were optimized by grid search with 10-fold stratified cross-validation on the training set (random seed\u0026thinsp;=\u0026thinsp;123), targeting the area under the receiver operating characteristic curve (AUC). The final tuned models were evaluated in the held-out validation set. Performance metrics included AUC, accuracy, sensitivity, specificity, F1 score, Cohen\u0026rsquo;s kappa, positive predictive value (PPV), and negative predictive value (NPV). Sensitivity, specificity, and F1 score were calculated using a default probability threshold of 0.5. Calibration and clinical utility were assessed using calibration curves and decision curve analysis (DCA). The optimal model was selected based on overall discrimination, calibration, and clinical utility.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section3\"\u003e \u003ch2\u003e2.4.4 Model Interpretation\u003c/h2\u003e \u003cp\u003eInterpretability remains a challenge in clinical machine learning. To address the \"black box\" issue, we utilized Shapley Additive exPlanations (SHAP) based on game theory. SHAP quantifies the marginal contribution of each variable, enabling both global assessment of variable importance and local explanation of individual predictions [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Although primarily designed for tree-based models, the LinearExplainer method extends SHAP to linear models like LR [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. This approach provided variable importance rankings and elucidated the direction and magnitude of each feature's effect, enhancing the transparency and clinical trust in the model.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section3\"\u003e \u003ch2\u003e2.4.5 Nomogram Construction\u003c/h2\u003e \u003cp\u003eTo enhance clinical applicability, a nomogram was constructed based on the final multivariate LR model. This tool, widely used in predictive modeling [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e], visualizes individual risk. Performance was validated using calibration curves and 1,000 bootstrap resamples to minimize overfitting [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. The nomogram assigns scores to variables to calculate MASLD probability, facilitating rapid risk stratification. This low-cost, easy-to-use tool is particularly valuable for resource-limited high-altitude regions, aiding frontline medical personnel in optimizing resource allocation [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e2.5 Statistical Analysis\u003c/h2\u003e \u003cp\u003eStatistical analyses were performed using R software (version 4.4.3). Normally distributed continuous variables were expressed as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation (SD), while non-normally distributed variables were reported as median (interquartile range) [M (IQR)]. Categorical variables were presented as frequencies and percentages [n (%)]. Between-group comparisons employed Student's t-test, Mann-Whitney U test, or χ\u0026sup2; test as appropriate. All hypothesis tests were two-tailed with statistical significance defined as α\u0026thinsp;=\u0026thinsp;0.05.\u003c/p\u003e \u003c/div\u003e"},{"header":"3.Results","content":"\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Baseline Characteristics\u003c/h2\u003e \u003cp\u003eThis study screened 834 lowlanders with long-term high-altitude exposure who attended the General Hospital of Western Theater Command. According to predefined exclusion criteria (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), 171 individuals were excluded, including incomplete clinical data (n\u0026thinsp;=\u0026thinsp;43), exposure to altitudes\u0026thinsp;\u0026lt;\u0026thinsp;2,500 m (n\u0026thinsp;=\u0026thinsp;21), high-altitude exposure\u0026thinsp;\u0026lt;\u0026thinsp;1 year (n\u0026thinsp;=\u0026thinsp;15), other liver diseases (n\u0026thinsp;=\u0026thinsp;24), and excessive alcohol consumption (n\u0026thinsp;=\u0026thinsp;68), leaving 663 eligible participants for analysis. The median age was 50.0 years (IQR: 45.0\u0026ndash;54.0), and 634 participants were male (95.6%). Long-term smoking was reported by 312 subjects (47.06%); 330 (49.77%) had hypertension, and 128 (19.31%) had diabetes. The mean residence altitude was 3,700 m, with a mean exposure duration of 204 months. Overall, 248 participants (37.41%) were stationed at 2,500\u0026ndash;3,500 m and 415 (62.59%) at \u0026ge;\u0026thinsp;3,500 m. Exposure duration was 2\u0026ndash;5 years in 93 individuals (14.03%) and \u0026gt;\u0026thinsp;5 years in 570 (85.97%). Physical activity was classified as low, moderate, and high in 216 (32.58%), 160 (24.13%), and 287 (43.29%) participants, respectively, and 356 (53.70%) reported habitual tea consumption.\u003c/p\u003e \u003cp\u003eCompared with the non-MASLD group, the MASLD group differed significantly across multiple variables (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05; Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), including general characteristics (BMI, SBP, DBP, and altitude), hematological indices (WBC, RBC, HGB, and leukocyte subsets), and metabolic/liver-related markers (TG, HDL-C, HSI, ALT, and AST).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cb\u003eBaseline characteristics of participants in this study.\u003c/b\u003e Continuous values were presented as median [interquartile range]. Categorical valueswere presented as numbers (percentages).\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariables\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTotal(n\u0026thinsp;=\u0026thinsp;663)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNon-MAFLD(n\u0026thinsp;=\u0026thinsp;373)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMAFLD(n\u0026thinsp;=\u0026thinsp;290)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003ep\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGender\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.018\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e29 ( 4.37%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e23 ( 6.17%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e6 ( 2.07%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e634 (95.63%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e350 (93.83%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e284 (97.93%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSmoking\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.527\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e351 (52.94%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e202 (54.16%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e149 (51.38%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e312 (47.06%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e171 (45.84%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e141 (48.62%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHTN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.335\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e333 (50.23%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e194 (52.01%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e139 (47.93%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e330 (49.77%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e179 (47.99%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e151 (52.07%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.969\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e532 (80.24%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e300 (80.43%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e232 (80.00%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e131 (19.76%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e73 (19.57%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e58 (20.00%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e248 (37.41%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e160 (42.90%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e88 (30.34%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVHA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e415 (62.59%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e213 (57.10%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e202 (69.66%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eED\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.021\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSTE\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3 ( 0.45)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2 ( 0.54)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1 ( 0.34)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMTE\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e87 (13.12%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e37 ( 9.92)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e50 (17.24%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLTE\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e573 (86.43%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e334 (89.54%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e239 (82.41%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.094\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLPA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e215 (32.43%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e132 (35.39%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e83 (28.62%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMPA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e167 (25.19%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e84 (22.52%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e83 (28.62%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHPA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e281 (42.38%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e157 (42.09%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e124 (42.76%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDietary\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.871\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLFD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e254 (38.31%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e141 (37.80%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e113 (38.97%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHFD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e134 (20.21%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e74 (19.84%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e60 (20.69%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e275 (41.48%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e158 (42.36%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e117 (40.34%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTea\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.009\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e311 (46.91%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e192 (51.47%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e119 (41.03%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e352 (53.09%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e181 (48.53%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e171 (58.97%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAltitude\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3700.00 [3000.00,3700.