Explainable Machine Learning Predicts Advanced HIV Disease Progression Using Easily Accessible Hematological Markers

Explainable Machine Learning Predicts Advanced HIV Disease Progression Using Easily Accessible Hematological Markers | 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 Article Explainable Machine Learning Predicts Advanced HIV Disease Progression Using Easily Accessible Hematological Markers Qian Zhang, Jingguo Li, Shanling Wang, Lingjuan Chen, Xueting Bai, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8226880/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background A large number of people living with HIV are not diagnosed until the advanced stage, and they face a high risk of death even after initiating antiretroviral therapy (ART). This study aimed to identify high-risk patients for advanced HIV disease using easily accessible hematological markers, explore effective predictive indicators in resource-limited settings, and provide a basis for early clinical intervention. Methods Data were collected from HIV/AIDS patients receiving ART in Linhai, Zhejiang Province, China, from 2010 to June 2025. Patients were classified into advanced infection (CD4 + T cell count < 200 cells/mm³) and non-advanced infection according to WHO criteria. Feature selection was performed using Lasso regression combined with the Boruta algorithm, and eight machine learning models were developed. Model performance was evaluated by discrimination, calibration, and clinical practicability. The optimal model was subjected to SHAP (SHapley Additive exPlanations) analysis to assess variable importance. Results A total of 709 patients were included. Among them, 260 individuals (accounting for 36.6%) had progressed to the advanced stage of HIV disease before starting ART.Seven variables were selected to construct the machine learning models. The ENET model demonstrated the highest AUC (0.801) in the validation set, along with satisfactory calibration and clinical utility. SHAP analysis revealed that CD8 + T cells had the highest average SHAP value, contributing the most to model prediction. Among the easily accessible hematological markers, total cholesterol had the greatest contribution. Conclusion The ENET model exhibited optimal performance for predicting advanced HIV disease, serving as an effective tool for identifying high-risk patients. CD8 + T cells are the core immune indicator for predicting disease progression, while total cholesterol is the most influential among easily accessible hematological markers. Combining these markers with others such as hemoglobin provides a convenient and reliable approach for assessing HIV disease progression risk in resource-limited settings. Health sciences/Biomarkers Biological sciences/Computational biology and bioinformatics Health sciences/Diseases Health sciences/Medical research HIV/AIDS hematological markers machine learning SHAP Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Background HIV/AIDS is a chronic infectious disease caused by the Human Immunodeficiency Virus (HIV), which impairs the immune system and makes individuals more susceptible to other infections and diseases [ 1 ].In China, despite the widespread promotion of antiretroviral therapy (ART) having significantly reduced AIDS-related mortality, new infections and disease burden remain severe [ 2 , 3 ]. HIV is detected in a large number of patients with advanced infection (defined by the World Health Organization (WHO) as a CD4 + T cell count of less than 200 cells/mm³) [ 4 , 5 ], which is characterized by severe immunosuppression, recurrent opportunistic infections, and a high dependence on medical and supportive care. A study on trends across 55 countries showed [ 6 ] that more than one-third (37%) of patients initiating ART in 2015 already had advanced HIV infection. Even after starting ART (which increases inflammatory responses), such patients have a high risk of death, and the risk increases with the decrease in CD4 + T cells [ 6 ]. If patients start ART as early as possible, their life expectancy is close to normal [ 7 ]. Early treatment can also reduce the risk of HIV transmission [ 8 ]. Currently, the main basis for assessing HIV/AIDS disease progression and staging in clinical practice is CD4 + T cell count and HIV-RNA [ 9 ]. However, these tests may not be immediately and routinely available in some primary medical institutions or economically underdeveloped areas, with certain technical, cost, and time barriers. Therefore, finding alternative or auxiliary predictive indicators that can be quickly, conveniently, and low-cost obtained before ART is of crucial significance for early identification of high-risk patients and advancing the threshold of clinical intervention.In recent years, an increasing number of studies have shown that some basic indicators in routine blood tests, such as hemoglobin [ 9 ], platelets [ 10 , 11 ], total cholesterol [ 12 ], etc., are significantly associated with the immune status and disease progression of HIV-infected individuals.These indicators are easily accessible, low-cost, and provide rapid results, laying a valuable data foundation for constructing efficient clinical prediction models. In recent years, artificial intelligence technology, especially explainable machine learning methods, has provided a powerful tool for mining in-depth information from routine clinical data [ 13 ]. However, there is currently a lack of research on constructing interpretable machine learning models based on these easily accessible blood indicators to accurately identify patients who are about to progress to the advanced stage of HIV disease. This study aims to utilize readily available hematological indicators to develop and validate an interpretable machine learning model. The goal is to provide clinicians with a simple and effective tool at the beginning of ART treatment for early identification of HIV-infected individuals with a high risk of progressing to advanced stages of the disease. This will provide scientific basis for achieving precise intervention and optimizing the allocation of medical resources. Moreover, it seeks to identify conventional hematological indicators that can effectively predict the progression to the advanced stage of HIV disease in resource-limited environments, ultimately providing scientific support for achieving precise intervention and optimizing the allocation of medical resources. Materials and Methods Data Source The data is derived from HIV/AIDS patients who received antiretroviral therapy in Linhai City, Taizhou City, Zhejiang Province from January 2010 to June 2025.The following indicators were included: Hematological indicators: viral load, white blood cell count (WBC), platelets (PLT), hemoglobin (HB), serum creatinine (SCR), triglycerides (TG), total cholesterol (TC), blood glucose, ALT, AST, T.BIL, HBsAG, HbeAg, AntiHCV, and CD8 + T cells measured for the first time after diagnosis of HIV/AIDS [ 14 ]. Demographic data: age at diagnosis, gender, ethnicity, educational level, occupation, and marital status. Other variables: history of sexually transmitted diseases (STD) and transmission route. Among the above variables, gender (female, male), ethnicity (Han, other), educational level (illiterate, primary school, junior high school, high school or technical secondary school, college and above), HBsAG (negative, positive), HbeAg (negative, positive), AntiHCV (negative, positive), occupation (commercial sex worker, other worker, unemployed), marital status (married or with spouse, unmarried, divorced or widowed), history of STD (yes, no), and transmission route (homosexual transmission, heterosexual transmission, other routes) were categorical variables, and the other variables were continuous variables. The outcome variable was progression to advanced HIV infection, defined as CD4 + T cell count < 200 cells/mm³ [ 3 , 4 , 8 ]. Patients aged 18–85 years who received ART during the study period were included, while those with missing post-diagnosis CD4 + T cell data were excluded. A total of 709 patients were included. The sample size of this study was calculated using the ‘pmsampsize’ package in R software, which was developed based on the guidelines proposed by RD Riley et al. in 2018 to calculate the minimum sample size required for multivariable prediction models in clinical research [ 15 ]. After calculation, the minimum sample size required for this study was 357, so the sample size in the study met the requirements. This study has been approved by the Ethics Review Committee of Taizhou Center for Disease Control and Prevention (Taizhou Health Supervision Institute) (Approval No.: Taizhou Disease Control Center Review and Approval Document No. 008 of 2025 Research Project).The study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki.Informed consent has been waived by using cases and biological specimens obtained from previous clinical diagnoses and treatments. All personal data in the study have been anonymized during the data organization and statistical analysis stages to ensure participant privacy. No compensation was provided to the participants. Statistical Analysis Variables with more than 30% missing values were removed from the dataset [ 16 ]. Variables with more than 30% missing values included viral load, blood glucose, HbeAg, and AntiHCV. For variables with less than 30% missing values, the ‘MissForest’ package in R software was used for imputation [ 17 , 18 ].Descriptive analysis was performed on the preprocessed data using the ‘compareGroups’ package in R software. Measurement data were expressed as mean ± standard deviation, and independent sample t-test was used for comparison between groups; count data were expressed as frequency (%), and chi-square test was used for comparison between groups. Data standardization plays an important role in improving the performance of machine learning models. It can scale data of different dimensions or scales to the same interval, thereby increasing the comparability between variables. In this study, the scale() function of the base package in R software was used for Z-score standardization of the data. The standardized data conforms to a standard normal distribution with a mean of 0 and a standard deviation of 1. The dataset was divided into a training set and a test set randomly at a ratio of 7:3 using the sample.split function in the ‘caTools’ package. Variable selection was performed by combining LASSO regression and the Boruta algorithm. LASSO regression is an improved linear regression method mainly used for feature selection and model regularization [ 19 ]. In contrast, the Boruta algorithm determines the relevance of features by comparing the importance of features with that of randomly permuted (noise) features. The Boruta algorithm adopts the feature selection method of random forest, and the number of trees in the random forest is set to 1000 to enhance the stability and accuracy of the model [ 19 ]. Five-fold cross-validation was used in the training set to optimize model parameters and establish eight machine learning models:Logistic Regression (LR), Support Vector Machine (SVM), Gradient Boosting Machine (GBM), Elastic Network (ENET), Neural Network (NNET), Extreme Gradient Boosting (XGBoost), k-Nearest Neighbor (KNN), and Linear Discriminant Analysis (LDA).LR is a generalized linear regression analysis model used to study the influence relationship between categorical dependent variables and independent variables [ 20 ]. SVM is a supervised learning model based on statistical learning theory, widely used in classification and regression tasks [ 21 ]. GBM is an ensemble learning model widely used in the field of machine learning [ 22 ]. KNN is a classic supervised learning algorithm, widely used in classification and regression tasks [ 23 ]. ENET is a linear regression method that combines the advantages of Lasso regression (L1 regularization) and ridge regression (L2 regularization), mainly used to solve problems such as high-dimensional data and feature correlation [ 24 ]. NNET overcomes the limitations of traditional artificial intelligence methods in processing complex and unstructured information [ 25 ]. XGBoost uses the second-order Taylor series to approximate the value of the loss function, and further reduces the possibility of overfitting through regularization [ 26 ]. Linear Discriminant Analysis (LDA) is a classic statistical machine learning method that aims to find a linear data transformation that increases class discrimination in the optimal discriminant subspace [ 27 ]. The predictive performance of the models was evaluated from three dimensions: discriminative ability, calibration, and clinical utility. The Area Under the Receiver Operating Characteristic Curve (AUC) was used to evaluate the discriminative ability of the model. The closer the AUC is to 1, the stronger the diagnostic discrimination ability of the model. The Calibration Curve was used to evaluate the calibration of the model. The x-axis and y-axis of the calibration curve represent the predicted probability and the actual probability, respectively. A model with good predictive performance should have a calibration curve close to the diagonal line, indicating that the predicted probability is basically consistent with the actual probability. Clinical utility was evaluated using Decision Curve Analysis (DCA). The decision curve is a smooth curve formed by connecting the net benefits under different threshold probabilities. There are two fixed lines in the decision curve graph: the line parallel to the x-axis refers to the net benefit of not treating any patients, and the other diagonal line refers to the net benefit of treating all patients. The larger the area under the curve formed by the model's decision curve and these two lines, the higher the clinical net benefit of the model. In addition, the DeLong test [ 28 ] was used to compare whether there were significant differences in AUC between different models, and a p-value < 0.05 was considered statistically significant. We used SHAP (SHapley Additive exPlanations) values to evaluate the overall feature importance in the ML model with the best predictive performance. SHAP measures the importance of each feature to the model by generating a contribution value (shapley) of each feature in the model to the predicted outcome [ 29 ]. SHAP values can show the positive or negative contribution of each predictor variable to the target variable. The data analysis process of this study was performed using R4.4.2, and a p-value < 0.05 was considered statistically significant. Figure 1 is the flowchart of this study. Results This study included 709 HIV/AIDS patients. Among them, 260 cases (36.6%) progressed to the advanced stage of HIV disease (with CD4 + T cell count below 200 cells/mm3).Table 1 shows the comparison of baseline characteristics between the training set and the test set. Among them, the average age of HIV/AIDS patients was 43.2 years old. The Han ethnicity accounted for the vast majority, with 684 cases (96.5%), and the majority were male patients, with 585 cases (82.5%). Most of the patients were married or had a spouse, with 402 cases (56.7%). Compared to other stages, a higher proportion had a junior high school education, with 260 cases (36.7%). There were 60 cases (8.5%) with positive HBsAg, 44 cases (6.2%) were sex workers, and 137 cases (19.3%) had a history of sexually transmitted diseases. The majority were transmitted through heterosexual contact, with 437 cases (61.6%).Except for the transmission route, there were no statistically significant differences in other variables between the two groups (P > 0.05), indicating that the two groups were balanced and comparable. Table 1 Comparison of baseline characteristics between the training set and test set of HIV/AIDS patients. Overall test train p.overall N = 709 N = 212 N = 497 CD8+(cells /ul) 936 (539) 929 (544) 938 (538) 0.839 WBC(10^9/L) 5.58 (2.31) 5.74 (3.18) 5.51 (1.82) 0.340 PLT(10^9/L) 208 (72.7) 199 (77.2) 212 (70.4) 0.038 HB(g/L) 138 (21.8) 137 (22.4) 138 (21.6) 0.488 SCR(µmol/L) 69.1 (20.3) 68.7 (14.6) 69.3 (22.3) 0.665 TG(mmol/L) 1.86 (2.51) 1.92 (1.73) 1.84 (2.78) 0.665 TC(mmol/L) 4.25 (0.864) 4.17 (0.85) 4.28 (0.87) 0.105 ALT(U/L) 30.6 (44.1) 30.2 (28.0) 30.8 (49.4) 0.849 AST(U/L) 29.1 (28.2) 28.8 (18.5) 29.3 (31.4) 0.819 T.BIL(µmol/L) 13.0 (24.8) 11.6 (5.48) 13.5 (29.4) 0.161 Age(years) 43.2 (15.5) 44.5 (15.8) 42.7 (15.4) 0.173 HBsAG: 0.472 Negative 649 (91.5%) 197 (92.9%) 452 (90.9%) Positive 60 (8.5%) 15 (7.08%) 45 (9.05%) Sex: 0.901 Female 124 (17.5%) 36 (17.0%) 88 (17.7%) Male 585 (82.5%) 176 (83.0%) 409 (82.3%) Nationality: 0.368 Han 684 (96.5%) 202 (95.3%) 482 (97.0%) Other 25 (3.5%) 10 (4.72%) 15 (3.02%) Education: 0.736 College 89 (12.6%) 26 (12.3%) 63 (12.7%) High 124 (17.5%) 37 (17.5%) 87 (17.5%) Illiterate 56 (7.9%) 18 (8.49%) 38 (7.65%) Middle 260 (36.7%) 71 (33.5%) 189 (38.0%) Primary 180 (25.4%) 60 (28.3%) 120 (24.1%) Work: 0.427 CSW 44 (6.2%) 12 (5.66%) 32 (6.44%) Other 566 (79.8%) 165 (77.8%) 401 (80.7%) Unemployed 99 (14.0%) 35 (16.5%) 64 (12.9%) Marital: 0.972 Divorced/widowed 93 (13.1%) 27 (12.7%) 66 (13.3%) Married 402 (56.7%) 120 (56.6%) 282 (56.7%) Unmarried 214 (30.2%) 65 (30.7%) 149 (30.0%) STD: 0.256 No 572 (80.7%) 177 (83.5%) 395 (79.5%) Yes 137 (19.3%) 35 (16.5%) 102 (20.5%) Route: 0.044 Heterosexual 437 (61.6%) 125 (59.0%) 312 (62.8%) Homosexual 267 (37.7%) 83 (39.2%) 184 (37.0%) Other 5 (0.7%) 4 (1.89%) 1 (0.20%) Feature Selection Variable selection was performed by combining Lasso regression and the Boruta algorithm. As shown in Fig. 7a,b, Lasso regression used ten-fold cross-validation to select λ. In this study, λ.min was used as the optimal λ value, and the selected variables were CD8 + T cells, WBC, PLT, HB, SCR, TG, TC, AST, T.BIL, gender, ethnicity, and marital status. As