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3600.00 [3000.00,3700.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3700.00 [3200.00,3800.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e204.00 [144.00,240.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e206.00 [160.00,240.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e202.50 [120.00,240.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.162\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e50.00 [45.00,54.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e50.00 [45.00,54.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e50.00 [46.00,55.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.687\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSBP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e123.00 [111.00,134.50]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e116.00 [108.00,126.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e131.00 [120.00,140.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDBP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e74.00 [67.00,82.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e72.00 [66.00,79.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e77.00 [70.00,84.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBMI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e24.83 (2.66)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e23.45 (2.21)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e26.61 (2.08)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWBC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5.77 [4.90,6.69]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e5.29 [4.52,6.22]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e6.35 [5.56,7.40]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRBC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5.13 (0.39)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4.99 (0.36)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e5.30 (0.36)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNEUT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.32 [2.71,4.03]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3.15 [2.58,3.88]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3.54 [2.93,4.20]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLYMPH\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.66 [1.35,2.02]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.61 [1.32,1.93]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.72 [1.42,2.12]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMONO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.33 [0.27,0.40]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.32 [0.26,0.39]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.34 [0.28,0.43]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHGB\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e155.82 (10.65)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e152.89 (10.06)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e159.59 (10.21)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePLT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e183.00 [152.00,219.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e183.00 [152.00,215.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e184.00 [152.00,222.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.466\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUrea\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5.25 [4.50,6.07]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e5.25 [4.48,6.04]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e5.26 [4.52,6.14]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.554\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCr\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e78.30 (10.53)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e79.14 (10.31)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e77.23 (10.72)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.02\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eALT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e24.40 [17.70,34.65]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e20.70 [15.70,27.40]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e30.30 [22.33,41.68]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAST\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e22.50 [19.05,27.60]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e21.50 [18.80,25.60]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e24.70 [20.00,29.90]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eALT/AST\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.08 [0.85,1.37]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.97 [0.78,1.19]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.29 [1.03,1.55]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHSI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e34.11 [30.54,37.02]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e31.81 [29.17,34.68]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e36.72 [34.24,39.25]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4.81 (0.81)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4.78 (0.80)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.85 (0.84)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.267\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTG\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.62 [1.15,2.32]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.30 [0.92,1.75]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.22 [1.59,2.91]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHDL-C\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.32 [1.13,1.54]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.48 [1.28,1.65]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.16 [1.04,1.30]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLDL-C\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.05 (0.66)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2.97 (0.66)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3.15 (0.64)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e73.62 (3.76)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e73.25 (3.79)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e74.11 (3.67)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.003\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eALB\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e47.27 (2.12)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e47.02 (2.12)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e47.59 (2.09)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGLOB\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e26.20 [24.30,28.35]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e26.20 [24.20,28.20]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e26.30 [24.30,28.58]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.421\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eA/G\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.81 [1.66,1.96]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.81 [1.66,1.96]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.81 [1.66,1.97]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.593\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTBIL\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e18.60 [15.05,23.15]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e19.10 [15.20,23.50]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e17.90 [14.83,22.50]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.105\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDBIL\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4.50 [3.70,5.40]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4.60 [3.70,5.50]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.30 [3.60,5.10]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.015\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIBIL\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e14.00 [11.40,18.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e14.40 [11.50,18.10]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e13.65 [11.30,17.60]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.161\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e10.70 [10.40,10.95]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e10.80 [10.50,11.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e10.60 [10.40,10.80]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePTA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e105.30 [102.35,109.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e104.10 [101.80,107.70]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e106.50 [104.10,109.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eINR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.92 [0.89,0.94]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.93 [0.90,0.95]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.91 [0.89,0.93]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFIB\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2.59 [2.30,2.98]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2.55 [2.27,2.93]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.59 [2.33,2.98]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.123\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAPTT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e26.88 (1.61)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e27.08 (1.62)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e26.64 (1.56)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e16.52 (0.77)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e16.53 (0.78)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e16.51 (0.75)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.696\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.11 [0.08,0.19]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.11 [0.08,0.19]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.11 [0.08,0.21]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.337\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHbA1c\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5.80 [5.50,6.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e5.70 [5.50,6.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e5.80 [5.60,6.10]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Data Partitioning and Feature Extraction\u003c/h2\u003e \u003cp\u003eThe dataset was randomly split into a training set (n\u0026thinsp;=\u0026thinsp;465) and a validation set (n\u0026thinsp;=\u0026thinsp;198) at a 7:3 ratio.To identify predictors closely associated with MASLD, a three-stage screening strategy comprising univariate analysis, LASSO regression, and the Boruta algorithm was employed in the training set.\u003c/p\u003e \u003cp\u003eIn univariate analyses, 27 variables were significantly associated with MASLD (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05), including gender, physical activity, tea consumption, altitude, SBP, DBP, BMI, WBC, RBC, and NEUT (\u003cb\u003eSupplementary Table\u0026nbsp;1\u003c/b\u003e). Boruta further evaluated feature relevance and identified 17 \u0026ldquo;confirmed important\u0026rdquo; variables (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA\u0026ndash;B). LASSO regression selected 12 predictors with non-zero coefficients at the optimal penalty parameter (λ\u0026thinsp;=\u0026thinsp;0.0067) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC\u0026ndash;D), including altitude, SBP, BMI, WBC, RBC, HGB, HSI, TG, HDL-C, ALB, APTT, and tea consumption.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIntersecting the three selection methods yielded nine stable predictors: altitude, SBP, BMI, WBC, RBC, HGB, HSI, TG, and HDL-C (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eE), which were used for subsequent model development.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Model Construction and Evaluation\u003c/h2\u003e \u003cp\u003eUsing the training set, we developed and compared six machine learning models (LR, XGBoost, LightGBM, RF, SVM, and DT). After hyperparameter optimization by grid search with 10-fold stratified cross-validation, model performance metrics were calculated (Tables\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e\u0026ndash;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). LR achieved the highest AUC in the validation set (AUC\u0026thinsp;=\u0026thinsp;0.898) with an accuracy of 0.828, sensitivity of 0.816, specificity of 0.838, and F1 score of 0.807 (probability threshold\u0026thinsp;=\u0026thinsp;0.5). ROC analysis showed good discrimination (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA\u0026ndash;B). Calibration curves indicated acceptable agreement between predicted and observed risks (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC), and decision curve analysis demonstrated net clinical benefit across a range of threshold probabilities (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD). Given its competitive performance and interpretability, LR was selected as the final model.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePerformance of each algorithm in the training set cohort.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"11\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePrecision\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eF1 Score\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eKappa\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eYouden's J\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003ePPV\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c11\"\u003e \u003cp\u003eNPV\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.920(0.895\u0026ndash;0.943)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.830\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.795\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.823\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.836\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.809\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.656\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.659\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.795\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.859\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.894(0.865\u0026ndash;0.992)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.815\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.745\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.877\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.767\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.805\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.631\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.644\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.745\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.889\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.986(0.977\u0026ndash;0.993)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.933\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.922\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.926\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.939\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.924\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.865\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.865\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.922\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.943\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eXGBoost\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.972(0.959\u0026ndash;0.983)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.899\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.868\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.906\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.893\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.887\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.796\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.800\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.868\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.925\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLightGBM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.976(0.965\u0026ndash;0.986)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.918\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.915\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.897\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.935\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.905\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.834\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.832\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.915\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.921\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.912(0.885\u0026ndash;0.936)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.839\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.811\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.823\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.851\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.817\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.673\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.674\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.811\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.861\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePerformance of each algorithm in the validation set cohort.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"11\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePrecision\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eF1 Score\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eKappa\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eYouden's J\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003ePPV\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c11\"\u003e \u003cp\u003eNPV\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.898(0.856\u0026ndash;0.939)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.828\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.798\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.816\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.838\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.807\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.652\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.654\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.798\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.853\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.815(0.759\u0026ndash;0.876)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.758\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.701\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.782\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.739\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.739\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.514\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.520\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.701\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.812\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.878(0.832\u0026ndash;0.923)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.788\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.737\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.805\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.775\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.769\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.574\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.579\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.737\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.835\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eXGBoost\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.878(0.832\u0026ndash;0.922)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.798\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.783\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.747\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.838\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.765\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.588\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.585\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.783\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.809\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLightGBM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.870(0.821.0.917)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.773\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.744\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.736\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.802\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.740\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.538\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.537\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.744\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.795\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.896(0.854\u0026ndash;0.936)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.808\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.769\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.805\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.811\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.787\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.612\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.615\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.769\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.841\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Model Interpretation\u003c/h2\u003e \u003cp\u003eSHAP was applied to interpret the final LR model. Global SHAP importance ranked the nine predictors (highest to lowest) as BMI, TG, SBP, HDL-C, HSI, RBC, WBC, HGB, and altitude (\u003cb\u003eFig.\u0026nbsp;4A\u003c/b\u003e). The SHAP beeswarm plot illustrated the direction of association with predicted risk (\u003cb\u003eFig.