shown in Fig. 7c,d, the variables most closely related to the progression of advanced HIV disease screened by the Boruta algorithm were CD8 + T cells, WBC, TC, HB, AST, PLT, and T.BIL. The variables jointly screened by the two methods were: CD8, WBC, TC, HB, AST, PLT, and T.BIL. Finally, these seven variables were included in the eight machine learning algorithm models. Establishment and Evaluation of Eight Machine Learning Algorithm Models The 709 subjects were randomly divided into a training set (497 people) and a test set (212 people) at a ratio of 7:3. Using the seven variables screened above, eight machine learning models (LR, SVM, GBM, ENET, NNET, XGBoost, KNN, and LDA) were constructed. The predictive ability of the models was evaluated from three dimensions: discriminative ability, calibration, and clinical utility. The ROC curves of the eight models in the training set and validation set are shown in Fig. 3 a,b. In the training set, XGBoost had the highest AUC (0.876), followed by GBM, KNN, SVM, NNET, LR, ENET, and LDA (AUC = 0.793) was the lowest. In the validation set, ENET was the highest (AUC = 0.801), followed by LR, GBM, LDA, XGBoost, NNET, SVM, and KNN (AUC = 0.768) was the lowest. In the validation set, pairwise comparison of the models by the DeLong test (Table 2 ) showed that only the difference in AUC between the ENET model (AUC = 0.801) and the LDA model (AUC = 0.794) was statistically significant (P 0.05). Table 3 lists the detailed comparison of specific performance indicators of each model. Although the accuracy, sensitivity, specificity, precision, and F1 score of the ENET model are not the highest, they are all at a relatively reasonable level, and overall, it can better solve the research problem(Table 3 ). Figure 4 provides the confusion matrices of each model in the training set and test set. Table 2 Pairwise comparison of AUC values of eight models in the test set. AUC1 Model2 AUC2 Z_statistic P_value Significant 0.797 SVM 0.769 1.627 0.1037 No 0.797 GBM 0.795 0.077 0.9384 No 0.797 ENET 0.801 -0.587 0.5569 No 0.797 NNET 0.781 1.307 0.1913 No 0.797 XGB 0.781 0.652 0.5143 No 0.797 KNN 0.768 1.397 0.1625 No 0.797 LDA 0.794 0.606 0.5448 No 0.769 GBM 0.795 -1.188 0.2348 No 0.769 ENET 0.801 -1.882 0.0599 No 0.769 NNET 0.781 -0.663 0.5073 No 0.769 XGB 0.781 -0.508 0.6112 No 0.769 KNN 0.768 0.052 0.9584 No 0.769 LDA 0.794 -1.429 0.1529 No 0.795 ENET 0.801 -0.209 0.8346 No 0.795 NNET 0.781 0.597 0.5507 No 0.795 XGB 0.781 1.584 0.1132 No 0.795 KNN 0.768 0.984 0.3253 No 0.795 LDA 0.794 0.055 0.9559 No 0.801 NNET 0.781 1.481 0.1387 No 0.801 XGB 0.781 0.797 0.4253 No 0.801 KNN 0.768 1.707 0.0878 No 0.801 LDA 0.794 2.021 0.0433 Yes 0.781 XGB 0.781 0.026 0.9793 No 0.781 KNN 0.768 0.543 0.587 No 0.781 LDA 0.794 -0.957 0.3385 No 0.781 KNN 0.768 0.461 0.6447 No 0.781 LDA 0.794 -0.509 0.611 No 0.768 LDA 0.794 -1.328 0.1843 No Table 3 Performance parameters of eight prediction models. Model Accuracy Sensitivity Specificity Precision F1 Train set LR 0.757 0.533 0.886 0.729 0.616 SVM 0.807 0.637 0.905 0.795 0.707 GBM 0.799 0.615 0.905 0.789 0.691 ENET 0.755 0.505 0.898 0.742 0.601 NNET 0.773 0.632 0.854 0.714 0.671 XGB 0.801 0.604 0.914 0.803 0.690 KNN 0.797 0.615 0.902 0.783 0.689 LDA 0.753 0.500 0.898 0.740 0.597 Test set LR 0.764 0.526 0.903 0.759 0.621 SVM 0.712 0.474 0.851 0.649 0.548 GBM 0.736 0.513 0.866 0.690 0.588 ENET 0.764 0.513 0.910 0.769 0.615 NNET 0.726 0.538 0.836 0.656 0.592 XGB 0.731 0.487 0.873 0.691 0.571 KNN 0.741 0.551 0.851 0.683 0.610 LDA 0.769 0.513 0.918 0.784 0.620 We plotted calibration curves and DCA curves based on the training and test sets. The former is used to evaluate the accuracy and reliability of model predictions, while the latter is used to evaluate the potential clinical utility of the model within different threshold ranges. The calibration curves of the eight models are shown in Fig. 5 a,b. In the training set, all models except the SVM model showed good calibration. In the test set, the NNET, GBM, and XGBoost models had poor calibration, and the remaining models had good calibration. The DCA curves of the eight models (Fig. 5 c,d) showed that within a wide range of thresholds, the net benefits of these models were higher than the "intervene" or "no intervene" strategies. In the training set, within different threshold probability ranges, the XGBoost and GBM models showed more significant clinical benefits compared with other models. In the test set, the ENET model had the highest net benefit and good clinical utility. Based on the comprehensive evaluation results of model performance, the ENET model exhibited superior discriminative ability and clinical utility, so it is the optimal model for identifying progression to advanced HIV disease. Model Interpretation - SHAP Analysis In this study, the SHAP algorithm was used to explain the importance of predictive variables in the ENET model with the best predictive performance (Fig. 6 ). The contribution of variables to the model was reflected by SHAP values. Figure 6 a ranks the importance of each variable in descending order according to the absolute value of the average SHAP value. CD8 + T cells had the highest average SHAP value, contributing the most to model prediction. Among the easily accessible hematological markers, TC had the greatest contribution, followed by HB, AST, PLT, WBC and T.BIL. Figure 6 b is a swarm plot of variable importance of the ENET model. The x-axis represents the size of the SHAP value, and the order of variables from top to bottom on the y-axis is arranged in descending order of importance. Each point in the figure represents a sample. The closer the color of the point is to purple, the lower its SHAP value; the closer it is to yellow, the higher its value. When the SHAP value is greater than 0, the variable has a positive contribution to the prediction result; when it is less than 0, it has a negative contribution. Figure 6 c is a SHAP force plot. If the feature block is yellow and the SHAP value is positive: the current value of the feature makes the prediction result higher than the "baseline value" (the average prediction value of the model for all samples, E[f(x)] = 0.245 in the figure is the baseline value). If the feature block is purple and the SHAP value is negative: the current value of the feature makes the prediction result lower than the "baseline value". Discussion Since the first AIDS-related death was reported in China in 1985 [ 30 ], the Chinese government has implemented a series of AIDS prevention and control policies and achieved certain results [ 31 ]. Early diagnosis of HIV in suspected populations is a prerequisite for successful treatment of AIDS.Although early diagnosis of HIV is of great significance in patient treatment, studies have shown that late diagnosis remains a major problem in global AIDS control and prevention [ 32 ]. Late diagnosis of HIV reduces the effectiveness of ART, thereby increasing the risk of morbidity and mortality in patients, which has adverse effects on individuals and society.On the other hand, such late diagnosis increases the possibility of disease transmission and spread in society, making disease control more difficult [ 33 , 34 ].In addition, the main basis for assessing HIV/AIDS disease progression and staging is CD4 + T cell count and HIV-RNA [ 9 ], which may not be achievable in economically underdeveloped areas or primary medical institutions, providing the possibility of using easily accessible routine hematological indicators to assist treatment. Therefore, this study also used routine, low-cost, and easily accessible hematological indicators based on machine learning algorithm models to quickly and accurately identify patients who seem stable but are actually at high risk of progressing to advanced HIV disease at the critical time point of initiating ART providing a scientific basis for early diagnosis and intervention. This study collected data on HIV/AIDS patients who received antiretroviral therapy in Linhai City, Taizhou City, Zhejiang Province from January 2010 to June 2025.A total of 709 HIV/AIDS patients were included. Among them, 260 cases (36.6%) had progressed to the advanced stage of HIV disease (with CD4 + T cell count below 200 cells/mm3) before starting ART.Seven predictors were screened by combining Lasso regression and the Boruta algorithm to construct eight machine learning models, and the model performance was evaluated from three aspects: discriminative ability, calibration, and clinical utility. Based on the comprehensive evaluation indicators of the training set and validation set, the ENET model was selected as the optimal model. Although its accuracy, sensitivity, specificity, precision, and F1 score are not the highest, they are all at a relatively reasonable level, and overall, it can better solve the research problem. The AUC of the ENET model in the training set was 0.794, and the AUC in the validation set was 0.801, with good calibration and clinical utility. After confirming that the prediction model for progression to advanced HIV disease constructed based on easily accessible routine hematological indicators before antiretroviral therapy has a good predictive effect, and the ENET model performs the best, it is particularly crucial to conduct an in-depth analysis of the important variables involved in model construction. In recent years, machine learning has been widely used in the field of infectious diseases. However, due to the "black box" nature of machine learning, its interpretability is relatively poor, so it is difficult to explain why specific predictions are made for patients [ 35 ].In this study, we used SHAP analysis to explain the optimal model in detail. SHAP analysis allowed us to evaluate the contribution of different variables to the model's prediction results. This study screened seven variables, and the importance of these seven variables was ranked based on the SHAP algorithm. The results of this study showed that CD8 + T cell count contributed the most to the model's prediction of HIV disease progression. As a key immune cell indicator, the indicating effect of changes in CD8 + T cell levels on disease progression is highlighted in the model. This is consistent with previous research conclusions [ 18 ] that a decrease in CD8 + T cells is the best independent predictor of HIV disease progression, especially when the CD4 + T cell count is below 200 cells/mm³, a decrease in CD8 + T cells will significantly increase the risk of HIV disease progression. From a pathological mechanism perspective, when the number of CD4 + T cells drops to an extremely low level of < 200 cells/µL, the immune system collapses completely; at this time, even CD8 + T cells, which act as the "emergency force" of the immune system, cannot maintain their numbers due to the complete failure of the immune system, and their functions and numbers will decrease significantly. At the same time, persistent immune activation caused by HIV infection leads to exhaustion of CD8 + T cells, making them gradually lose immune function and eventually undergo apoptosis, which is also an important driving factor for the decrease in the number of CD8 + T cells in progressive HIV infection [ 18 ].In our study, SHAP analysis showed that among the easily accessible routine hematological indicators, TC and HB also had relatively high contribution values, second only to CD8 + T cells. A study by Melaku et al. [ 36 ] showed that low serum total cholesterol and its combination with anemia showed high sensitivity > 80.0. Total cholesterol can be used as an important biomarker because lipids play a role in virus entry, uncoating, replication, protein synthesis, assembly, budding, and infectivity [ 37 , 38 ]. Cellular cholesterol is crucial for HIV replication and may control HIV transmission [ 39 ]. Compared with individuals with low cholesterol levels, immune system cells of patients with hypercholesterolemia have higher phagocytic activity, more circulating lymphocytes, more total T cells, more CD8 + T cells, more immunoglobulin production, more proliferation and differentiation, and migration of lymphocytes [ 40 ]. This can be explained by the role of intermediates in the cholesterol biosynthesis pathway and the role of downstream oxysterol metabolites that have been found to affect various functions of immune system cells. In addition, the effect of ART is impaired in HIV-infected patients with hypocholesterolemia [ 41 ].This may mean that in this era of testing and treatment, low serum TC may be used as a predictive marker for ART efficacy.Anemia is a common complication, occurring in 20–80% of HIV-infected individuals, and is associated with faster disease progression and higher mortality rates [ 42 ].HB levels reflect the speed of disease progression and independently predict prognosis in different populations [ 43 , 44 ].The rate of HB decline is also associated with a decrease in CD4 count [ 45 ]. A study by Christian et al. [ 9 ] showed that HB measurement can be used as an indicator of HIV/AIDS progression in resource-limited settings.In addition, changes in indicators such as AST, PLT, WBC, and T.BIL are also helpful for identifying progression to advanced HIV disease.These biomarkers can improve the performance of doctors in accurately examining the disease prognosis of patients receiving ART. If doctors only have CD4 + T cell counts, it is not easy to switch treatment because the recovery rate of CD4 + T cells is not high enough, especially in patients who start ART late [ 46 ].Using alternative biomarkers to monitor ART efficacy by extending the interval between testing CD4 + T cell counts and HIV RNA load is a very useful tool. Therefore, in this case, basic laboratory tests of alternative biomarkers and effective clinical monitoring will be very helpful [ 47 ]. Therefore, HIV/AIDS patients should pay special attention to the fluctuations of these indicators before starting ART. This model is based on easily accessible routine hematological indicators and shows considerable effectiveness even in resource-limited settings where CD4 + cell counts and HIV-RNA are not available. However, this study also has its limitations. This study was conducted only in one county-level city in Taizhou, Zhejiang Province, introducing potential selection bias. Therefore, external validation from other regions is needed to improve the generalization of the model. Secondly, the study is cross-sectional, which can identify the association between risk variables and outcome variables, but cannot identify causal relationships. The current model fails to capture the dynamic changes of key indicators, which may limit its application in the dynamic monitoring and management of HIV/AIDS patients.Future studies should include longitudinal data to explore the performance of the new model at different time points and analyze the impact of changes in key indicators over time on outcomes. In addition, due to limitations in detection technology, a large number of data such as viral load and serological evidence of hepatitis B/C in earlier years are missing. Due to limitations in data access rights, comorbidities such as tuberculosis, family history of dyslipidemia, emotional changes, depression, and lifestyle-related factors (smoking and lack of exercise) have not been explained. Conclusion In this study, the ENET model constructed based on seven screened variables showed the optimal performance, with an AUC of 0.801 in the test set, along with good calibration and clinical utility, making it an effective tool for identifying high-risk patients of advanced disease. CD8 + T cells are the core immune indicator for predicting HIV progression to the advanced stage, and TC is the primary predictive factor among clinically easily accessible hematological markers. The combination of these two indicators with other hematological markers (such as HB) can provide a convenient and reliable reference for assessing HIV disease progression risk in resource-limited settings. Abbreviations HIV Human Immunodeficiency Virus AIDS Acquired Immune Deficiency Syndrome ART Antiretroviral Therapy STD Sexually Transmitted Diseases WBC White Blood Cell Count PLT Platelets HB Hemoglobin SCR Serum Creatinine TG Triglycerides TC Total Cholesterol AUC Area Under the Curve DCA Decision Curve Analysis LR Logistic Regression SVM Support Vector Machine GBM Gradient Boosting Machine KNN k-Nearest Neighbor ENET Elastic Network NNET Neural Network XGBoost Extreme Gradient Boosting LDA Linear Discriminant Analysis SHAP SHapley Additive exPlanations Declarations Ethics approval and consent to participate Data were obtained from HIV/AIDS patients receiving ART in Linhai City, Taizhou, Zhejiang Province, from January 2010 to June 2025. The study was approved by the Ethics Review Committee of Taizhou Center for Disease Control and Prevention (Taizhou Health Supervision Institute) (Approval NO.: Taizhou Disease Control Center Review and Approval Document No. 008 of 2025 Research Project) in accordance with the Declaration of Helsinki. Informed consent has been waived by using cases and biological specimens obtained from previous clinical diagnoses and treatments. The ethics review committee of Taizhou Center for Disease Control and Prevention (Taizhou Health Supervision Institute) approved the exemption of informed consent.All personal data in the study have been anonymized during the data organization and statistical analysis stages to ensure participant privacy. No compensation was provided to the participants. Consent for publication All authors reviewed and approved the final manuscript. Clinical trial number Not applicable. Availability of data and materials The data for this study were obtained from the Disease Prevention and Control Center of Linhai City, Taizhou City, Zhejiang Province, China.Data sources and handling of these data are described in the “Materials and Methods”.The data of HIV/AIDS patients in Linhai City, Taizhou City, Zhejiang Province were obtained by the corresponding author:Email: [email protected] . Competing interests The authors declare no conflict of interest. Funding None. Authors' contributions X.L. proposed this idea, designed the research plan and supervised the entire process. Z.Z. supervised the entire work and provided clinical knowledge.Q. Z. was responsible for the conception and design of the research, data analysis, article drafting and writing. J. L. collected the data and wrote the article. S. W., revised the content.L. C.，X. B., and W. Z. reviewed the relevant literature.Z.L. and Y.Z. provided methodological guidance. Acknowledgements We are extremely grateful to Linhai Center for Disease Control and Prevention for providing us with the data support.The authors highly appreciate all the members who were involved in the present study. References Li, X. C. et al. Global burden of viral infectious diseases of poverty based on Global Burden of Diseases Study 2021. Infect Dis Poverty. ;13(1):71. Published 2024 Oct 8. (2024). 10.1186/s40249-024-01234-z Xu, B., Li, J. & Wang, M. Epidemiological and time series analysis on the incidence and death of AIDS and HIV in China. BMC Public. Health . 20 (1), 1906. 10.1186/s12889-020-09977-8 (2020). Published 2020 Dec 14. Zhai, M. et al. The trend of AIDS in China: A prediction and comparative analysis with G20 countries based on the Global Burden of Disease Study 2019. J Glob Health. ;14:04029. Published 2024 Mar 1. (2024). 10.7189/jogh.14.04029 Zeng, Y. et al. Exploring palliative care needs of advanced AIDS patients in China: insights from a descriptive phenomenological study. BMC Palliat. Care . 24 (1), 42. 10.1186/s12904-025-01675-9 (2025). Published 2025 Feb 13. Ford, N. & Doherty, M. The Enduring Challenge of Advanced HIV Infection. N Engl. J. Med. 377 (3), 283–284. 10.1056/NEJMe1707598 (2017). IeDEA et al. Immunodeficiency at the start of combination antiretroviral therapy in low-, middle-, and high-income countries. J. Acquir. Immune Defic. Syndr. 65 (1), e8–e16. 10.1097/QAI.0b013e3182a39979 (2014). Antiretroviral Therapy Cohort Collaboration. Survival of HIV-positive patients starting antiretroviral therapy between 1996 and 2013: a collaborative analysis of cohort studies. Lancet HIV . 4 (8), e349–e356. 10.1016/S2352-3018(17)30066-8 (2017). Cohen, M. S. et al. Antiretroviral Therapy for the Prevention of HIV-1 Transmission. N Engl. J. Med. 375 (9), 830–839. 10.1056/NEJMoa1600693 (2016). Obirikorang, C. & Yeboah, F. A. Blood haemoglobin measurement as a predictive indicator for the progression of HIV/AIDS in resource-limited setting. J. Biomed. Sci. 16 (1), 102. 10.1186/1423-0127-16-102 (2009). Published 2009 Nov 18. Wu, F. et al. Increased platelet-CD8 + T-cell aggregates displaying high activation, exhaustion, and tendency to death correlate with disease progression in people with HIV-1. J. Leukoc. Biol. 116 (1), 166–176. 10.1093/jleuko/qiae048 (2024). Deng, M. et al. Over-activation and dysfunction of platelet-NK cell aggregates in HIV-infected individuals. J Transl Med. ;23(1):584. Published 2025 May 26. (2025). 10.1186/s12967-025-06591-3 Samuel, M. et al. The effects of age on associations between markers of HIV progression and markers of metabolic function including albumin, haemoglobin and lipid concentrations. HIV Med. 15 (5), 311–316. 10.1111/hiv.12103 (2014). Ponce-Bobadilla, A. V., Schmitt, V., Maier, C. S., Mensing, S. & Stodtmann, S. Practical guide to SHAP analysis: Explaining supervised machine learning model predictions in drug development. Clin. Transl Sci. 17 (11), e70056. 10.1111/cts.70056 (2024). Kvale, D., Aukrust, P., Osnes, K., Müller, F. & Frøland, S. S. CD4 + and CD8 + lymphocytes and HIV RNA in HIV infection: high baseline counts and in particular rapid decrease of CD8 + lymphocytes predict AIDS. AIDS 13 (2), 195–201. 10.1097/00002030-199902040-00007 (1999). Riley, R. D. et al. Calculating the sample size required for developing a clinical prediction model. BMJ 368 , m441. 10.1136/bmj.m441 (2020). Published 2020 Mar 18. Hu, C. et al. Interpretable Machine Learning for Early Prediction of Prognosis in Sepsis: A Discovery and Validation Study. Infect. Dis. Ther. 11 (3), 1117–1132. 10.1007/s40121-022-00628-6 (2022). Stekhoven, D. J. & Bühlmann, P. MissForest–non-parametric missing value imputation for mixed-type data. Bioinformatics 28 (1), 112–118. 10.1093/bioinformatics/btr597 (2012). You, J. et al. Comparison of imputation methods for missing production data of dairy cattle. Animal 17 (Suppl 5), 100921. 10.1016/j.animal.2023.100921 (2023). Fu, Y., Zhao, J. & Wang, Y. LASSO regression and Boruta algorithm to explore the relationship between neutrophil percentage to albumin ratio and asthma: results from the NHANES 2001 to 2018. Clin Exp Med. ;25(1):149. Published 2025 May 10. (2025). 10.1007/s10238-025-01701-3 Musa, A. B. Comparative study on classification performance between support vector machine and logistic regression. Int. J. Mach. Learn. Cyber . 4 , 13–24. https://doi.org/10.1007/s13042-012-0068-x (2013). Valkenborg, D., Rousseau, A. J., Geubbelmans, M. & Burzykowski, T. Support vector machines. Am. J. Orthod. Dentofac. Orthop. 164 (5), 754–757. 10.1016/j.ajodo.2023.08.003 (2023). Xie, Y. et al. JOURNAL CLUB: Use of Gradient Boosting Machine Learning to Predict Patient Outcome in Acute Ischemic Stroke on the Basis of Imaging, Demographic, and Clinical Information. AJR Am. J. Roentgenol. 212 (1), 44–51. 10.2214/AJR.18.20260 (2019). Rajaguru, H., S & R SC. Analysis of Decision Tree and K-Nearest Neighbor Algorithm in the Classification of Breast Cancer. Asian Pac. J. Cancer Prev. 20 (12), 3777–3781. 10.31557/APJCP.2019.20.12.3777 (2019). Published 2019 Dec 1. Wang, X. W. & Liu, Y. Y. Comparative study of classifiers for human microbiome data. Med. Microecol . 4 , 100013. 10.1016/j.medmic.2020.100013 (2020). Xiao, Y. Application of Neural Network Algorithm in Medical Artificial Intelligence Product Development. Comput. Math. Methods Med. 10.1155/2022/5413202 (2022). 2022:5413202. Published 2022 Jun 8. Wang, K. et al. Interpretable prediction of 3-year all-cause mortality in patients with heart failure caused by coronary heart disease based on machine learning and SHAP. Comput. Biol. Med. 137 , 104813. 10.1016/j.compbiomed.2021.104813 (2021). Xu, L., Raitoharju, J., Iosifidis, A. & Gabbouj, M. Saliency-Based Multilabel Linear Discriminant Analysis. IEEE Trans. Cybern . 52 (10), 10200–10213. 10.1109/TCYB.2021.3069338 (2022). DeLong, E. R., DeLong, D. M. & Clarke-Pearson, D. L. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 44 (3), 837–845 (1988). Lundberg, S. M. & Lee, S. I. A unified approach to interpreting model predictions. NIPS'17: Proceedings of the 31st International Conference on Neural Information Processing Systems; NIPS'17; Dec 4–9; Long Beach, CA. pp. 4768–4777. (2017). Settle, E. AIDS in China: An annotated chronology 1985–2003 (China AIDS Survey, Chinese Center for Disease Control and Prevention, 2003). Wu, Z., Wang, Y., Detels, R. & Rotheram-Borus, M. J. China AIDS policy implementation: reversing the HIV/AIDS epidemic by 2015. Int. J. Epidemiol. 39 (Suppl 2), ii1–ii3. 10.1093/ije/dyq220 (2010). Smith, M. K. et al. The detection and management of early HIV infection: a clinical and public health emergency. J. Acquir. Immune Defic. Syndr. 63 (0 2), S187–S199. 10.1097/QAI.0b013e31829871e0 (2013). Mugavero, M. J., Castellano, C., Edelman, D. & Hicks, C. Late diagnosis of HIV infection: the role of age and sex. Am. J. Med. 120 (4), 370–373. 10.1016/j.amjmed.2006.05.050 (2007). Mayer, K. H. & Venkatesh, K. K. Antiretroviral therapy as HIV prevention: status and prospects. Am. J. Public. Health . 100 (10), 1867–1876. 10.2105/AJPH.2009.184796 (2010). McDonald, L., Ramagopalan, S. V., Cox, A. P. & Oguz, M. Unintended consequences of machine learning in medicine? F1000Res 10.12688/f1000research.12693.1 (2017). 6:1707. Published 2017 Sep 19. Adal, M., Howe, R., Kassa, D., Aseffa, A. & Petros, B. Associations of gender and serum total cholesterol with CD4 + T cell count and HIV RNA load in antiretroviral-naïve individuals in Addis Ababa. BMC Public Health. ;18(1):943. Published 2018 Jul 31. (2018). 10.1186/s12889-018-5852-4 Lorizate, M. & Kräusslich, H. G. Role of lipids in virus replication. Cold Spring Harb Perspect Biol. ;3(10):a004820. Published 2011 Oct 1. (2011). 10.1101/cshperspect.a004820 Mazzon, M. & Mercer, J. Lipid interactions during virus entry and infection. Cell. Microbiol. 16 (10), 1493–1502. 10.1111/cmi.12340 (2014). Negredo, E. et al. Association between HIV replication and cholesterol in peripheral blood mononuclear cells in HIV-infected patients interrupting HAART. J. Antimicrob. Chemother. 61 (2), 400–404. 10.1093/jac/dkm470 (2008). Hannedouche, S. et al. Oxysterols direct immune cell migration via EBI2. Nature 475 (7357), 524–527. 10.1038/nature10280 (2011). Published 2011 Jul 27. Míguez MJ, Lewis JE, Bryant VE, et al. Low cholesterol? Don't brag yet … hypocholesterolemia blunts HAART effectiveness: a longitudinal study. J Int AIDS Soc. 2010;13:25. Published 2010 Jul 13. doi:10.1186/1758-2652-13-25. Belperio, P. S. & Rhew, D. C. Prevalence and outcomes of anemia in individuals with human immunodeficiency virus: a systematic review of the literature. Am. J. Med. 116 (Suppl 7A), 27S–43S. 10.1016/j.amjmed.2003.12.010 (2004). Paton, N. I., Sangeetha, S., Earnest, A. & Bellamy, R. The impact of malnutrition on survival and the CD4 count response in HIV-infected patients starting antiretroviral therapy. HIV Med. 7 (5), 323–330. 10.1111/j.1468-1293.2006.00383.x (2006). Wang, D. et al. Plasma hemoglobin and the risk of death in HIV/AIDS patients treated with antiretroviral therapy. Aging (Albany NY) . 13 (9), 13061–13072. 10.18632/aging.202987 (2021). Costello, C. et al. HIV-1 subtype E progression among northern Thai couples: traditional and non-traditional predictors of survival. Int. J. Epidemiol. 34 (3), 577–584. 10.1093/ije/dyi023 (2005). WHO. Guideline on when to start antiretroviral therapy and on pre-exposure prophylaxis for HIV (Geneva, 2015). Collini, P. et al. Sustained immunological responses to highly active antiretroviral therapy at 36 months in a Ghanaian HIV cohort. Clin. Infect. Dis. 48 (7), 988–991. 10.1086/597353 (2009). Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted 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-8226880","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":555190216,"identity":"5cbf34f1-b182-4ef8-b583-dc56991a8443","order_by":0,"name":"Qian Zhang","email":"","orcid":"","institution":"Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Qian","middleName":"","lastName":"Zhang","suffix":""},{"id":555190217,"identity":"8eea88fd-39ad-4115-8d74-a3b0190aa9c2","order_by":1,"name":"Jingguo Li","email":"","orcid":"","institution":"Linhai Center for Disease Control and Prevention","correspondingAuthor":false,"prefix":"","firstName":"Jingguo","middleName":"","lastName":"Li","suffix":""},{"id":555190218,"identity":"9d0b490e-7f8f-407c-8c98-e93afb55b8b3","order_by":2,"name":"Shanling Wang","email":"","orcid":"","institution":"Taizhou Center for Disease Control and Prevention","correspondingAuthor":false,"prefix":"","firstName":"Shanling","middleName":"","lastName":"Wang","suffix":""},{"id":555190219,"identity":"1bc35167-5230-4cb0-b1a2-6bec88c86076","order_by":3,"name":"Lingjuan Chen","email":"","orcid":"","institution":"Linhai Center for Disease Control and Prevention","correspondingAuthor":false,"prefix":"","firstName":"Lingjuan","middleName":"","lastName":"Chen","suffix":""},{"id":555190220,"identity":"2891eb07-e4d1-41a4-aa9c-477a32abf5ca","order_by":4,"name":"Xueting Bai","email":"","orcid":"","institution":"Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xueting","middleName":"","lastName":"Bai","suffix":""},{"id":555190221,"identity":"e646e8c9-28fe-4be7-92e1-281bbbfc4428","order_by":5,"name":"Wenchong Zhen","email":"","orcid":"","institution":"Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Wenchong","middleName":"","lastName":"Zhen","suffix":""},{"id":555190222,"identity":"fe666950-fa3a-4941-92cb-c27c13a83167","order_by":6,"name":"Zihan Li","email":"","orcid":"","institution":"Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zihan","middleName":"","lastName":"Li","suffix":""},{"id":555190223,"identity":"4dbc7665-9d79-41ef-b1e8-748124604aa3","order_by":7,"name":"Yixin Zhang","email":"","orcid":"","institution":"Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yixin","middleName":"","lastName":"Zhang","suffix":""},{"id":555190226,"identity":"a4b46566-5ca4-409b-b337-9eecdc1a7592","order_by":8,"name":"Zhen Zhang","email":"","orcid":"","institution":"Linhai Center for Disease Control and Prevention","correspondingAuthor":false,"prefix":"","firstName":"Zhen","middleName":"","lastName":"Zhang","suffix":""},{"id":555190227,"identity":"68ccfd9c-75a1-4533-ad66-4fc9973c4f6e","order_by":9,"name":"Xiaofeng Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAz0lEQVRIiWNgGAWjYBACPgbGBwyMDQwM/AwJID4zYS1sDMwGYC2SDSRrMThAtBaJZAaGnzts8oyPZ6dJMFRYJzawnz2AXwvPYQbG3jNpxWZn3m6TYDiTntjAk5eAXwt7/wFmxrbDidtu5G6TADEaJHgM8GthZmYAavmfuHkGSMs/YrSwN4O0HEjcIAHS0kCMFrBf2pITZ5x5u9ki4Vi6cRtPDn4t/OAQa7NL7G/P3XjjQ421bD/7GfxagID9B5yZALKXkPpRMApGwSgYBYQBAEDlQAI3tZsfAAAAAElFTkSuQmCC","orcid":"","institution":"Dalian Medical University","correspondingAuthor":true,"prefix":"","firstName":"Xiaofeng","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2025-11-28 05:53:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8226880/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8226880/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":97894243,"identity":"308f066b-a774-4d62-9656-3cb6b07f7c8a","added_by":"auto","created_at":"2025-12-10 15:32:05","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1184504,"visible":true,"origin":"","legend":"","description":"","filename":"ExplainableMachineLearningPredictsAdvancedHIV.docx","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/68c44ebb1195e6769711840f.docx"},{"id":97693295,"identity":"fd474c1c-93b6-4528-a350-70b6671f5930","added_by":"auto","created_at":"2025-12-08 11:16:39","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":11277,"visible":true,"origin":"","legend":"","description":"","filename":"aa6a6746cf974d85b4d317aeba30664c.json","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/08c5304f44db2eb856219f5b.json"},{"id":97892946,"identity":"e89349ef-b19f-4e7f-b291-ddee6bf14480","added_by":"auto","created_at":"2025-12-10 15:24:23","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":160129,"visible":true,"origin":"","legend":"","description":"","filename":"aa6a6746cf974d85b4d317aeba30664c1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/65e2dd7b650f40aadeb76ad5.xml"},{"id":97693298,"identity":"72072a5c-1196-48ee-923d-7bb297fe7658","added_by":"auto","created_at":"2025-12-08 11:16:39","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":149379,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/020735b3e3bdefceba93d708.png"},{"id":97693312,"identity":"94a5ce06-c27b-4b43-834d-e02dc084aaf0","added_by":"auto","created_at":"2025-12-08 11:16:40","extension":"jpeg","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1224324,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage10.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/875e5d1c6a027aa5238cfdc3.jpeg"},{"id":97693304,"identity":"920d6224-8683-4386-8400-e79bde803c3c","added_by":"auto","created_at":"2025-12-08 11:16:39","extension":"jpeg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":157039,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/9daa2b7015df46cfca4d05e8.jpeg"},{"id":97693300,"identity":"e82a8c97-4075-47a0-8eab-d995c823cae6","added_by":"auto","created_at":"2025-12-08 11:16:39","extension":"jpeg","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":216251,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/f707d140b129f9c9081f46a9.jpeg"},{"id":97894319,"identity":"d71884b0-ef32-4b06-9521-07a702be2be5","added_by":"auto","created_at":"2025-12-10 15:32:19","extension":"jpeg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":225548,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/1495b9b11b04058f579030ce.jpeg"},{"id":97894311,"identity":"1c084aaf-fc30-465c-ba82-1865c53ee391","added_by":"auto","created_at":"2025-12-10 15:32:18","extension":"jpeg","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1892484,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/3015185920a734e354041b90.jpeg"},{"id":97693309,"identity":"b263c8be-c5d4-4e7f-9e4f-ded04720ba5a","added_by":"auto","created_at":"2025-12-08 11:16:39","extension":"jpeg","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1892484,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/abe4a056dc3cfd9e762afa02.jpeg"},{"id":97893001,"identity":"9ced4b8d-f568-46d1-b095-2df3fcc152e1","added_by":"auto","created_at":"2025-12-10 15:25:25","extension":"jpeg","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":247227,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/299009929b8426cdd3861cec.jpeg"},{"id":97693318,"identity":"40879594-fd43-4e42-a59d-e0dbe239f104","added_by":"auto","created_at":"2025-12-08 11:16:40","extension":"jpeg","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":246776,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage8.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/f9e1a2360642250d759c6aa2.jpeg"},{"id":97693315,"identity":"ec42b3e2-bc70-4471-b0fe-f82b41a175f1","added_by":"auto","created_at":"2025-12-08 11:16:40","extension":"jpeg","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":181788,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage9.