\u0026nbsp;4B\u003c/b\u003e): higher BMI, TG, SBP, HSI, RBC, WBC, HGB, and altitude were associated with higher predicted MASLD risk, whereas higher HDL-C values were associated with lower predicted risk. Overall, SHAP provided transparent, patient-level and population-level explanations of the model output.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003e3.5 Nomogram and Online Prediction Tool\u003c/h2\u003e \u003cp\u003eTo facilitate clinical translation and visualization, a nomogram was constructed based on the multivariate LR model (\u003cb\u003eFig.\u0026nbsp;5A\u003c/b\u003e). This tool converts the regression equation into an intuitive scoring system, allowing clinicians to quantify an individual\u0026rsquo;s probability of MASLD by summing the points assigned to each predictor. To overcome the limitations of static paper charts and enhance accessibility, we developed a web-based dynamic risk calculator (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://altitude-masld-predictor.shinyapps.io/dynnomapp\u003c/span\u003e\u003cspan address=\"https://altitude-masld-predictor.shinyapps.io/dynnomapp\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e, \u003cb\u003eFig.\u0026nbsp;5B\u003c/b\u003e). This platform allows clinicians and end-users to access the tool via smartphones or computers. By inputting routine clinical parameters, users can obtain real-time, personalized MASLD risk assessments. This easy-to-use, low-cost, and efficient tool provides a scalable decision support solution, particularly valuable for high-altitude regions with limited medical resources.\u003c/p\u003e \u003cp\u003eCalibration curves demonstrated good agreement between predicted and observed probabilities. In the training set, the apparent and bootstrap bias-corrected curves (1,000 resamples) closely approximated the ideal line (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e6\u003c/span\u003eA). In the validation set, calibration also remained acceptable (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e6\u003c/span\u003eB), supporting the robustness of the model.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eIn this study, we developed and validated a machine learning\u0026ndash;based model to predict MASLD risk in lowlanders with long-term high-altitude exposure. Among 834 screened individuals, 663 met the eligibility criteria and were included in the analysis. All six algorithms showed good discrimination, and LR achieved the highest AUC in the validation set (AUC\u0026thinsp;=\u0026thinsp;0.898). Given its competitive performance and straightforward interpretability, the LR model may support early risk stratification and individualized clinical decision-making in resource-limited high-altitude settings.\u003c/p\u003e \u003cp\u003eUsing a three-step feature selection strategy (univariate analysis, LASSO regression, and the Boruta algorithm), we identified nine predictors: BMI, TG, SBP, HDL-C, HSI, WBC, RBC, HGB, and altitude. SHAP further supported the relative contributions and directional associations of these variables. Overall, metabolic indicators (BMI, TG, HDL-C), a steatosis-related index (HSI), blood pressure (SBP), hematological parameters related to hypoxic adaptation (RBC, HGB, altitude), and an inflammatory marker (WBC) were associated with MASLD risk in this cohort. These findings suggest that metabolic dysregulation, altitude-related physiological adaptation, and low-grade inflammation may jointly contribute to MASLD susceptibility under prolonged high-altitude exposure. This interpretation is consistent with evidence that chronic hypoxia and cold stress can induce metabolic reprogramming, oxidative stress, and sympathetic activation, thereby promoting hepatic lipid accumulation and inflammation through pathways including HIF-1α signaling, impaired fatty acid oxidation, insulin resistance, and mitochondrial dysfunction [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. Notably, the prominent contributions of metabolic and inflammatory markers in our SHAP analysis align with these biological processes.\u003c/p\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e4.1 The pivotal role of metabolic indicators (BMI, TG, and HDL-C) in MASLD risk\u003c/h2\u003e \u003cp\u003eObesity is a well-established risk factor for MASLD. In our cohort, BMI showed stable predictive value and ranked as the most influential feature in the final model. Although high-altitude hypoxia has been proposed to promote weight loss by increasing energy expenditure, our results indicate that adiposity remains a dominant driver of MASLD risk in contemporary high-altitude-exposed lowlanders. This is in line with evidence identifying BMI as an important determinant of MASLD progression and remission [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. Furthermore, epidemiological studies suggest that cardiometabolic disorders\u0026mdash;including obesity, hypertension, and diabetes\u0026mdash;may remain common or even increase in hypoxic high-altitude environments, all of which are closely linked to MASLD [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. Our univariate results in the training set are consistent with these observations.\u003c/p\u003e \u003cp\u003eRegarding lipid metabolism, elevated TG and reduced HDL-C were among the strongest predictors of MASLD, consistent with prior population-based studies. Flores et al. reported that higher TG and lower HDL-C were associated with NAFLD risk in adults [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Hou et al. showed that increased TG, including postprandial TG assessed by an oral fat tolerance test, was closely related to NAFLD [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. Peng et al. also observed an association between dyslipidemia and NAFLD in adult men [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. Mechanistically, chronic hypoxia may alter hepatic lipid handling by activating HIF signaling, suppressing mitochondrial fatty acid β-oxidation, impairing VLDL assembly/export, and upregulating lipogenic programs, which collectively favor intrahepatic lipid accumulation [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. In animal models, improving oxygenation has been reported to attenuate hepatic steatosis, supporting a link between hypoxia and lipid deposition [\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eConsistent with these findings, SHAP ranked TG and HDL-C among the top contributors, with TG positively and HDL-C inversely associated with predicted MASLD risk. Taken together, these results underscore dyslipidemia as a core pathway captured by the model in high-altitude-exposed lowlanders. Because BMI, TG, and HDL-C are inexpensive and routinely available, they may serve as pragmatic indicators for MASLD screening and risk assessment in high-altitude regions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e4.2 The critical role of HSI in predicting MASLD in high-altitude populations\u003c/h2\u003e \u003cp\u003eIn our study, the hepatic steatosis index (HSI) showed strong predictive value for MASLD. As a composite score derived from BMI, the ALT/AST ratio, and sex, HSI integrates information related to adiposity and hepatocellular injury. Lee et al. originally developed HSI in a health-screening cohort of over 10,000 individuals and validated it in independent cohorts, demonstrating good sensitivity and specificity for identifying NAFLD as a simple and non-invasive screening tool [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]. Subsequent studies further supported associations between HSI and liver fat content, insulin resistance, and cardiovascular risk factors, with diagnostic performance comparable to other established indices such as the Fatty Liver Index (FLI) and the NAFLD Liver Fat Score [\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eWith the updated disease nomenclature, accumulating evidence indicates that HSI also performs well in MASLD/MAFLD-related phenotyping and risk stratification. Elevated HSI has been linked to insulin resistance, visceral adiposity, and atherosclerotic burden, supporting its value for stratifying cardiometabolic risk [\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e]. Large-scale epidemiological studies have further incorporated HSI with cardiometabolic factors to define MASLD-related phenotypes and have reported positive associations between higher HSI and risks of cardiovascular disease, type 2 diabetes, and all-cause mortality [\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e]. Collectively, these findings suggest that HSI reflects not only hepatic steatosis but also broader metabolic dysfunction.\u003c/p\u003e \u003cp\u003eIn the context of chronic high-altitude hypoxia, the components captured by HSI may be particularly relevant. Hypoxia has been reported to alter lipid handling through HIF-related pathways and impaired oxidative metabolism, potentially favoring intrahepatic lipid accumulation [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. In addition, sympathetic activation and systemic inflammation at high altitude may exacerbate insulin resistance and hepatocellular stress, which can influence transaminase levels [\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e]. These altitude-related physiological stressors may amplify the clinical signal captured by HSI in predicting MASLD risk.\u003c/p\u003e \u003cp\u003eConsistent with this interpretation, SHAP ranked HSI among the leading predictors, indicating a substantial contribution to the model output. Given that HSI relies on routinely available clinical variables and does not require imaging, it may serve as a pragmatic indicator for MASLD screening and risk stratification in resource-limited high-altitude settings, including remote military and field engineering sites.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e4.3 The role of high-altitude hematological adaptation markers (RBC, HGB, and altitude) in MASLD risk prediction\u003c/h2\u003e \u003cp\u003eIn the present model, altitude, RBC count, and hemoglobin emerged as important predictors, suggesting that altitude-related hematological adaptation may be informative for MASLD risk stratification in lowlanders with long-term high-altitude exposure. Chronic exposure to high altitude (typically\u0026thinsp;\u0026ge;\u0026thinsp;2,500 m) stimulates erythropoietin-driven erythropoiesis, resulting in sustained increases in RBC count and hemoglobin as a compensatory response to maintain oxygen delivery [\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e]. However, excessive erythrocytosis may increase blood viscosity and vascular resistance, potentially compromising microcirculatory perfusion, including within hepatic sinusoids, and thereby aggravating metabolic stress. These findings are consistent with epidemiological evidence suggesting that erythrocyte-related indices may be associated with fatty liver disease. Analyses including NHANES-derived cohorts have reported associations between higher RBC/hemoglobin levels and NAFLD risk after adjustment for conventional metabolic factors [\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e, \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e]. Potential explanations include hemoconcentration-related iron dysregulation and oxidative stress [\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e], which may contribute to hepatic injury and metabolic perturbations.\u003c/p\u003e \u003cp\u003eIn high-altitude settings, maladaptive responses to chronic hypoxia can progress toward syndromes such as chronic mountain sickness, which are characterized by pronounced erythrocytosis, hyperviscosity, and thrombotic risk [\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e]. In our model, altitude served as an exposure-related variable and showed a positive contribution to predicted risk, potentially reflecting the cumulative physiological burden of chronic hypoxia. Together, these observations support the inclusion of altitude and erythrocyte-related indices as accessible markers that may improve MASLD risk assessment in high-altitude-exposed lowlanders [\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eOverall, RBC count, hemoglobin, and altitude may capture clinically relevant aspects of acclimatization and exposure intensity. Incorporating these indicators into prediction frameworks may help identify individuals who warrant closer metabolic evaluation and follow-up in high-altitude environments.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section2\"\u003e \u003ch2\u003e4.4 The role of cardiometabolic indicators (SBP) in MASLD and their amplification in high-altitude environments\u003c/h2\u003e \u003cp\u003eIn our cohort, SBP was positively associated with MASLD risk and contributed substantially to model prediction. This finding is consistent with evidence showing a close relationship between steatotic liver disease and hypertension/cardiovascular disease. Meta-analyses and cohort studies have reported that individuals with NAFLD have a higher risk of incident hypertension; for example, a meta-analysis including more than 46,000 participants estimated an approximately 1.5\u0026ndash;1.6-fold increased risk [\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e]. Conversely, hypertension has also been associated with an increased risk of NAFLD in population-based analyses [\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e]. In addition, NAFLD/MASLD has been recognized as a risk factor for atherosclerotic cardiovascular disease (ASCVD), and its prognostic relevance for major adverse cardiovascular events and mortality has been documented in prospective studies and reviews [\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eSeveral biological pathways may underlie the co-occurrence of elevated blood pressure and MASLD, including insulin resistance, chronic low-grade inflammation, endothelial dysfunction, activation of the renin\u0026ndash;angiotensin\u0026ndash;aldosterone system (RAAS), and sympathetic nervous system activity [\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e]. Hepatic steatosis and metabolic dysfunction may contribute to vascular stiffness and blood pressure elevation through inflammatory signaling and oxidative stress, while sustained hypertension may impair hepatic microcirculation and promote endothelial injury, potentially accelerating the transition from steatosis to steatohepatitis and fibrosis [\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e, \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e]. Together, these observations support a clinically relevant liver\u0026ndash;cardiovascular interaction.\u003c/p\u003e \u003cp\u003eHigh-altitude hypoxia may further accentuate this interplay. Hypoxia is a potent trigger of sympathetic activation and may increase vasomotor tone and blood pressure. Prior studies have reported higher SBP and peripheral vascular resistance in lowlanders relocating to high altitude as well as in high-altitude residents [\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e, \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e]. In addition, altitude-related erythrocytosis and increased blood viscosity may increase vascular resistance and cardiac workload, potentially compounding blood pressure\u0026ndash;related metabolic stress.\u003c/p\u003e \u003cp\u003eIn our study, the identification of SBP as a key predictor suggests that cardiometabolic burden is an important component of MASLD risk stratification in high-altitude-exposed lowlanders. These findings support the practical value of incorporating blood pressure assessment into screening and follow-up strategies for MASLD in high-altitude settings, alongside broader cardiovascular risk evaluation.\u003c/p\u003e \u003cp\u003e \u003cb\u003e4.5 The role of inflammatory markers (WBC) in MASLD risk prediction and their amplification in high-altitude environments\u003c/b\u003e \u003c/p\u003e \u003cp\u003eChronic low-grade inflammation is widely recognized as a key feature of MASLD pathophysiology. WBC count is an accessible peripheral marker of systemic inflammation and has been associated with NAFLD/MASLD risk in multiple studies. In a prospective cohort of more than 26,000 Han Chinese adults, higher baseline WBC predicted incident NAFLD after adjustment for adiposity, lipid profiles, and lifestyle factors [\u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e]. Consistently, WBC levels have been reported to be higher in individuals with hepatic steatosis and to increase with steatosis severity [\u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e66\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eRecent genetic analyses further support a potential contributory role of inflammatory and hematological traits. A Mendelian randomization study published in 2024 reported genetic evidence linking hematological traits, including WBC and hemoglobin, with NAFLD risk [\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e]. In addition, immunometabolic studies highlight that innate immune activation\u0026mdash;such as neutrophil and monocyte recruitment and Kupffer cell activation\u0026mdash;contributes to progression from steatosis to MASH and fibrosis [\u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e], processes that may be partially captured by peripheral leukocyte indices.\u003c/p\u003e \u003cp\u003eIn our cohort, WBC was retained among the final predictors and contributed to the model output in SHAP analysis, underscoring the relevance of systemic inflammation in risk stratification. High-altitude hypoxia may further influence inflammatory tone: hypoxic stress can activate HIF-1α\u0026ndash;related pathways and promote pro-inflammatory cytokine signaling, potentially enhancing leukocyte recruitment and sterile inflammation [\u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e69\u003c/span\u003e]. Moreover, oxidative stress and sympathetic activation during hypoxic exposure may modulate immune activity and metabolic homeostasis [\u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e]. Together, these data suggest that inflammatory burden\u0026mdash;reflected in part by WBC\u0026mdash;may be particularly pertinent when assessing MASLD risk in high-altitude-exposed lowlanders.\u003c/p\u003e \u003cp\u003eGiven its low cost and routine availability, WBC may serve as a pragmatic marker to complement metabolic indicators in early screening and risk stratification in resource-limited high-altitude settings. However, WBC is non-specific and may be influenced by infection, smoking, hydration status, and other stressors; therefore, residual confounding cannot be excluded. Future studies, particularly prospective cohorts and external validations, are needed to clarify causal pathways and to determine whether targeting hypoxia-related stress or inflammation improves MASLD-related outcomes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003e4.6 Strengths and clinical utility of the machine learning model\u003c/h2\u003e \u003cp\u003eCompared with prior MASLD-related prediction studies, our model offers several practical advantages for use in resource-limited high-altitude settings. Although a growing number of machine learning (ML) models for MASLD/NAFLD have been reported, many rely on inputs that are difficult to implement at scale in remote regions. For example, Chen et al. developed ML models in individuals with hypertension and prehypertension and reported an AUC of 0.889, with ALT, BMI, waist circumference, and HDL-C among the key predictors [\u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e]. Other studies have incorporated transcriptomic, metabolomic, lipidomic, or other novel biomarkers to improve predictive performance [\u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e72\u003c/span\u003e]; however, these approaches are often costly and may not be feasible for population-level screening in high-altitude areas with limited infrastructure. Similarly, Rohit et al. proposed a non-invasive score combining liver enzymes and MRI-derived parameters (AUC\u0026thinsp;=\u0026thinsp;0.81) [\u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e73\u003c/span\u003e], but MRI availability and cost constrain routine deployment in field settings. Moreover, although some models have focused on specific comorbid populations such as diabetes or cardiovascular disease [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e], few have addressed early, non-invasive risk stratification in lowlanders with long-term high-altitude exposure, including military personnel, construction workers, and migrants.\u003c/p\u003e \u003cp\u003eIn our cohort, the final model achieved good discrimination in the validation set (AUC\u0026thinsp;=\u0026thinsp;0.898), consistent with the performance range commonly reported in general or metabolically enriched populations. This performance may be attributable to: (1) a structured, three-stage feature selection strategy that improved predictor robustness; (2) inclusion of HSI, which integrates adiposity and liver injury signals and may enhance clinical interpretability; and (3) the distinctive physiological and metabolic profile associated with prolonged high-altitude exposure, which may be captured by routinely measured indicators.\u003c/p\u003e \u003cp\u003eFrom a clinical and public health perspective, the model uses only routinely collected examination and laboratory variables, supporting scalable and non-invasive screening in high-altitude regions where imaging and specialist resources are scarce. In addition, SHAP-based explanations provide transparent, individualized contributions of predictors to model output, which may facilitate risk communication and guide follow-up intensity. Finally, the accompanying web-based dynamic calculator improves accessibility and usability, potentially enabling integration into health management programs for high-altitude-exposed lowlanders, while acknowledging that external validation is needed before broad implementation.