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/9c50bb86155b3e52e1bb5749.jpeg"},{"id":97893754,"identity":"8628ebf6-8a61-4fcd-9ac9-0265d148e6c8","added_by":"auto","created_at":"2025-12-10 15:31:08","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":31760,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/9671ad9af428efcd6caccae4.png"},{"id":97693306,"identity":"a189bc99-91aa-416e-8769-27c5c5a060ec","added_by":"auto","created_at":"2025-12-08 11:16:39","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7996,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage10.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/4a6c98936389ad3aa2ce602e.png"},{"id":97693319,"identity":"f58f427d-0b85-48c2-9692-bcbdf0cee417","added_by":"auto","created_at":"2025-12-08 11:16:40","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":28327,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/3abca2054b5a79e5ade1cb08.png"},{"id":97693322,"identity":"f4b1f6cb-c4fe-4300-9c41-2bc824e1e622","added_by":"auto","created_at":"2025-12-08 11:16:40","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":46591,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/e1d3fa3ce7f7192c33117906.png"},{"id":97693305,"identity":"d5ed4943-6b70-4839-88fe-33ba5340b79e","added_by":"auto","created_at":"2025-12-08 11:16:39","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":48508,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/a66ea3d8e92018ca07ddfa4d.png"},{"id":97693321,"identity":"be2a4c8b-f33d-4bb2-a550-57636b162132","added_by":"auto","created_at":"2025-12-08 11:16:40","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":35649,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/55285b5fb76531e596753d68.png"},{"id":97894286,"identity":"8f66427f-2f5e-4164-903a-f8a6b358fd5a","added_by":"auto","created_at":"2025-12-10 15:32:13","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":36076,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/b793412a05e2ed96e720ebd5.png"},{"id":97693324,"identity":"8cc5d796-14d1-4933-9b50-a14ae45214d9","added_by":"auto","created_at":"2025-12-08 11:16:40","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":57190,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/47e3c17a167dd5353f8c7171.png"},{"id":97892776,"identity":"6b9ce661-53c0-4be2-a35b-dbc8375e8eec","added_by":"auto","created_at":"2025-12-10 15:20:54","extension":"png","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":56999,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/4f5440fc94efa675ccbaf88f.png"},{"id":97894113,"identity":"00df3e69-fd1c-4936-a2ee-58b0d53c6a5e","added_by":"auto","created_at":"2025-12-10 15:31:56","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":36182,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/7c745aa9e3229239314fb71d.png"},{"id":97693323,"identity":"2599aaef-8560-4aa9-89ea-b32667facf5c","added_by":"auto","created_at":"2025-12-08 11:16:40","extension":"xml","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":157016,"visible":true,"origin":"","legend":"","description":"","filename":"aa6a6746cf974d85b4d317aeba30664c1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/486e589bee26e9bd5521f96d.xml"},{"id":97894104,"identity":"df6a5ec5-b757-4f37-8da4-95e96a3beb20","added_by":"auto","created_at":"2025-12-10 15:31:56","extension":"html","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":171570,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/dfd5668dd863dc010f06c4b8.html"},{"id":97693297,"identity":"e0356cc4-4bc0-431a-bf1e-4448d64df459","added_by":"auto","created_at":"2025-12-08 11:16:39","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":179050,"visible":true,"origin":"","legend":"\u003cp\u003eResearch flowchart.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/b6a0eca19e87c98bf1aebf69.png"},{"id":97892800,"identity":"b618f56d-5fc0-4532-9f61-56db734c990a","added_by":"auto","created_at":"2025-12-10 15:22:04","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":449229,"visible":true,"origin":"","legend":"\u003cp\u003eVariable selection by Lasso regression and Boruta algorithm. \u003cstrong\u003ea.\u003c/strong\u003e Lasso coefficient path diagram of variables.\u003cstrong\u003e b.\u003c/strong\u003e Cross-validation curve (ten-fold cross-validation). The left dashed line indicates lambda.min, and the right dashed line indicates λ.1se. \u003cstrong\u003ec.\u003c/strong\u003e Iterative process of the Boruta algorithm. \u003cstrong\u003ed.\u003c/strong\u003e Feature selection results of the Boruta algorithm. The horizontal axis is the name of each variable, and the vertical axis is the Z value of each variable. The box plot shows the Z value of each variable during the model calculation process. The green box represents important variables, and the red box represents unimportant variables.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/df02e7b3d300eb497cd2bde3.png"},{"id":97693293,"identity":"4eb798d8-8519-4e4b-99eb-1c9f83f2112a","added_by":"auto","created_at":"2025-12-08 11:16:39","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":125898,"visible":true,"origin":"","legend":"\u003cp\u003eROC comparison of eight machine learning algorithm models. \u003cstrong\u003ea.\u003c/strong\u003e Training set ROC.\u003cstrong\u003e b. \u003c/strong\u003eTest set ROC. logit (LR): Logistic Regression; gbm (GBM): Gradient Boosting Machine; svm (SVM): Support Vector Machine; xgb (XGB): Extreme Gradient Boosting; enet (ENET): Elastic Network; knn (KNN): k-Nearest Neighbor; lda (LDA): Linear Discriminant Analysis; nnet (NNET): Neural Network.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/b018443c2b6d07dfe11b06eb.png"},{"id":97693299,"identity":"3e97abf2-0060-4bc8-b413-17874b594248","added_by":"auto","created_at":"2025-12-08 11:16:39","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":901126,"visible":true,"origin":"","legend":"\u003cp\u003eConfusion matrices of the eight models in the training set and validation set.\u003cstrong\u003ea.\u003c/strong\u003eConfusion matrices in the training set. \u003cstrong\u003eb\u003c/strong\u003e. Confusion matrices in the test set.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/95672dec96b32ff90753f982.png"},{"id":97894263,"identity":"8334e1fa-ffcf-4850-ba6b-37b0b5f1a04c","added_by":"auto","created_at":"2025-12-10 15:32:08","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":430088,"visible":true,"origin":"","legend":"\u003cp\u003eCalibration curves and DCA curves of the eight models in the training set and validation set. \u003cstrong\u003ea\u003c/strong\u003e. Training set calibration curve. \u003cstrong\u003eb\u003c/strong\u003e. Test set calibration curve.\u003cstrong\u003e c.\u003c/strong\u003e Training set DCA curve. \u003cstrong\u003ed.\u003c/strong\u003e Test set DCA curve. logit (LR): Logistic Regression; gbm (GBM): Gradient Boosting Machine; svm (SVM): Support Vector Machine; xgb (XGB): Extreme Gradient Boosting; enet (ENET): Elastic Network; knn (KNN): k-Nearest Neighbor; lda (LDA): Linear Discriminant Analysis; nnet (NNET): Neural Network\u003cstrong\u003e.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/3377d08d54d68c3a90ab1330.png"},{"id":97894061,"identity":"672eca72-1514-4a3f-8a18-bc6a62ba5bb2","added_by":"auto","created_at":"2025-12-10 15:31:54","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":388739,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP analysis based on the ENET model. \u003cstrong\u003ea\u003c/strong\u003e. Variable importance bar chart. \u003cstrong\u003eb.\u003c/strong\u003eVariable importance swarm plot. \u003cstrong\u003ec\u003c/strong\u003e. SHAP force plot. CD8: CD8+ T cell count. TC: Total cholesterol. HB: Hemoglobin. PLT: Platelets. WBC: White blood cell count.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/2500e1c4d0b592f311a6a8f3.png"},{"id":100363333,"identity":"fd2bbcaf-c8c2-42bc-9fd5-07ff32127407","added_by":"auto","created_at":"2026-01-16 07:49:27","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3316334,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8226880/v1/81938aba-102d-4f48-b60f-c2192b0768e5.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Explainable Machine Learning Predicts Advanced HIV Disease Progression Using Easily Accessible Hematological Markers","fulltext":[{"header":"Background","content":"\u003cp\u003eHIV/AIDS is a chronic infectious disease caused by the Human Immunodeficiency Virus (HIV), which impairs the immune system and makes individuals more susceptible to other infections and diseases [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e].In China, despite the widespread promotion of antiretroviral therapy (ART) having significantly reduced AIDS-related mortality, new infections and disease burden remain severe [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. HIV is detected in a large number of patients with advanced infection (defined by the World Health Organization (WHO) as a CD4\u0026thinsp;+\u0026thinsp;T cell count of less than 200 cells/mm\u0026sup3;) [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e], which is characterized by severe immunosuppression, recurrent opportunistic infections, and a high dependence on medical and supportive care. A study on trends across 55 countries showed [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e] that more than one-third (37%) of patients initiating ART in 2015 already had advanced HIV infection. Even after starting ART (which increases inflammatory responses), such patients have a high risk of death, and the risk increases with the decrease in CD4\u0026thinsp;+\u0026thinsp;T cells [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. If patients start ART as early as possible, their life expectancy is close to normal [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Early treatment can also reduce the risk of HIV transmission [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eCurrently, the main basis for assessing HIV/AIDS disease progression and staging in clinical practice is CD4\u0026thinsp;+\u0026thinsp;T cell count and HIV-RNA [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. However, these tests may not be immediately and routinely available in some primary medical institutions or economically underdeveloped areas, with certain technical, cost, and time barriers. Therefore, finding alternative or auxiliary predictive indicators that can be quickly, conveniently, and low-cost obtained before ART is of crucial significance for early identification of high-risk patients and advancing the threshold of clinical intervention.In recent years, an increasing number of studies have shown that some basic indicators in routine blood tests, such as hemoglobin [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e], platelets [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e], total cholesterol [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e], etc., are significantly associated with the immune status and disease progression of HIV-infected individuals.These indicators are easily accessible, low-cost, and provide rapid results, laying a valuable data foundation for constructing efficient clinical prediction models. In recent years, artificial intelligence technology, especially explainable machine learning methods, has provided a powerful tool for mining in-depth information from routine clinical data [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. However, there is currently a lack of research on constructing interpretable machine learning models based on these easily accessible blood indicators to accurately identify patients who are about to progress to the advanced stage of HIV disease.\u003c/p\u003e\u003cp\u003eThis study aims to utilize readily available hematological indicators to develop and validate an interpretable machine learning model. The goal is to provide clinicians with a simple and effective tool at the beginning of ART treatment for early identification of HIV-infected individuals with a high risk of progressing to advanced stages of the disease. This will provide scientific basis for achieving precise intervention and optimizing the allocation of medical resources. Moreover, it seeks to identify conventional hematological indicators that can effectively predict the progression to the advanced stage of HIV disease in resource-limited environments, ultimately providing scientific support for achieving precise intervention and optimizing the allocation of medical resources.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eData Source\u003c/h2\u003e\u003cp\u003eThe data is derived from HIV/AIDS patients who received antiretroviral therapy in Linhai City, Taizhou City, Zhejiang Province from January 2010 to June 2025.The following indicators were included:\u003c/p\u003e\u003cp\u003e\u003col\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eHematological indicators: viral load, white blood cell count (WBC), platelets (PLT), hemoglobin (HB), serum creatinine (SCR), triglycerides (TG), total cholesterol (TC), blood glucose, ALT, AST, T.BIL, HBsAG, HbeAg, AntiHCV, and CD8\u0026thinsp;+\u0026thinsp;T cells measured for the first time after diagnosis of HIV/AIDS [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e].\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eDemographic data: age at diagnosis, gender, ethnicity, educational level, occupation, and marital status.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eOther variables: history of sexually transmitted diseases (STD) and transmission route. Among the above variables, gender (female, male), ethnicity (Han, other), educational level (illiterate, primary school, junior high school, high school or technical secondary school, college and above), HBsAG (negative, positive), HbeAg (negative, positive), AntiHCV (negative, positive), occupation (commercial sex worker, other worker, unemployed), marital status (married or with spouse, unmarried, divorced or widowed), history of STD (yes, no), and transmission route (homosexual transmission, heterosexual transmission, other routes) were categorical variables, and the other variables were continuous variables.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003c/p\u003e\u003cp\u003eThe outcome variable was progression to advanced HIV infection, defined as CD4\u0026thinsp;+\u0026thinsp;T cell count\u0026thinsp;\u0026lt;\u0026thinsp;200 cells/mm\u0026sup3; [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Patients aged 18\u0026ndash;85 years who received ART during the study period were included, while those with missing post-diagnosis CD4\u0026thinsp;+\u0026thinsp;T cell data were excluded. A total of 709 patients were included.\u003c/p\u003e\u003cp\u003e The sample size of this study was calculated using the \u0026lsquo;pmsampsize\u0026rsquo; package in R software, which was developed based on the guidelines proposed by RD Riley et al. in 2018 to calculate the minimum sample size required for multivariable prediction models in clinical research [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. After calculation, the minimum sample size required for this study was 357, so the sample size in the study met the requirements. This study has been approved by the Ethics Review Committee of Taizhou Center for Disease Control and Prevention (Taizhou Health Supervision Institute) (Approval No.: Taizhou Disease Control Center Review and Approval Document No. 008 of 2025 Research Project).The study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki.Informed consent has been waived by using cases and biological specimens obtained from previous clinical diagnoses and treatments. All personal data in the study have been anonymized during the data organization and statistical analysis stages to ensure participant privacy. No compensation was provided to the participants.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003eStatistical Analysis\u003c/h2\u003e\u003cp\u003eVariables with more than 30% missing values were removed from the dataset [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Variables with more than 30% missing values included viral load, blood glucose, HbeAg, and AntiHCV. For variables with less than 30% missing values, the \u0026lsquo;MissForest\u0026rsquo; package in R software was used for imputation [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e].Descriptive analysis was performed on the preprocessed data using the \u0026lsquo;compareGroups\u0026rsquo; package in R software. Measurement data were expressed as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation, and independent sample t-test was used for comparison between groups; count data were expressed as frequency (%), and chi-square test was used for comparison between groups. Data standardization plays an important role in improving the performance of machine learning models. It can scale data of different dimensions or scales to the same interval, thereby increasing the comparability between variables. In this study, the scale() function of the base package in R software was used for Z-score standardization of the data. The standardized data conforms to a standard normal distribution with a mean of 0 and a standard deviation of 1.