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section2\"\u003e \u003ch2\u003e4.7 Limitations and future perspectives\u003c/h2\u003e \u003cp\u003eSeveral limitations should be considered when interpreting our findings. First, this was a single-center retrospective study, which may introduce selection bias and limit generalizability. Future multicenter, prospective studies are needed to validate model performance across diverse settings and populations. Second, MASLD was ascertained by abdominal ultrasonography rather than liver biopsy. Although biopsy remains the reference standard, ultrasonography is non-invasive, widely available, and suitable for large-scale screening; however, some degree of misclassification, especially for mild steatosis, cannot be excluded. Third, although altitude and exposure duration were captured, more granular indicators of hypoxic burden (e.g., ambient partial pressure of oxygen, oxygen saturation [SpO₂], and acclimatization status) were not available and should be incorporated in future studies to refine exposure characterization and improve model transportability. Fourth, the cohort was predominantly male (95.6%), reflecting the occupational composition of high-altitude deployment and construction work in our setting. Consequently, model applicability to females remains uncertain and warrants dedicated validation in female and more sex-balanced cohorts. Finally, we did not include an independent external validation cohort. External validation using geographically and demographically distinct datasets is essential before broader clinical implementation, and future work should also evaluate prospective clinical impact and calibration drift over time.\u003c/p\u003e \u003c/div\u003e"},{"header":"5. Conclusions","content":"\u003cp\u003eThis study developed and internally validated a machine learning\u0026ndash;based model to predict MASLD risk in lowlanders with long-term high-altitude exposure using routinely available clinical variables. The final LR model demonstrated good discrimination and calibration, and SHAP-based interpretation highlighted the relative contributions of metabolic factors (e.g., BMI, TG, HDL-C, and HSI), blood pressure, and altitude/hematological indices to predicted risk. Given its reliance on low-cost, routinely collected data and its interpretability, the model may support scalable screening and risk stratification in resource-limited high-altitude settings. The accompanying web-based dynamic calculator further facilitates real-world use among high-altitude-exposed populations such as military personnel, construction workers, and long-term residents. External validation in independent and more diverse cohorts is warranted before broader implementation.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eA/G, albumin/globulin ratio\u003c/p\u003e\n\u003cp\u003eALB, albumin\u003c/p\u003e\n\u003cp\u003eALT, alanine aminotransferase\u003c/p\u003e\n\u003cp\u003eAPTT, activated partial thromboplastin time\u003c/p\u003e\n\u003cp\u003eAST, aspartate aminotransferase\u003c/p\u003e\n\u003cp\u003eASCVD, atherosclerotic cardiovascular disease\u003c/p\u003e\n\u003cp\u003eAUC, area under the receiver operating characteristic curve\u003c/p\u003e\n\u003cp\u003eBMI, body mass index\u003c/p\u003e\n\u003cp\u003eBoruta, Boruta feature selection algorithm\u003c/p\u003e\n\u003cp\u003eCr, creatinine\u003c/p\u003e\n\u003cp\u003eDCA, decision curve analysis\u003c/p\u003e\n\u003cp\u003eDBIL, direct bilirubin\u003c/p\u003e\n\u003cp\u003eDBP, diastolic blood pressure\u003c/p\u003e\n\u003cp\u003eDT, decision tree\u003c/p\u003e\n\u003cp\u003eFIB, fibrinogen\u003c/p\u003e\n\u003cp\u003eFLI, Fatty Liver Index\u003c/p\u003e\n\u003cp\u003eGLB, globulin\u003c/p\u003e\n\u003cp\u003eHbA1c, glycated hemoglobin\u003c/p\u003e\n\u003cp\u003eHDL-C, high-density lipoprotein cholesterol\u003c/p\u003e\n\u003cp\u003eHGB, hemoglobin\u003c/p\u003e\n\u003cp\u003eHIF, hypoxia-inducible factor\u003c/p\u003e\n\u003cp\u003eHIF-1\u0026alpha;, hypoxia-inducible factor-1 alpha\u003c/p\u003e\n\u003cp\u003eHSI, hepatic steatosis index\u003c/p\u003e\n\u003cp\u003eIBIL, indirect bilirubin\u003c/p\u003e\n\u003cp\u003eINR, international normalized ratio\u003c/p\u003e\n\u003cp\u003eIQR, interquartile range\u003c/p\u003e\n\u003cp\u003eLASSO, least absolute shrinkage and selection operator\u003c/p\u003e\n\u003cp\u003eLightGBM, Light Gradient Boosting Machine\u003c/p\u003e\n\u003cp\u003eLR, logistic regression\u003c/p\u003e\n\u003cp\u003eLYMPH, lymphocyte count\u003c/p\u003e\n\u003cp\u003eMAFLD, metabolic dysfunction\u0026ndash;associated fatty liver disease\u003c/p\u003e\n\u003cp\u003eMASH, metabolic dysfunction\u0026ndash;associated steatohepatitis\u003c/p\u003e\n\u003cp\u003eMASLD, metabolic dysfunction\u0026ndash;associated steatotic liver disease\u003c/p\u003e\n\u003cp\u003eMissForest, random forest\u0026ndash;based imputation algorithm (MissForest)\u003c/p\u003e\n\u003cp\u003eML, machine learning\u003c/p\u003e\n\u003cp\u003eMONO, monocyte count\u003c/p\u003e\n\u003cp\u003eNAFLD, non-alcoholic fatty liver disease\u003c/p\u003e\n\u003cp\u003eNEUT, neutrophil count\u003c/p\u003e\n\u003cp\u003eNPV, negative predictive value\u003c/p\u003e\n\u003cp\u003ePLT, platelet count\u003c/p\u003e\n\u003cp\u003ePPV, positive predictive value\u003c/p\u003e\n\u003cp\u003ePT, prothrombin time\u003c/p\u003e\n\u003cp\u003ePTA, prothrombin activity\u003c/p\u003e\n\u003cp\u003eRAAS, renin\u0026ndash;angiotensin\u0026ndash;aldosterone system\u003c/p\u003e\n\u003cp\u003eRBC, red blood cell count\u003c/p\u003e\n\u003cp\u003eRF, random forest\u003c/p\u003e\n\u003cp\u003eROC, receiver operating characteristic\u003c/p\u003e\n\u003cp\u003eSBP, systolic blood pressure\u003c/p\u003e\n\u003cp\u003eSD, standard deviation\u003c/p\u003e\n\u003cp\u003eSHAP, Shapley Additive exPlanations\u003c/p\u003e\n\u003cp\u003eSVM, support vector machine\u003c/p\u003e\n\u003cp\u003eTBIL, total bilirubin\u003c/p\u003e\n\u003cp\u003eTC, total cholesterol\u003c/p\u003e\n\u003cp\u003eTG, triglycerides\u003c/p\u003e\n\u003cp\u003eTP, total protein\u003c/p\u003e\n\u003cp\u003eTT, thrombin time\u003c/p\u003e\n\u003cp\u003eVLDL, very-low-density lipoprotein\u003c/p\u003e\n\u003cp\u003eWBC, white blood cell count\u003c/p\u003e\n\u003cp\u003eXGBoost, Extreme Gradient Boosting\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate.\u0026nbsp;\u003c/strong\u003eThis study received approval from the Ethics Committee of the General Hospital of Western Theater Command and was conducted in accordance with the principles of the Declaration of Helsinki.The ethical approval number is 2024EC4-ky024. Due to the retrospective design of the study and the analysis of anonymized data, the Ethics Committee granted a waiver for informed consent.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets analysed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was supported by Hospital Management of the General Hospital of Western Theater Command (2024-YGLC-A01).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors above appropriately contributed to the development of this manuscript. The conceptualization of the aims of the article was made by YZ and RD. The formal acquisition and analysis of the data were carried out by JF, QW, and HD. YZ, YH and XL were involved in interpreting the data, drafting, revising, and approving the final version for submission. RD was responsible for funding acquisition and supervision. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors acknowledge all the clinical and research staff from the research centers.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eBoldys A, Buldak L. Metabolic dysfunction-associated steatotic liver disease. Navigating terminological evolution, diagnostic frontiers and therapeutic horizon-an editorial exploration. \u003cem\u003eWorld J Gastroenterol. \u003c/em\u003e2024; 30\u003cstrong\u003e:\u003c/strong\u003e2387-90.\u003c/li\u003e\n\u003cli\u003eRinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multisociety Delphi consensus statement on new fatty liver disease nomenclature. \u003cem\u003eJ Hepatol. \u003c/em\u003e2023; 79\u003cstrong\u003e:\u003c/strong\u003e1542-56.\u003c/li\u003e\n\u003cli\u003eEl-Kassas M, Awad A, Elbadry M, Arab JP. Tailored Model of Care for Patients with Metabolic Dysfunction-Associated Steatotic Liver Disease. \u003cem\u003eSemin Liver Dis. \u003c/em\u003e2024; 44\u003cstrong\u003e:\u003c/strong\u003e54-68.\u003c/li\u003e\n\u003cli\u003eDeprince A, Hennuyer N, Kooijman S, Pronk ACM, Baug\u0026eacute; E, Lienard V, et al. Apolipoprotein F is reduced in humans with steatosis and controls plasma triglyceride-rich lipoprotein metabolism. \u003cem\u003eHepatology. \u003c/em\u003e2023; 77\u003cstrong\u003e:\u003c/strong\u003e1287-302.\u003c/li\u003e\n\u003cli\u003eFriedman SL, Neuschwander-Tetri BA, Rinella M, Sanyal AJ. Mechanisms of NAFLD development and therapeutic strategies. \u003cem\u003eNat Med. \u003c/em\u003e2018; 24\u003cstrong\u003e:\u003c/strong\u003e908-22.\u003c/li\u003e\n\u003cli\u003ePowell EE, Wong VW, Rinella M. Non-alcoholic fatty liver disease. \u003cem\u003eLancet. \u003c/em\u003e2021; 397\u003cstrong\u003e:\u003c/strong\u003e2212-24.\u003c/li\u003e\n\u003cli\u003eCariou B. The metabolic triad of non-alcoholic fatty liver disease, visceral adiposity and type 2 diabetes. Implications for treatment. \u003cem\u003eDiabetes Obes Metab. \u003c/em\u003e2022; 24 Suppl 2\u003cstrong\u003e:\u003c/strong\u003e15-27.\u003c/li\u003e\n\u003cli\u003eTymko MM, Lawley JS, Ainslie PN, Hansen AB, Hofstaetter F, Rainer S, et al. Global Reach 2018 Heightened \u0026alpha;-Adrenergic Signaling Impairs Endothelial Function During Chronic Exposure to Hypobaric Hypoxia. \u003cem\u003eCirc Res. \u003c/em\u003e2020; 127\u003cstrong\u003e:\u003c/strong\u003ee1-13.\u003c/li\u003e\n\u003cli\u003eWest JB. High-altitude medicine. \u003cem\u003eAm J Respir Crit Care Med. \u003c/em\u003e2012; 186\u003cstrong\u003e:\u003c/strong\u003e1229-37.\u003c/li\u003e\n\u003cli\u003eHuang L. High altitude medicine in China in the 21st century. opportunities and challenges. \u003cem\u003eMil Med Res. \u003c/em\u003e2014; 1\u003cstrong\u003e:\u003c/strong\u003e17.\u003c/li\u003e\n\u003cli\u003eWu TY, Ding SQ, Liu JL, Yu MT, Jia JH, Chai ZC, et al. Who should not go high: chronic disease and work at altitude during construction of the Qinghai-Tibet railroad. \u003cem\u003eHigh Alt Med Biol. \u003c/em\u003e2007; 8\u003cstrong\u003e:\u003c/strong\u003e88-107.\u003c/li\u003e\n\u003cli\u003eZhang W, Jiao L, Liu R, Zhang Y, Ji Q, Zhang H, et al. The effect of exposure to high altitude and low oxygen on intestinal microbial communities in mice. \u003cem\u003ePLoS One. \u003c/em\u003e2018; 13\u003cstrong\u003e:\u003c/strong\u003ee0203701.\u003c/li\u003e\n\u003cli\u003eBasnyat B, Murdoch DR. High-altitude illness. \u003cem\u003eLancet. \u003c/em\u003e2003; 361\u003cstrong\u003e:\u003c/strong\u003e1967-74.\u003c/li\u003e\n\u003cli\u003eSiebenmann C, Rasmussen P, Hug M, Keiser S, Fl\u0026uuml;ck D, Fisher JP, et al. Parasympathetic withdrawal increases heart rate after 2 weeks at 3454 m altitude. \u003cem\u003eJ Physiol. \u003c/em\u003e2017; 595\u003cstrong\u003e:\u003c/strong\u003e1619-26.\u003c/li\u003e\n\u003cli\u003eSavla JJ, Levine BD, Sadek HA. The Effect of Hypoxia on Cardiovascular Disease. Friend or Foe? \u003cem\u003eHigh Alt Med Biol. \u003c/em\u003e2018; 19\u003cstrong\u003e:\u003c/strong\u003e124-30.\u003c/li\u003e\n\u003cli\u003eMurray AJ, Horscroft JA. Mitochondrial function at extreme high altitude. \u003cem\u003eJ Physiol. \u003c/em\u003e2016; 594\u003cstrong\u003e:\u003c/strong\u003e1137-49.\u003c/li\u003e\n\u003cli\u003eMesarwi OA, Loomba R, Malhotra A. Obstructive Sleep Apnea, Hypoxia, and Nonalcoholic Fatty Liver Disease. \u003cem\u003eAm J Respir Crit Care Med. \u003c/em\u003e2019; 199\u003cstrong\u003e:\u003c/strong\u003e830-41.\u003c/li\u003e\n\u003cli\u003eLefere S, Van Steenkiste C, Verhelst X, Van Vlierberghe H, Devisscher L, Geerts A. Hypoxia-regulated mechanisms in the pathogenesis of obesity and non-alcoholic fatty liver disease. \u003cem\u003eCell Mol Life Sci. \u003c/em\u003e2016; 73\u003cstrong\u003e:\u003c/strong\u003e3419-31.\u003c/li\u003e\n\u003cli\u003eSakurai A, Sakurai T, Ho HJ, Chiba H, Hui SP. Kaempferol Improves Cardiolipin and ATP in Hepatic Cells. A Cellular Model Perspective in the Context of Metabolic Dysfunction-Associated Steatotic Liver Disease. \u003cem\u003eNutrients. \u003c/em\u003e2024; 16(4):508.\u003c/li\u003e\n\u003cli\u003eSavransky V, Nanayakkara A, Li J, Bevans S, Smith PL, Rodriguez A, et al. Chronic intermittent hypoxia induces atherosclerosis. \u003cem\u003eAm J Respir Crit Care Med. \u003c/em\u003e2007; 175\u003cstrong\u003e:\u003c/strong\u003e1290-7.\u003c/li\u003e\n\u003cli\u003eChalasani N, Younossi Z, Lavine JE, Charlton M, Cusi K, Rinella M, et al. The diagnosis and management of nonalcoholic fatty liver disease. Practice guidance from the American Association for the Study of Liver Diseases. \u003cem\u003eHepatology. \u003c/em\u003e2018; 67\u003cstrong\u003e:\u003c/strong\u003e328-57.\u003c/li\u003e\n\u003cli\u003eIwaki M, Fujii H, Hayashi H, Toyoda H, Oeda S, Hyogo H, et al. Prognosis of biopsy-confirmed metabolic dysfunction- associated steatotic liver disease. A sub-analysis of the CLIONE study. \u003cem\u003eClin Mol Hepatol. \u003c/em\u003e2024; 30\u003cstrong\u003e:\u003c/strong\u003e225-34.\u003c/li\u003e\n\u003cli\u003eAbdelhameed F, Kite C, Lagojda L, Dallaway A, Chatha KK, Chaggar SS, et al. Non-invasive Scores and Serum Biomarkers for Fatty Liver in the Era of Metabolic Dysfunction-associated Steatotic Liver Disease (MASLD). A Comprehensive Review From NAFLD to MAFLD and MASLD. \u003cem\u003eCurr Obes Rep. \u003c/em\u003e2024; 13\u003cstrong\u003e:\u003c/strong\u003e510-31.\u003c/li\u003e\n\u003cli\u003eDeng J, Ji W, Liu H, Li L, Wang Z, Hu Y, et al. Development and validation of a machine learning-based framework for assessing metabolic-associated fatty liver disease risk. \u003cem\u003eBMC Public Health. \u003c/em\u003e2024; 24\u003cstrong\u003e:\u003c/strong\u003e2545.\u003c/li\u003e\n\u003cli\u003eMasaebi F, Azizmohammad Looha M, Mohammadzadeh M, Pahlevani V, Farjam M, Zayeri F, et al. Machine-Learning Application for Predicting Metabolic Dysfunction-Associated Steatotic Liver Disease Using Laboratory and Body Composition Indicators. \u003cem\u003eArch Iran Med. \u003c/em\u003e2024; 27\u003cstrong\u003e:\u003c/strong\u003e551-62.\u003c/li\u003e\n\u003cli\u003eNabrdalik K, Kwiendacz H, Irlik K, Hendel M, Drożdż K, Wijata AM, et al. Machine Learning Identifies Metabolic Dysfunction-Associated Steatotic Liver Disease in Patients With Diabetes Mellitus. \u003cem\u003eJ Clin Endocrinol Metab. \u003c/em\u003e2024; 109\u003cstrong\u003e:\u003c/strong\u003e2029-38.\u003c/li\u003e\n\u003cli\u003eCanivet CM, Boursier J, Loomba R. New Nomenclature for Nonalcoholic Fatty Liver Disease. Understanding Metabolic Dysfunction-Associated Steatotic Liver Disease, Metabolic Dysfunction- and Alcohol-Associated Liver Disease, and Their Implications in Clinical Practice. \u003cem\u003eSemin Liver Dis. \u003c/em\u003e2024; 44\u003cstrong\u003e:\u003c/strong\u003e35-42.\u003c/li\u003e\n\u003cli\u003eHong S, Lynn HS. Accuracy of random-forest-based imputation of missing data in the presence of non-normality, non-linearity, and interaction. \u003cem\u003eBMC Med Res Methodol. \u003c/em\u003e2020; 20\u003cstrong\u003e:\u003c/strong\u003e199.\u003c/li\u003e\n\u003cli\u003eLee Y, Leite WL. A comparison of random forest-based missing imputation methods for covariates in propensity score analysis. \u003cem\u003ePsychol Methods. \u003c/em\u003e2024; doi: 10.1037/met0000676.\u003c/li\u003e\n\u003cli\u003ePfeifer B, Holzinger A, Schimek MG. Robust Random Forest-Based All-Relevant Feature Ranks for Trustworthy AI. \u003cem\u003eStud Health Technol Inform. \u003c/em\u003e2022; 294\u003cstrong\u003e:\u003c/strong\u003e137-8.\u003c/li\u003e\n\u003cli\u003eLi W, Zhou Q, Liu W, Xu C, Tang ZR, Dong S, et al. A Machine Learning-Based Predictive Model for Predicting Lymph Node Metastasis in Patients With Ewing\u0026apos;s Sarcoma. \u003cem\u003eFront Med (Lausanne). \u003c/em\u003e2022; 9\u003cstrong\u003e:\u003c/strong\u003e832108.\u003c/li\u003e\n\u003cli\u003eEjiyi CJ, Qin Z, Ukwuoma CC, Nneji GU, Monday HN, Ejiyi MB, et al. Comparative performance analysis of Boruta, SHAP, and Borutashap for disease diagnosis. A study with multiple machine learning algorithms. \u003cem\u003eNetwork. \u003c/em\u003e2025; 36\u003cstrong\u003e:\u003c/strong\u003e507-44.\u003c/li\u003e\n\u003cli\u003eNing Y, Ong MEH, Chakraborty B, Goldstein BA, Ting DSW, Vaughan R, et al. Shapley variable importance cloud for interpretable machine learning. \u003cem\u003ePatterns (N Y). \u003c/em\u003e2022; 3\u003cstrong\u003e:\u003c/strong\u003e100452.\u003c/li\u003e\n\u003cli\u003eHe W, Xu C, Wang X, Lei J, Qiu Q, Hu Y, et al. Development and validation of a risk prediction nomogram for in-stent restenosis in patients undergoing percutaneous coronary intervention. \u003cem\u003eBMC Cardiovasc Disord. \u003c/em\u003e2021; 21\u003cstrong\u003e:\u003c/strong\u003e435.\u003c/li\u003e\n\u003cli\u003eZhu C, You Y, Liu S, Ji Y, Yu J. A Nomogram to Predict Distant Metastasis for Patients with Esophageal Cancer. \u003cem\u003eOncol Res Treat. \u003c/em\u003e2020; 43\u003cstrong\u003e:\u003c/strong\u003e2-9.\u003c/li\u003e\n\u003cli\u003eLiu P, Kuang Y, Ye L, Peng C, Chen W, Shen M, et al. Predicting the Risk of Psoriatic Arthritis in Plaque Psoriasis Patients. Development and Assessment of a New Predictive Nomogram. \u003cem\u003eFront Immunol. \u003c/em\u003e2021; 12\u003cstrong\u003e:\u003c/strong\u003e740968.\u003c/li\u003e\n\u003cli\u003eFu X, Fu Y, Liu Y, Guo J, Liu Y, Guo Y, et al. Genetic polymorphisms of 26 Y-STR loci in the Mongolian minority from Horqin district, China. \u003cem\u003eInt J Legal Med. \u003c/em\u003e2016; 130\u003cstrong\u003e:\u003c/strong\u003e941-6.\u003c/li\u003e\n\u003cli\u003eHuang CF, Dai CY, Lin YH, Wang CW, Jang TY, Liang PC, et al. Dynamic change of metabolic dysfunction-associated steatotic liver disease in chronic hepatitis C patients after viral eradication. A nationwide registry study in Taiwan. \u003cem\u003eClin Mol Hepatol. \u003c/em\u003e2024; 30\u003cstrong\u003e:\u003c/strong\u003e883-94.\u003c/li\u003e\n\u003cli\u003eZhang W, Song WJ, Chen W, Pan Z, Zhang J, Fan L, et al. Metabolic dysfunction-associated steatotic liver disease-related hepatic fibrosis increases risk of insulin resistance, type 2 diabetes, and chronic kidney disease. \u003cem\u003eEur J Gastroenterol Hepatol. \u003c/em\u003e2024; 36\u003cstrong\u003e:\u003c/strong\u003e802-10.\u003c/li\u003e\n\u003cli\u003eFlores YN, Amoon AT, Su B, Velazquez-Cruz R, Ram\u0026iacute;rez-Palacios P, Salmer\u0026oacute;n J, et al. Serum lipids are associated with nonalcoholic fatty liver disease. a pilot case-control study in Mexico. \u003cem\u003eLipids Health Dis. \u003c/em\u003e2021; 20\u003cstrong\u003e:\u003c/strong\u003e136.\u003c/li\u003e\n\u003cli\u003eHou X, Guan Y, Tang Y, Song A, Zhao J, Ren L, et al. A correlation study of the relationships between nonalcoholic fatty liver disease and serum triglyceride concentration after an oral fat tolerance test. \u003cem\u003eLipids Health Dis. \u003c/em\u003e2021; 20\u003cstrong\u003e:\u003c/strong\u003e54.\u003c/li\u003e\n\u003cli\u003ePeng K, Mo Z, Tian G. Serum Lipid Abnormalities and Nonalcoholic Fatty Liver Disease in Adult Males. \u003cem\u003eAm J Med Sci. \u003c/em\u003e2017; 353\u003cstrong\u003e:\u003c/strong\u003e236-41.\u003c/li\u003e\n\u003cli\u003eIsaza SC, Del Pozo-Maroto E, Dom\u0026iacute;nguez-Alc\u0026oacute;n L, Elbouayadi L, Gonz\u0026aacute;lez-Rodr\u0026iacute;guez \u0026Aacute;, Garc\u0026iacute;a-Monz\u0026oacute;n C. Hypoxia and Non-alcoholic Fatty Liver Disease. \u003cem\u003eFront Med (Lausanne). \u003c/em\u003e2020; 7\u003cstrong\u003e:\u003c/strong\u003e578001.\u003c/li\u003e\n\u003cli\u003eYu L, Wang H, Han X, Liu H, Zhu D, Feng W, et al. Oxygen therapy alleviates hepatic steatosis by inhibiting hypoxia-inducible factor-2\u0026alpha;. \u003cem\u003eJ Endocrinol. \u003c/em\u003e2020; 246\u003cstrong\u003e:\u003c/strong\u003e57-67.\u003c/li\u003e\n\u003cli\u003eLee JH, Kim D, Kim HJ, Lee CH, Yang JI, Kim W, et al. Hepatic steatosis index. a simple screening tool reflecting nonalcoholic fatty liver disease. \u003cem\u003eDig Liver Dis. \u003c/em\u003e2010; 42\u003cstrong\u003e:\u003c/strong\u003e503-8.\u003c/li\u003e\n\u003cli\u003eCastera L, Friedrich-Rust M, Loomba R. Noninvasive Assessment of Liver Disease in Patients With Nonalcoholic Fatty Liver Disease. \u003cem\u003eGastroenterology. \u003c/em\u003e2019; 156\u003cstrong\u003e:\u003c/strong\u003e1264-81.\u003c/li\u003e\n\u003cli\u003eSacks D, Baxter B, Campbell BCV, Carpenter JS, Cognard C, Dippel D, et al. Multisociety Consensus Quality Improvement Revised Consensus Statement for Endovascular Therapy of Acute Ischemic Stroke. \u003cem\u003eInt J Stroke. \u003c/em\u003e2018; 13\u003cstrong\u003e:\u003c/strong\u003e612-32.\u003c/li\u003e\n\u003cli\u003eChen X, Du L, Peng J. Association between estimated glucose disposal rate and cardiovascular disease prevalence and mortality outcomes in metabolic dysfunction-associated steatotic liver disease. a comparative analysis of insulin resistance markers. \u003cem\u003eJ Glob Health. \u003c/em\u003e2025; 15\u003cstrong\u003e:\u003c/strong\u003e04249.\u003c/li\u003e\n\u003cli\u003eTian M, Zhang Y, Liu Z, Zhao N, Huang D, Hou J, et al. A new investigation of nonalcoholic fatty liver disease. Effects of hypoxia on mitochondrial function and lipid droplet autophagy. \u003cem\u003ePLoS One. \u003c/em\u003e2025; 20\u003cstrong\u003e:\u003c/strong\u003ee0331305.\u003c/li\u003e\n\u003cli\u003eVillafuerte FC, Simonson TS, Bermudez D, Le\u0026oacute;n-Velarde F. High-Altitude Erythrocytosis. Mechanisms of Adaptive and Maladaptive Responses. \u003cem\u003ePhysiology (Bethesda). \u003c/em\u003e2022; 37\u003cstrong\u003e:\u003c/strong\u003e0.\u003c/li\u003e\n\u003cli\u003eZhong F, Guan L, Lin H, Zhao M, Qin Y, Li Q, et al. Red Blood Cell Count. An Unrecognized Risk Factor for Nonalcoholic Fatty Liver Disease. \u003cem\u003eFront Endocrinol (Lausanne). \u003c/em\u003e2021; 12\u003cstrong\u003e:\u003c/strong\u003e760981.\u003c/li\u003e\n\u003cli\u003eDai X, Zhou G, Xu L. Associations between red blood cell count and metabolic dysfunction-associated fatty liver disease(MAFLD). \u003cem\u003ePLoS One. \u003c/em\u003e2022; 17\u003cstrong\u003e:\u003c/strong\u003ee0279274.\u003c/li\u003e\n\u003cli\u003eChung GE, Yim JY, Kim D, Kwak MS, Yang JI, Chung SJ, et al. Associations between hemoglobin concentrations and the development of incidental metabolic syndrome or nonalcoholic fatty liver disease. \u003cem\u003eDig Liver Dis. \u003c/em\u003e2017; 49\u003cstrong\u003e:\u003c/strong\u003e57-62.\u003c/li\u003e\n\u003cli\u003eSwenson ER. Chronic Mountain Sickness Evolving Over Time. New Data From on High. \u003cem\u003eChest. \u003c/em\u003e2022; 161\u003cstrong\u003e:\u003c/strong\u003e1136-7.\u003c/li\u003e\n\u003cli\u003eZhu N, Wang X, Zhu H, Zheng Y. Blood cell parameters and risk of nonalcoholic fatty liver disease. a comprehensive Mendelian randomization study. \u003cem\u003eBMC Med Genomics. \u003c/em\u003e2024; 17\u003cstrong\u003e:\u003c/strong\u003e102.\u003c/li\u003e\n\u003cli\u003eCiardullo S, Grassi G, Mancia G, Perseghin G. Nonalcoholic fatty liver disease and risk of incident hypertension. a systematic review and meta-analysis. \u003cem\u003eEur J Gastroenterol Hepatol. \u003c/em\u003e2022; 34\u003cstrong\u003e:\u003c/strong\u003e365-71.\u003c/li\u003e\n\u003cli\u003eLi G, Peng Y, Chen Z, Li H, Liu D, Ye X. Bidirectional Association between Hypertension and NAFLD. A Systematic Review and Meta-Analysis of Observational Studies. \u003cem\u003eInt J Endocrinol. \u003c/em\u003e2022; 2022\u003cstrong\u003e:\u003c/strong\u003e8463640.\u003c/li\u003e\n\u003cli\u003eMantovani A, Csermely A, Petracca G, Beatrice G, Corey KE, Simon TG, et al. Non-alcoholic fatty liver disease and risk of fatal and non-fatal cardiovascular events. an updated systematic review and meta-analysis. \u003cem\u003eLancet Gastroenterol Hepatol. \u003c/em\u003e2021; 6\u003cstrong\u003e:\u003c/strong\u003e903-13.\u003c/li\u003e\n\u003cli\u003eDuell PB, Welty FK, Miller M, Chait A, Hammond G, Ahmad Z, et al. Nonalcoholic Fatty Liver Disease and Cardiovascular Risk. A Scientific Statement From the American Heart Association. \u003cem\u003eArterioscler Thromb Vasc Biol. \u003c/em\u003e2022; 42\u003cstrong\u003e:\u003c/strong\u003ee168-85.\u003c/li\u003e\n\u003cli\u003ePelusi S, Macchi C, Malvestiti F, Margarita S, De Matteis I, Periti G, et al. Interplay among lipoprotein(a), hepatic and vascular damage in individuals with metabolic dysfunction. \u003cem\u003eCardiovasc Diabetol. \u003c/em\u003e2025; 24\u003cstrong\u003e:\u003c/strong\u003e447.\u003c/li\u003e\n\u003cli\u003eDoumas SA, Tripathi S, Kashikar A, Khuttan A, Kumar A, Singh H, et al. Nonalcoholic Fatty Liver Disease (NAFLD) and Cardiovascular Risk. Is Imaging Helpful? \u003cem\u003eCurr Probl Cardiol. \u003c/em\u003e2024; 49\u003cstrong\u003e:\u003c/strong\u003e102065.\u003c/li\u003e\n\u003cli\u003eKtenopoulos N, Sagris M, Gerogianni M, Pamporis K, Apostolos A, Balampanis K, et al. Non-Alcoholic Fatty Liver Disease and Coronary Artery Disease. A Bidirectional Association Based on Endothelial Dysfunction. \u003cem\u003eInt J Mol Sci. \u003c/em\u003e2024; 25.\u003c/li\u003e\n\u003cli\u003eSimpson LL, Steinback CD, Stembridge M, Moore JP. A sympathetic view of blood pressure control at high altitude. new insights from microneurographic studies. \u003cem\u003eExp Physiol. \u003c/em\u003e2021; 106\u003cstrong\u003e:\u003c/strong\u003e377-84.\u003c/li\u003e\n\u003cli\u003eBilo G, Caravita S, Torlasco C, Parati G. Blood pressure at high altitude. physiology and clinical implications. \u003cem\u003eKardiol Pol. \u003c/em\u003e2019; 77\u003cstrong\u003e:\u003c/strong\u003e596-603.\u003c/li\u003e\n\u003cli\u003eWang S, Zhang C, Zhang G, Yuan Z, Liu Y, Ding L, et al. Association between white blood cell count and non-alcoholic fatty liver disease in urban Han Chinese. a prospective cohort study. \u003cem\u003eBMJ Open. \u003c/em\u003e2016; 6\u003cstrong\u003e:\u003c/strong\u003ee010342.\u003c/li\u003e\n\u003cli\u003eChao YL, Wu PY, Huang JC, Chiu YW, Lee JJ, Chen SC, et al. Hepatic Steatosis Is Associated with High White Blood Cell and Platelet Counts. \u003cem\u003eBiomedicines. \u003c/em\u003e2022; 10:892.\u003c/li\u003e\n\u003cli\u003eHu B, Wan AH, Xiang XQ, Wei YH, Chen Y, Tang Z, et al. Blood cell counts and nonalcoholic fatty liver disease. Evidence from Mendelian randomization analysis. \u003cem\u003eWorld J Hepatol. \u003c/em\u003e2024; 16\u003cstrong\u003e:\u003c/strong\u003e1145-55.\u003c/li\u003e\n\u003cli\u003ePaquissi FC. Immune Imbalances in Non-Alcoholic Fatty Liver Disease. From General Biomarkers and Neutrophils to Interleukin-17 Axis Activation and New Therapeutic Targets. \u003cem\u003eFront Immunol. \u003c/em\u003e2016; 7\u003cstrong\u003e:\u003c/strong\u003e490.\u003c/li\u003e\n\u003cli\u003eHern\u0026aacute;ndez A, Geng Y, Sep\u0026uacute;lveda R, Sol\u0026iacute;s N, Torres J, Arab JP, et al. Chemical hypoxia induces pro-inflammatory signals in fat-laden hepatocytes and contributes to cellular crosstalk with Kupffer cells through extracellular vesicles. \u003cem\u003eBiochim Biophys Acta Mol Basis Dis. \u003c/em\u003e2020; 1866\u003cstrong\u003e:\u003c/strong\u003e165753.\u003c/li\u003e\n\u003cli\u003eDrager LF, Polotsky VY, O\u0026apos;Donnell CP, Cravo SL, Lorenzi-Filho G, Machado BH. Translational approaches to understanding metabolic dysfunction and cardiovascular consequences of obstructive sleep apnea. \u003cem\u003eAm J Physiol Heart Circ Physiol. \u003c/em\u003e2015; 309\u003cstrong\u003e:\u003c/strong\u003eH1101-11.\u003c/li\u003e\n\u003cli\u003eChen C, Zhang W, Yan G, Tang C. Identifying metabolic dysfunction-associated steatotic liver disease in patients with hypertension and pre-hypertension. An interpretable machine learning approach. \u003cem\u003eDigit Health. \u003c/em\u003e2024; 10\u003cstrong\u003e:\u003c/strong\u003e20552076241233135.\u003c/li\u003e\n\u003cli\u003eHan N, He J, Shi L, Zhang M, Zheng J, Fan Y. Identification of biomarkers in nonalcoholic fatty liver disease. A machine learning method and experimental study. \u003cem\u003eFront Genet. \u003c/em\u003e2022; 13\u003cstrong\u003e:\u003c/strong\u003e1020899.\u003c/li\u003e\n\u003cli\u003eLoomba R, Amangurbanova M, Bettencourt R, Madamba E, Siddiqi H, Richards L, et al. MASH Resolution Index. development and validation of a non-invasive score to detect histological resolution of MASH. \u003cem\u003eGut. \u003c/em\u003e2024; 73\u003cstrong\u003e:\u003c/strong\u003e1343-9.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-gastroenterology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmge","sideBox":"Learn more about [BMC Gastroenterology](http://bmcgastroenterol.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmge/default.aspx","title":"BMC Gastroenterology","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Metabolic dysfunction–associated steatotic liver disease, MASLD, High-altitude exposure, Machine learning, Risk prediction model, Nomogram, Military health","lastPublishedDoi":"10.21203/rs.3.rs-9109584/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9109584/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eThe impact of long-term high-altitude exposure on metabolic dysfunction\u0026ndash;associated steatotic liver disease (MASLD) risk in lowlanders remains poorly defined. Given the scarcity of medical resources in high-altitude regions, a practical tool for early identification of MASLD is critical. This study aimed to develop and validate a MASLD risk prediction model tailored for lowlanders with long-term high-altitude exposure using routine clinical indicators.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eThis retrospective study analyzed 663 lowlanders with long-term high-altitude exposure from July 2022 to June 2023. Missing data were imputed using the MissForest algorithm. A rigorous feature selection strategy comprising univariate analysis, LASSO regression, and the Boruta algorithm was employed. Six machine learning models\u0026mdash;Logistic Regression (LR), Random Forest (RF), Decision Tree (DT), XGBoost, Support Vector Machine (SVM), and LightGBM\u0026mdash;were trained and evaluated using 10-fold cross-validation. Performance was assessed via AUC, accuracy, sensitivity, specificity, F1 score, calibration curves, and Decision Curve Analysis (DCA). Model interpretability was quantified using SHapley Additive exPlanations (SHAP). A nomogram and an online dynamic calculator were developed based on the optimal model.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eNine predictors were identified: altitude, systolic blood pressure (SBP), body mass index (BMI), white blood cell count (WBC), red blood cell count (RBC), hemoglobin (HGB), hepatic steatosis index (HSI), triglycerides (TG), and high-density lipoprotein cholesterol (HDL-C). Among the six models, Logistic Regression. SHAP analysis further elucidated the model's interpretability, ranking BMI as the primary risk driver followed by TG and SBP. The nomogram and web-based calculator demonstrated good calibration and provided net clinical benefit across clinically relevant thresholds.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eWe developed and internally validated a machine learning\u0026ndash;based model to predict MASLD risk in lowlanders with long-term high-altitude exposure. The online calculator (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://altitude-masld-predictor.shinyapps.io/dynnomapp\u003c/span\u003e\u003cspan address=\"https://altitude-masld-predictor.shinyapps.io/dynnomapp\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) may facilitate early screening and individualized risk stratification in resource-limited high-altitude settings.\u003c/p\u003e","manuscriptTitle":"Development and validation of a machine learning model for MASLD risk prediction in lowlanders with long-term high-altitude exposure","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-24 15:10:44","doi":"10.21203/rs.3.rs-9109584/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-04-29T14:11:19+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-26T01:41:27+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"330612729208070185168488404876210604863","date":"2026-04-23T14:07:15+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"154329609274424945869235742482841118033","date":"2026-04-23T07:24:27+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-16T11:41:57+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-03-20T07:14:44+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-03-19T11:27:41+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-03-19T11:26:41+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Gastroenterology","date":"2026-03-13T03:03:51+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-gastroenterology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmge","sideBox":"Learn more about [BMC Gastroenterology](http://bmcgastroenterol.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmge/default.aspx","title":"BMC Gastroenterology","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"695b31a8-d342-49be-bb6a-18f9f3e10c53","owner":[],"postedDate":"April 24th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-04-29T14:11:19+00:00","index":60,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-24T15:10:44+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-24 15:10:44","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9109584","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9109584","identity":"rs-9109584","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}