\u003c/p\u003e\u003cp\u003eThe dataset was divided into a training set and a test set randomly at a ratio of 7:3 using the sample.split function in the \u0026lsquo;caTools\u0026rsquo; package. Variable selection was performed by combining LASSO regression and the Boruta algorithm. LASSO regression is an improved linear regression method mainly used for feature selection and model regularization [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. In contrast, the Boruta algorithm determines the relevance of features by comparing the importance of features with that of randomly permuted (noise) features. The Boruta algorithm adopts the feature selection method of random forest, and the number of trees in the random forest is set to 1000 to enhance the stability and accuracy of the model [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eFive-fold cross-validation was used in the training set to optimize model parameters and establish eight machine learning models:Logistic Regression (LR), Support Vector Machine (SVM), Gradient Boosting Machine (GBM), Elastic Network (ENET), Neural Network (NNET), Extreme Gradient Boosting (XGBoost), k-Nearest Neighbor (KNN), and Linear Discriminant Analysis (LDA).LR is a generalized linear regression analysis model used to study the influence relationship between categorical dependent variables and independent variables [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. SVM is a supervised learning model based on statistical learning theory, widely used in classification and regression tasks [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. GBM is an ensemble learning model widely used in the field of machine learning [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. KNN is a classic supervised learning algorithm, widely used in classification and regression tasks [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. ENET is a linear regression method that combines the advantages of Lasso regression (L1 regularization) and ridge regression (L2 regularization), mainly used to solve problems such as high-dimensional data and feature correlation [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. NNET overcomes the limitations of traditional artificial intelligence methods in processing complex and unstructured information [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. XGBoost uses the second-order Taylor series to approximate the value of the loss function, and further reduces the possibility of overfitting through regularization [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. Linear Discriminant Analysis (LDA) is a classic statistical machine learning method that aims to find a linear data transformation that increases class discrimination in the optimal discriminant subspace [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eThe predictive performance of the models was evaluated from three dimensions: discriminative ability, calibration, and clinical utility. The Area Under the Receiver Operating Characteristic Curve (AUC) was used to evaluate the discriminative ability of the model. The closer the AUC is to 1, the stronger the diagnostic discrimination ability of the model. The Calibration Curve was used to evaluate the calibration of the model. The x-axis and y-axis of the calibration curve represent the predicted probability and the actual probability, respectively. A model with good predictive performance should have a calibration curve close to the diagonal line, indicating that the predicted probability is basically consistent with the actual probability. Clinical utility was evaluated using Decision Curve Analysis (DCA). The decision curve is a smooth curve formed by connecting the net benefits under different threshold probabilities. There are two fixed lines in the decision curve graph: the line parallel to the x-axis refers to the net benefit of not treating any patients, and the other diagonal line refers to the net benefit of treating all patients. The larger the area under the curve formed by the model's decision curve and these two lines, the higher the clinical net benefit of the model. In addition, the DeLong test [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e] was used to compare whether there were significant differences in AUC between different models, and a p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically significant.\u003c/p\u003e\u003cp\u003eWe used SHAP (SHapley Additive exPlanations) values to evaluate the overall feature importance in the ML model with the best predictive performance. SHAP measures the importance of each feature to the model by generating a contribution value (shapley) of each feature in the model to the predicted outcome [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. SHAP values can show the positive or negative contribution of each predictor variable to the target variable.\u003c/p\u003e\u003cp\u003eThe data analysis process of this study was performed using R4.4.2, and a p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically significant. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e is the flowchart of this study.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003eThis study included 709 HIV/AIDS patients. Among them, 260 cases (36.6%) progressed to the advanced stage of HIV disease (with CD4\u0026thinsp;+\u0026thinsp;T cell count below 200 cells/mm3).Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e shows the comparison of baseline characteristics between the training set and the test set. Among them, the average age of HIV/AIDS patients was 43.2 years old. The Han ethnicity accounted for the vast majority, with 684 cases (96.5%), and the majority were male patients, with 585 cases (82.5%). Most of the patients were married or had a spouse, with 402 cases (56.7%). Compared to other stages, a higher proportion had a junior high school education, with 260 cases (36.7%). There were 60 cases (8.5%) with positive HBsAg, 44 cases (6.2%) were sex workers, and 137 cases (19.3%) had a history of sexually transmitted diseases. The majority were transmitted through heterosexual contact, with 437 cases (61.6%).Except for the transmission route, there were no statistically significant differences in other variables between the two groups (P\u0026thinsp;\u0026gt;\u0026thinsp;0.05), indicating that the two groups were balanced and comparable.\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\u003eComparison of baseline characteristics between the training set and test set of HIV/AIDS patients.\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=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" 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\" morerows=\"1\" rowspan=\"2\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eOverall\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003etest\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003etrain\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003ep.overall\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eN\u0026thinsp;=\u0026thinsp;709\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003eN\u0026thinsp;=\u0026thinsp;212\u003c/b\u003e\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003eN\u0026thinsp;=\u0026thinsp;497\u003c/b\u003e\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCD8+(cells /ul)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e936 (539)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e929 (544)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e938 (538)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.839\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWBC(10^9/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e5.58 (2.31)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5.74 (3.18)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e5.51 (1.82)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.340\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePLT(10^9/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e208 (72.7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e199 (77.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e212 (70.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.038\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHB(g/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e138 (21.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e137 (22.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e138 (21.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.488\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSCR(\u0026micro;mol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e69.1 (20.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e68.7 (14.6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e69.3 (22.3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.665\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTG(mmol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.86 (2.51)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.92 (1.73)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.84 (2.78)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.665\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTC(mmol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e4.25 (0.864)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4.17 (0.85)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e4.28 (0.87)\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\u003eALT(U/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e30.6 (44.1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e30.2 (28.0)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e30.8 (49.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.849\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAST(U/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e29.1 (28.2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e28.8 (18.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e29.3 (31.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.819\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eT.BIL(\u0026micro;mol/L)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e13.0 (24.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11.6 (5.48)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e13.5 (29.4)\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\u003eAge(years)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e43.2 (15.5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e44.5 (15.8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e42.7 (15.4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.173\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHBsAG:\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\u003e0.472\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNegative\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e649 (91.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e197 (92.9%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e452 (90.9%)\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\u003ePositive\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e60 (8.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e15 (7.08%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e45 (9.05%)\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\u003eSex:\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\u003e0.901\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=\"left\" colname=\"c2\"\u003e\u003cp\u003e124 (17.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e36 (17.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e88 (17.7%)\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=\"left\" colname=\"c2\"\u003e\u003cp\u003e585 (82.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e176 (83.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e409 (82.3%)\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\u003eNationality:\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\u003e0.368\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHan\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e684 (96.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e202 (95.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e482 (97.0%)\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=\"left\" colname=\"c2\"\u003e\u003cp\u003e25 (3.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e10 (4.72%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e15 (3.02%)\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\u003eEducation:\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\u003e0.736\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCollege\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e89 (12.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e26 (12.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e63 (12.7%)\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\u003eHigh\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e124 (17.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e37 (17.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e87 (17.5%)\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\u003eIlliterate\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e56 (7.9%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e18 (8.49%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e38 (7.65%)\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\u003eMiddle\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e260 (36.7%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e71 (33.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e189 (38.0%)\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\u003ePrimary\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e180 (25.4%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e60 (28.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e120 (24.1%)\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\u003eWork:\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\u003e0.427\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCSW\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e44 (6.2%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e12 (5.66%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e32 (6.44%)\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=\"left\" colname=\"c2\"\u003e\u003cp\u003e566 (79.8%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e165 (77.8%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e401 (80.7%)\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\u003eUnemployed\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e99 (14.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e35 (16.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e64 (12.9%)\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\u003eMarital:\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\u003e0.972\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDivorced/widowed\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e93 (13.1%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e27 (12.7%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e66 (13.3%)\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\u003eMarried\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e402 (56.7%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e120 (56.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e282 (56.7%)\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\u003eUnmarried\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e214 (30.2%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e65 (30.7%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e149 (30.0%)\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\u003eSTD:\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\u003e0.256\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=\"left\" colname=\"c2\"\u003e\u003cp\u003e572 (80.7%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e177 (83.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e395 (79.5%)\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=\"left\" colname=\"c2\"\u003e\u003cp\u003e137 (19.3%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e35 (16.5%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e102 (20.5%)\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\u003eRoute:\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\u003e0.044\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHeterosexual\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e437 (61.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e125 (59.0%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e312 (62.8%)\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\u003eHomosexual\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e267 (37.7%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e83 (39.2%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e184 (37.0%)\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=\"left\" colname=\"c2\"\u003e\u003cp\u003e5 (0.7%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4 (1.89%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1 (0.20%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\n\u003ch3\u003eFeature Selection\u003c/h3\u003e\n\u003cp\u003eVariable selection was performed by combining Lasso regression and the Boruta algorithm. As shown in Fig.\u0026nbsp;7a,b, Lasso regression used ten-fold cross-validation to select λ. In this study, λ.min was used as the optimal λ value, and the selected variables were CD8\u0026thinsp;+\u0026thinsp;T cells, WBC, PLT, HB, SCR, TG, TC, AST, T.BIL, gender, ethnicity, and marital status. As shown in Fig.\u0026nbsp;7c,d, the variables most closely related to the progression of advanced HIV disease screened by the Boruta algorithm were CD8\u0026thinsp;+\u0026thinsp;T cells, WBC, TC, HB, AST, PLT, and T.BIL. The variables jointly screened by the two methods were: CD8, WBC, TC, HB, AST, PLT, and T.BIL. Finally, these seven variables were included in the eight machine learning algorithm models.\u003c/p\u003e\n\u003ch3\u003eEstablishment and Evaluation of Eight Machine Learning Algorithm Models\u003c/h3\u003e\n\u003cp\u003eThe 709 subjects were randomly divided into a training set (497 people) and a test set (212 people) at a ratio of 7:3. Using the seven variables screened above, eight machine learning models (LR, SVM, GBM, ENET, NNET, XGBoost, KNN, and LDA) were constructed. The predictive ability of the models was evaluated from three dimensions: discriminative ability, calibration, and clinical utility. The ROC curves of the eight models in the training set and validation set are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea,b. In the training set, XGBoost had the highest AUC (0.876), followed by GBM, KNN, SVM, NNET, LR, ENET, and LDA (AUC\u0026thinsp;=\u0026thinsp;0.793) was the lowest. In the validation set, ENET was the highest (AUC\u0026thinsp;=\u0026thinsp;0.801), followed by LR, GBM, LDA, XGBoost, NNET, SVM, and KNN (AUC\u0026thinsp;=\u0026thinsp;0.768) was the lowest. In the validation set, pairwise comparison of the models by the DeLong test (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) showed that only the difference in AUC between the ENET model (AUC\u0026thinsp;=\u0026thinsp;0.801) and the LDA model (AUC\u0026thinsp;=\u0026thinsp;0.794) was statistically significant (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05), and the differences in AUC between other models were not significant (P\u0026thinsp;\u0026gt;\u0026thinsp;0.05). Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e lists the detailed comparison of specific performance indicators of each model. Although the accuracy, sensitivity, specificity, precision, and F1 score of the ENET model are not the highest, they are all at a relatively reasonable level, and overall, it can better solve the research problem(Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e provides the confusion matrices of each model in the training set and test set.\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\u003ePairwise comparison of AUC values of eight models in the test set.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" 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=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAUC1\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eModel2\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAUC2\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eZ_statistic\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eP_value\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eSignificant\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.797\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSVM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.769\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.627\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.1037\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.797\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGBM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.795\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.077\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.9384\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.797\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eENET\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.801\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.587\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.5569\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.797\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNNET\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.307\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.1913\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.797\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eXGB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.652\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.5143\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.797\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eKNN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.768\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.397\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.1625\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.797\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLDA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.794\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.606\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.5448\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.769\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGBM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.795\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-1.188\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.2348\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.769\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eENET\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.801\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-1.882\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.0599\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.769\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNNET\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.663\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.5073\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.769\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eXGB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.508\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.6112\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.769\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eKNN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.768\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.052\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.9584\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.769\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLDA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.794\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-1.429\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.1529\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.795\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eENET\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.801\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.209\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.8346\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.795\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNNET\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.597\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.5507\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.795\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eXGB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.584\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.1132\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.795\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eKNN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.768\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.984\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.3253\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.795\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLDA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.794\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.055\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.9559\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.801\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNNET\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.481\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.1387\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.801\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eXGB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.797\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.4253\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.801\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eKNN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.768\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e1.707\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.0878\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.801\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLDA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.794\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e2.021\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.0433\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eXGB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.026\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.9793\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eKNN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.768\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.543\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.587\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLDA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.794\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.957\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.3385\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eKNN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.768\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.461\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.6447\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.781\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLDA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.794\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.509\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.611\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e0.768\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLDA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.794\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-1.328\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.1843\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eNo\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 parameters of eight prediction models.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"7\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" 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\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eModel\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\u003eSensitivity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eSpecificity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003ePrecision\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003eF1\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"7\" rowspan=\"8\"\u003e\u003cp\u003e\u003cb\u003eTrain set\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.757\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.533\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.886\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.729\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.616\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSVM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.807\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.637\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.905\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.795\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e0.707\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGBM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.799\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.615\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.905\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.789\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.691\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eENET\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.755\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.505\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.898\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.742\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.601\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNNET\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\u003e0.632\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.854\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.714\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.671\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eXGB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.801\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.604\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.914\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.803\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.690\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eKNN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.797\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.615\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.902\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.783\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.689\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLDA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.753\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.500\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.898\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.740\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.597\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"7\" rowspan=\"8\"\u003e\u003cp\u003e\u003cb\u003eTest set\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLR\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.764\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.526\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.903\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.759\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e0.621\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSVM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.712\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.474\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.851\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.649\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.548\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGBM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.736\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.513\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.866\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.690\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.588\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eENET\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.764\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.513\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.910\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.769\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.615\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNNET\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.726\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.538\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.836\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.656\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.592\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eXGB\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.731\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.487\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.873\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.691\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.571\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eKNN\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.741\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.551\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.851\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.683\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.610\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLDA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.769\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.513\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.918\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.784\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.620\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\u003eWe plotted calibration curves and DCA curves based on the training and test sets. The former is used to evaluate the accuracy and reliability of model predictions, while the latter is used to evaluate the potential clinical utility of the model within different threshold ranges. The calibration curves of the eight models are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea,b. In the training set, all models except the SVM model showed good calibration. In the test set, the NNET, GBM, and XGBoost models had poor calibration, and the remaining models had good calibration. The DCA curves of the eight models (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec,d) showed that within a wide range of thresholds, the net benefits of these models were higher than the \"intervene\" or \"no intervene\" strategies. In the training set, within different threshold probability ranges, the XGBoost and GBM models showed more significant clinical benefits compared with other models. In the test set, the ENET model had the highest net benefit and good clinical utility. Based on the comprehensive evaluation results of model performance, the ENET model exhibited superior discriminative ability and clinical utility, so it is the optimal model for identifying progression to advanced HIV disease.\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eModel Interpretation - SHAP Analysis\u003c/h2\u003e\u003cp\u003eIn this study, the SHAP algorithm was used to explain the importance of predictive variables in the ENET model with the best predictive performance (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). The contribution of variables to the model was reflected by SHAP values. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea ranks the importance of each variable in descending order according to the absolute value of the average SHAP value. CD8\u0026thinsp;+\u0026thinsp;T cells had the highest average SHAP value, contributing the most to model prediction. Among the easily accessible hematological markers, TC had the greatest contribution, followed by HB, AST, PLT, WBC and T.BIL. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eb is a swarm plot of variable importance of the ENET model. The x-axis represents the size of the SHAP value, and the order of variables from top to bottom on the y-axis is arranged in descending order of importance. Each point in the figure represents a sample. The closer the color of the point is to purple, the lower its SHAP value; the closer it is to yellow, the higher its value. When the SHAP value is greater than 0, the variable has a positive contribution to the prediction result; when it is less than 0, it has a negative contribution. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ec is a SHAP force plot. If the feature block is yellow and the SHAP value is positive: the current value of the feature makes the prediction result higher than the \"baseline value\" (the average prediction value of the model for all samples, E[f(x)]\u0026thinsp;=\u0026thinsp;0.245 in the figure is the baseline value). If the feature block is purple and the SHAP value is negative: the current value of the feature makes the prediction result lower than the \"baseline value\".\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eSince the first AIDS-related death was reported in China in 1985 [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e], the Chinese government has implemented a series of AIDS prevention and control policies and achieved certain results [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. Early diagnosis of HIV in suspected populations is a prerequisite for successful treatment of AIDS.Although early diagnosis of HIV is of great significance in patient treatment, studies have shown that late diagnosis remains a major problem in global AIDS control and prevention [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Late diagnosis of HIV reduces the effectiveness of ART, thereby increasing the risk of morbidity and mortality in patients, which has adverse effects on individuals and society.On the other hand, such late diagnosis increases the possibility of disease transmission and spread in society, making disease control more difficult [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e].In addition, the main basis for assessing HIV/AIDS disease progression and staging is CD4\u0026thinsp;+\u0026thinsp;T cell count and HIV-RNA [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e], which may not be achievable in economically underdeveloped areas or primary medical institutions, providing the possibility of using easily accessible routine hematological indicators to assist treatment. Therefore, this study also used routine, low-cost, and easily accessible hematological indicators based on machine learning algorithm models to quickly and accurately identify patients who seem stable but are actually at high risk of progressing to advanced HIV disease at the critical time point of initiating ART providing a scientific basis for early diagnosis and intervention.\u003c/p\u003e\u003cp\u003eThis study collected data on HIV/AIDS patients who received antiretroviral therapy in Linhai City, Taizhou City, Zhejiang Province from January 2010 to June 2025.A total of 709 HIV/AIDS patients were included. Among them, 260 cases (36.6%) had progressed to the advanced stage of HIV disease (with CD4\u0026thinsp;+\u0026thinsp;T cell count below 200 cells/mm3) before starting ART.Seven predictors were screened by combining Lasso regression and the Boruta algorithm to construct eight machine learning models, and the model performance was evaluated from three aspects: discriminative ability, calibration, and clinical utility. Based on the comprehensive evaluation indicators of the training set and validation set, the ENET model was selected as the optimal model. Although its accuracy, sensitivity, specificity, precision, and F1 score are not the highest, they are all at a relatively reasonable level, and overall, it can better solve the research problem. The AUC of the ENET model in the training set was 0.794, and the AUC in the validation set was 0.801, with good calibration and clinical utility. After confirming that the prediction model for progression to advanced HIV disease constructed based on easily accessible routine hematological indicators before antiretroviral therapy has a good predictive effect, and the ENET model performs the best, it is particularly crucial to conduct an in-depth analysis of the important variables involved in model construction.\u003c/p\u003e\u003cp\u003eIn recent years, machine learning has been widely used in the field of infectious diseases. However, due to the \"black box\" nature of machine learning, its interpretability is relatively poor, so it is difficult to explain why specific predictions are made for patients [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e].In this study, we used SHAP analysis to explain the optimal model in detail. SHAP analysis allowed us to evaluate the contribution of different variables to the model's prediction results. This study screened seven variables, and the importance of these seven variables was ranked based on the SHAP algorithm. The results of this study showed that CD8\u0026thinsp;+\u0026thinsp;T cell count contributed the most to the model's prediction of HIV disease progression. As a key immune cell indicator, the indicating effect of changes in CD8\u0026thinsp;+\u0026thinsp;T cell levels on disease progression is highlighted in the model. This is consistent with previous research conclusions [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e] that a decrease in CD8\u0026thinsp;+\u0026thinsp;T cells is the best independent predictor of HIV disease progression, especially when the CD4\u0026thinsp;+\u0026thinsp;T cell count is below 200 cells/mm\u0026sup3;, a decrease in CD8\u0026thinsp;+\u0026thinsp;T cells will significantly increase the risk of HIV disease progression. From a pathological mechanism perspective, when the number of CD4\u0026thinsp;+\u0026thinsp;T cells drops to an extremely low level of \u0026lt;\u0026thinsp;200 cells/\u0026micro;L, the immune system collapses completely; at this time, even CD8\u0026thinsp;+\u0026thinsp;T cells, which act as the \"emergency force\" of the immune system, cannot maintain their numbers due to the complete failure of the immune system, and their functions and numbers will decrease significantly. At the same time, persistent immune activation caused by HIV infection leads to exhaustion of CD8\u0026thinsp;+\u0026thinsp;T cells, making them gradually lose immune function and eventually undergo apoptosis, which is also an important driving factor for the decrease in the number of CD8\u0026thinsp;+\u0026thinsp;T cells in progressive HIV infection [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e].In our study, SHAP analysis showed that among the easily accessible routine hematological indicators, TC and HB also had relatively high contribution values, second only to CD8\u0026thinsp;+\u0026thinsp;T cells. A study by Melaku et al. [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e] showed that low serum total cholesterol and its combination with anemia showed high sensitivity\u0026thinsp;\u0026gt;\u0026thinsp;80.0. Total cholesterol can be used as an important biomarker because lipids play a role in virus entry, uncoating, replication, protein synthesis, assembly, budding, and infectivity [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. Cellular cholesterol is crucial for HIV replication and may control HIV transmission [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. Compared with individuals with low cholesterol levels, immune system cells of patients with hypercholesterolemia have higher phagocytic activity, more circulating lymphocytes, more total T cells, more CD8\u0026thinsp;+\u0026thinsp;T cells, more immunoglobulin production, more proliferation and differentiation, and migration of lymphocytes [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. This can be explained by the role of intermediates in the cholesterol biosynthesis pathway and the role of downstream oxysterol metabolites that have been found to affect various functions of immune system cells. In addition, the effect of ART is impaired in HIV-infected patients with hypocholesterolemia [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e].This may mean that in this era of testing and treatment, low serum TC may be used as a predictive marker for ART efficacy.Anemia is a common complication, occurring in 20\u0026ndash;80% of HIV-infected individuals, and is associated with faster disease progression and higher mortality rates [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e].HB levels reflect the speed of disease progression and independently predict prognosis in different populations [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e].The rate of HB decline is also associated with a decrease in CD4 count [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]. A study by Christian et al. [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e] showed that HB measurement can be used as an indicator of HIV/AIDS progression in resource-limited settings.In addition, changes in indicators such as AST, PLT, WBC, and T.BIL are also helpful for identifying progression to advanced HIV disease.These biomarkers can improve the performance of doctors in accurately examining the disease prognosis of patients receiving ART. If doctors only have CD4\u0026thinsp;+\u0026thinsp;T cell counts, it is not easy to switch treatment because the recovery rate of CD4\u0026thinsp;+\u0026thinsp;T cells is not high enough, especially in patients who start ART late [\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e].Using alternative biomarkers to monitor ART efficacy by extending the interval between testing CD4\u0026thinsp;+\u0026thinsp;T cell counts and HIV RNA load is a very useful tool. Therefore, in this case, basic laboratory tests of alternative biomarkers and effective clinical monitoring will be very helpful [\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e]. Therefore, HIV/AIDS patients should pay special attention to the fluctuations of these indicators before starting ART.\u003c/p\u003e\u003cp\u003eThis model is based on easily accessible routine hematological indicators and shows considerable effectiveness even in resource-limited settings where CD4\u0026thinsp;+\u0026thinsp;cell counts and HIV-RNA are not available. However, this study also has its limitations. This study was conducted only in one county-level city in Taizhou, Zhejiang Province, introducing potential selection bias. Therefore, external validation from other regions is needed to improve the generalization of the model. Secondly, the study is cross-sectional, which can identify the association between risk variables and outcome variables, but cannot identify causal relationships. The current model fails to capture the dynamic changes of key indicators, which may limit its application in the dynamic monitoring and management of HIV/AIDS patients.Future studies should include longitudinal data to explore the performance of the new model at different time points and analyze the impact of changes in key indicators over time on outcomes. In addition, due to limitations in detection technology, a large number of data such as viral load and serological evidence of hepatitis B/C in earlier years are missing. Due to limitations in data access rights, comorbidities such as tuberculosis, family history of dyslipidemia, emotional changes, depression, and lifestyle-related factors (smoking and lack of exercise) have not been explained.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn this study, the ENET model constructed based on seven screened variables showed the optimal performance, with an AUC of 0.801 in the test set, along with good calibration and clinical utility, making it an effective tool for identifying high-risk patients of advanced disease. CD8\u0026thinsp;+\u0026thinsp;T cells are the core immune indicator for predicting HIV progression to the advanced stage, and TC is the primary predictive factor among clinically easily accessible hematological markers. The combination of these two indicators with other hematological markers (such as HB) can provide a convenient and reliable reference for assessing HIV disease progression risk in resource-limited settings.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eHIV\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eHuman Immunodeficiency Virus\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eAIDS\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eAcquired Immune Deficiency Syndrome\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eART\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eAntiretroviral Therapy\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eSTD\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eSexually Transmitted Diseases\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eWBC\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eWhite Blood Cell Count\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003ePLT\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003ePlatelets\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eHB\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eHemoglobin\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eSCR\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eSerum Creatinine\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eTG\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eTriglycerides\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eTC\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eTotal Cholesterol\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eAUC\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eArea Under the Curve\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eDCA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eDecision Curve Analysis\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eLR\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eLogistic Regression\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eSVM\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eSupport Vector Machine\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eGBM\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eGradient Boosting Machine\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eKNN\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003ek-Nearest Neighbor\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eENET\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eElastic Network\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eNNET\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eNeural Network\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eXGBoost\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eExtreme Gradient Boosting\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eLDA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eLinear Discriminant Analysis\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eSHAP\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eSHapley Additive exPlanations\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData were obtained from HIV/AIDS patients receiving ART in Linhai City, Taizhou, Zhejiang Province, from January 2010 to June 2025. The study was approved by the Ethics Review Committee of Taizhou Center for Disease Control and Prevention (Taizhou Health Supervision Institute) (Approval NO.: Taizhou Disease Control Center Review and Approval Document No. 008 of 2025 Research Project) in accordance with the Declaration of Helsinki. Informed consent has been waived by using cases and biological specimens obtained from previous clinical diagnoses and treatments. The ethics review committee of Taizhou Center for Disease Control and Prevention (Taizhou Health Supervision Institute) approved the exemption of informed consent.All personal data in the study have been anonymized during the data organization and statistical analysis stages to ensure participant privacy. No compensation was provided to the participants.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors reviewed and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial number\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 data for this study were obtained from the Disease Prevention and Control Center of Linhai City, Taizhou City, Zhejiang Province, China.Data sources and handling of these data are described in the \u0026ldquo;Materials and Methods\u0026rdquo;.The data of HIV/AIDS patients in Linhai City, Taizhou City, Zhejiang Province were obtained by the corresponding author:Email: [email protected].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests \u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no conflict of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNone.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors' contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eX.L. proposed this idea, designed the research plan and supervised the entire process. Z.Z. supervised the entire work and provided clinical knowledge.Q. Z. was responsible for the conception and design of the research, data analysis, article drafting and writing. J. L. collected the data and wrote the article. S. W., revised the content.L. C.，X. B., and W. Z. reviewed the relevant literature.Z.L. and Y.Z. provided methodological guidance.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements \u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe are extremely grateful to Linhai Center for Disease Control and Prevention for providing us with the data support.The authors highly appreciate all the members who were involved in the present study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eLi, X. C. et al. Global burden of viral infectious diseases of poverty based on Global Burden of Diseases Study 2021. Infect Dis Poverty. ;13(1):71. Published 2024 Oct 8. (2024). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s40249-024-01234-z\u003c/span\u003e\u003cspan address=\"10.1186/s40249-024-01234-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXu, B., Li, J. \u0026amp; Wang, M. Epidemiological and time series analysis on the incidence and death of AIDS and HIV in China. \u003cem\u003eBMC Public. Health\u003c/em\u003e. \u003cb\u003e20\u003c/b\u003e (1), 1906. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12889-020-09977-8\u003c/span\u003e\u003cspan address=\"10.1186/s12889-020-09977-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020). Published 2020 Dec 14.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhai, M. et al. The trend of AIDS in China: A prediction and comparative analysis with G20 countries based on the Global Burden of Disease Study 2019. J Glob Health. ;14:04029. Published 2024 Mar 1. (2024). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7189/jogh.14.04029\u003c/span\u003e\u003cspan address=\"10.7189/jogh.14.04029\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZeng, Y. et al. Exploring palliative care needs of advanced AIDS patients in China: insights from a descriptive phenomenological study. \u003cem\u003eBMC Palliat. Care\u003c/em\u003e. \u003cb\u003e24\u003c/b\u003e (1), 42. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12904-025-01675-9\u003c/span\u003e\u003cspan address=\"10.1186/s12904-025-01675-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2025). Published 2025 Feb 13.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFord, N. \u0026amp; Doherty, M. The Enduring Challenge of Advanced HIV Infection. \u003cem\u003eN Engl. J. Med.\u003c/em\u003e \u003cb\u003e377\u003c/b\u003e (3), 283\u0026ndash;284. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1056/NEJMe1707598\u003c/span\u003e\u003cspan address=\"10.1056/NEJMe1707598\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eIeDEA et al. Immunodeficiency at the start of combination antiretroviral therapy in low-, middle-, and high-income countries. \u003cem\u003eJ. Acquir. Immune Defic. Syndr.\u003c/em\u003e \u003cb\u003e65\u003c/b\u003e (1), e8\u0026ndash;e16. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/QAI.0b013e3182a39979\u003c/span\u003e\u003cspan address=\"10.1097/QAI.0b013e3182a39979\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2014).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAntiretroviral Therapy Cohort Collaboration. Survival of HIV-positive patients starting antiretroviral therapy between 1996 and 2013: a collaborative analysis of cohort studies. \u003cem\u003eLancet HIV\u003c/em\u003e. \u003cb\u003e4\u003c/b\u003e (8), e349\u0026ndash;e356. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S2352-3018(17)30066-8\u003c/span\u003e\u003cspan address=\"10.1016/S2352-3018(17)30066-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCohen, M. S. et al. Antiretroviral Therapy for the Prevention of HIV-1 Transmission. \u003cem\u003eN Engl. J. Med.\u003c/em\u003e \u003cb\u003e375\u003c/b\u003e (9), 830\u0026ndash;839. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1056/NEJMoa1600693\u003c/span\u003e\u003cspan address=\"10.1056/NEJMoa1600693\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2016).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eObirikorang, C. \u0026amp; Yeboah, F. A. Blood haemoglobin measurement as a predictive indicator for the progression of HIV/AIDS in resource-limited setting. \u003cem\u003eJ. Biomed. Sci.\u003c/em\u003e \u003cb\u003e16\u003c/b\u003e (1), 102. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/1423-0127-16-102\u003c/span\u003e\u003cspan address=\"10.1186/1423-0127-16-102\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2009). Published 2009 Nov 18.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWu, F. et al. Increased platelet-CD8\u0026thinsp;+\u0026thinsp;T-cell aggregates displaying high activation, exhaustion, and tendency to death correlate with disease progression in people with HIV-1. \u003cem\u003eJ. Leukoc. Biol.\u003c/em\u003e \u003cb\u003e116\u003c/b\u003e (1), 166\u0026ndash;176. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/jleuko/qiae048\u003c/span\u003e\u003cspan address=\"10.1093/jleuko/qiae048\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDeng, M. et al. Over-activation and dysfunction of platelet-NK cell aggregates in HIV-infected individuals. J Transl Med. ;23(1):584. Published 2025 May 26. (2025). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12967-025-06591-3\u003c/span\u003e\u003cspan address=\"10.1186/s12967-025-06591-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSamuel, M. et al. The effects of age on associations between markers of HIV progression and markers of metabolic function including albumin, haemoglobin and lipid concentrations. \u003cem\u003eHIV Med.\u003c/em\u003e \u003cb\u003e15\u003c/b\u003e (5), 311\u0026ndash;316. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/hiv.12103\u003c/span\u003e\u003cspan address=\"10.1111/hiv.12103\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2014).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePonce-Bobadilla, A. V., Schmitt, V., Maier, C. S., Mensing, S. \u0026amp; Stodtmann, S. Practical guide to SHAP analysis: Explaining supervised machine learning model predictions in drug development. \u003cem\u003eClin. Transl Sci.\u003c/em\u003e \u003cb\u003e17\u003c/b\u003e (11), e70056. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/cts.70056\u003c/span\u003e\u003cspan address=\"10.1111/cts.70056\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKvale, D., Aukrust, P., Osnes, K., M\u0026uuml;ller, F. \u0026amp; Fr\u0026oslash;land, S. S. CD4\u0026thinsp;+\u0026thinsp;and CD8\u0026thinsp;+\u0026thinsp;lymphocytes and HIV RNA in HIV infection: high baseline counts and in particular rapid decrease of CD8\u0026thinsp;+\u0026thinsp;lymphocytes predict AIDS. \u003cem\u003eAIDS\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e (2), 195\u0026ndash;201. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/00002030-199902040-00007\u003c/span\u003e\u003cspan address=\"10.1097/00002030-199902040-00007\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (1999).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRiley, R. D. et al. Calculating the sample size required for developing a clinical prediction model. \u003cem\u003eBMJ\u003c/em\u003e \u003cb\u003e368\u003c/b\u003e, m441. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmj.m441\u003c/span\u003e\u003cspan address=\"10.1136/bmj.m441\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020). Published 2020 Mar 18.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHu, C. et al. Interpretable Machine Learning for Early Prediction of Prognosis in Sepsis: A Discovery and Validation Study. \u003cem\u003eInfect. Dis. Ther.\u003c/em\u003e \u003cb\u003e11\u003c/b\u003e (3), 1117\u0026ndash;1132. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s40121-022-00628-6\u003c/span\u003e\u003cspan address=\"10.1007/s40121-022-00628-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eStekhoven, D. J. \u0026amp; B\u0026uuml;hlmann, P. MissForest\u0026ndash;non-parametric missing value imputation for mixed-type data. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cb\u003e28\u003c/b\u003e (1), 112\u0026ndash;118. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/bioinformatics/btr597\u003c/span\u003e\u003cspan address=\"10.1093/bioinformatics/btr597\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2012).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYou, J. et al. Comparison of imputation methods for missing production data of dairy cattle. \u003cem\u003eAnimal\u003c/em\u003e \u003cb\u003e17\u003c/b\u003e (Suppl 5), 100921. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.animal.2023.100921\u003c/span\u003e\u003cspan address=\"10.1016/j.animal.2023.100921\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFu, Y., Zhao, J. \u0026amp; Wang, Y. LASSO regression and Boruta algorithm to explore the relationship between neutrophil percentage to albumin ratio and asthma: results from the NHANES 2001 to 2018. Clin Exp Med. ;25(1):149. Published 2025 May 10. (2025). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s10238-025-01701-3\u003c/span\u003e\u003cspan address=\"10.1007/s10238-025-01701-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMusa, A. B. Comparative study on classification performance between support vector machine and logistic regression. \u003cem\u003eInt. J. Mach. Learn. Cyber\u003c/em\u003e. \u003cb\u003e4\u003c/b\u003e, 13\u0026ndash;24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s13042-012-0068-x\u003c/span\u003e\u003cspan address=\"10.1007/s13042-012-0068-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2013).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eValkenborg, D., Rousseau, A. J., Geubbelmans, M. \u0026amp; Burzykowski, T. Support vector machines. \u003cem\u003eAm. J. Orthod. Dentofac. Orthop.\u003c/em\u003e \u003cb\u003e164\u003c/b\u003e (5), 754\u0026ndash;757. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.ajodo.2023.08.003\u003c/span\u003e\u003cspan address=\"10.1016/j.ajodo.2023.08.003\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXie, Y. et al. JOURNAL CLUB: Use of Gradient Boosting Machine Learning to Predict Patient Outcome in Acute Ischemic Stroke on the Basis of Imaging, Demographic, and Clinical Information. \u003cem\u003eAJR Am. J. Roentgenol.\u003c/em\u003e \u003cb\u003e212\u003c/b\u003e (1), 44\u0026ndash;51. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2214/AJR.18.20260\u003c/span\u003e\u003cspan address=\"10.2214/AJR.18.20260\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRajaguru, H., S \u0026amp; R SC. Analysis of Decision Tree and K-Nearest Neighbor Algorithm in the Classification of Breast Cancer. \u003cem\u003eAsian Pac. J. Cancer Prev.\u003c/em\u003e \u003cb\u003e20\u003c/b\u003e (12), 3777\u0026ndash;3781. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.31557/APJCP.2019.20.12.3777\u003c/span\u003e\u003cspan address=\"10.31557/APJCP.2019.20.12.3777\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019). Published 2019 Dec 1.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang, X. W. \u0026amp; Liu, Y. Y. Comparative study of classifiers for human microbiome data. \u003cem\u003eMed. Microecol\u003c/em\u003e. \u003cb\u003e4\u003c/b\u003e, 100013. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.medmic.2020.100013\u003c/span\u003e\u003cspan address=\"10.1016/j.medmic.2020.100013\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXiao, Y. Application of Neural Network Algorithm in Medical Artificial Intelligence Product Development. \u003cem\u003eComput. Math. Methods Med.\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1155/2022/5413202\u003c/span\u003e\u003cspan address=\"10.1155/2022/5413202\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022). 2022:5413202. Published 2022 Jun 8.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang, K. et al. Interpretable prediction of 3-year all-cause mortality in patients with heart failure caused by coronary heart disease based on machine learning and SHAP. \u003cem\u003eComput. Biol. Med.\u003c/em\u003e \u003cb\u003e137\u003c/b\u003e, 104813. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.compbiomed.2021.104813\u003c/span\u003e\u003cspan address=\"10.1016/j.compbiomed.2021.104813\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXu, L., Raitoharju, J., Iosifidis, A. \u0026amp; Gabbouj, M. Saliency-Based Multilabel Linear Discriminant Analysis. \u003cem\u003eIEEE Trans. Cybern\u003c/em\u003e. \u003cb\u003e52\u003c/b\u003e (10), 10200\u0026ndash;10213. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1109/TCYB.2021.3069338\u003c/span\u003e\u003cspan address=\"10.1109/TCYB.2021.3069338\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDeLong, E. R., DeLong, D. M. \u0026amp; Clarke-Pearson, D. L. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. \u003cem\u003eBiometrics\u003c/em\u003e \u003cb\u003e44\u003c/b\u003e (3), 837\u0026ndash;845 (1988).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLundberg, S. M. \u0026amp; Lee, S. I. A unified approach to interpreting model predictions. NIPS'17: Proceedings of the 31st International Conference on Neural Information Processing Systems; NIPS'17; Dec 4\u0026ndash;9; Long Beach, CA. pp. 4768\u0026ndash;4777. (2017).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSettle, E. \u003cem\u003eAIDS in China: An annotated chronology 1985\u0026ndash;2003\u003c/em\u003e (China AIDS Survey, Chinese Center for Disease Control and Prevention, 2003).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWu, Z., Wang, Y., Detels, R. \u0026amp; Rotheram-Borus, M. J. China AIDS policy implementation: reversing the HIV/AIDS epidemic by 2015. \u003cem\u003eInt. J. Epidemiol.\u003c/em\u003e \u003cb\u003e39\u003c/b\u003e (Suppl 2), ii1\u0026ndash;ii3. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/ije/dyq220\u003c/span\u003e\u003cspan address=\"10.1093/ije/dyq220\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2010).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSmith, M. K. et al. The detection and management of early HIV infection: a clinical and public health emergency. \u003cem\u003eJ. Acquir. Immune Defic. Syndr.\u003c/em\u003e \u003cb\u003e63\u003c/b\u003e (0 2), S187\u0026ndash;S199. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/QAI.0b013e31829871e0\u003c/span\u003e\u003cspan address=\"10.1097/QAI.0b013e31829871e0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2013).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMugavero, M. J., Castellano, C., Edelman, D. \u0026amp; Hicks, C. Late diagnosis of HIV infection: the role of age and sex. \u003cem\u003eAm. J. Med.\u003c/em\u003e \u003cb\u003e120\u003c/b\u003e (4), 370\u0026ndash;373. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.amjmed.2006.05.050\u003c/span\u003e\u003cspan address=\"10.1016/j.amjmed.2006.05.050\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2007).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMayer, K. H. \u0026amp; Venkatesh, K. K. Antiretroviral therapy as HIV prevention: status and prospects. \u003cem\u003eAm. J. Public. Health\u003c/em\u003e. \u003cb\u003e100\u003c/b\u003e (10), 1867\u0026ndash;1876. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2105/AJPH.2009.184796\u003c/span\u003e\u003cspan address=\"10.2105/AJPH.2009.184796\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2010).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMcDonald, L., Ramagopalan, S. V., Cox, A. P. \u0026amp; Oguz, M. Unintended consequences of machine learning in medicine? \u003cem\u003eF1000Res\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.12688/f1000research.12693.1\u003c/span\u003e\u003cspan address=\"10.12688/f1000research.12693.1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017). 6:1707. Published 2017 Sep 19.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAdal, M., Howe, R., Kassa, D., Aseffa, A. \u0026amp; Petros, B. Associations of gender and serum total cholesterol with CD4\u0026thinsp;+\u0026thinsp;T cell count and HIV RNA load in antiretroviral-na\u0026iuml;ve individuals in Addis Ababa. BMC Public Health. ;18(1):943. Published 2018 Jul 31. (2018). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12889-018-5852-4\u003c/span\u003e\u003cspan address=\"10.1186/s12889-018-5852-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLorizate, M. \u0026amp; Kr\u0026auml;usslich, H. G. Role of lipids in virus replication. Cold Spring Harb Perspect Biol. ;3(10):a004820. Published 2011 Oct 1. (2011). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1101/cshperspect.a004820\u003c/span\u003e\u003cspan address=\"10.1101/cshperspect.a004820\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMazzon, M. \u0026amp; Mercer, J. Lipid interactions during virus entry and infection. \u003cem\u003eCell. Microbiol.\u003c/em\u003e \u003cb\u003e16\u003c/b\u003e (10), 1493\u0026ndash;1502. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/cmi.12340\u003c/span\u003e\u003cspan address=\"10.1111/cmi.12340\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2014).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNegredo, E. et al. Association between HIV replication and cholesterol in peripheral blood mononuclear cells in HIV-infected patients interrupting HAART. \u003cem\u003eJ. Antimicrob. Chemother.\u003c/em\u003e \u003cb\u003e61\u003c/b\u003e (2), 400\u0026ndash;404. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/jac/dkm470\u003c/span\u003e\u003cspan address=\"10.1093/jac/dkm470\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2008).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHannedouche, S. et al. Oxysterols direct immune cell migration via EBI2. \u003cem\u003eNature\u003c/em\u003e \u003cb\u003e475\u003c/b\u003e (7357), 524\u0026ndash;527. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nature10280\u003c/span\u003e\u003cspan address=\"10.1038/nature10280\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2011). Published 2011 Jul 27.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eM\u0026iacute;guez MJ, Lewis JE, Bryant VE, et al. Low cholesterol? Don't brag yet \u0026hellip; hypocholesterolemia blunts HAART effectiveness: a longitudinal study. J Int AIDS Soc. 2010;13:25. Published 2010 Jul 13. doi:10.1186/1758-2652-13-25.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBelperio, P. S. \u0026amp; Rhew, D. C. Prevalence and outcomes of anemia in individuals with human immunodeficiency virus: a systematic review of the literature. \u003cem\u003eAm. J. Med.\u003c/em\u003e \u003cb\u003e116\u003c/b\u003e (Suppl 7A), 27S\u0026ndash;43S. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.amjmed.2003.12.010\u003c/span\u003e\u003cspan address=\"10.1016/j.amjmed.2003.12.010\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2004).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePaton, N. I., Sangeetha, S., Earnest, A. \u0026amp; Bellamy, R. The impact of malnutrition on survival and the CD4 count response in HIV-infected patients starting antiretroviral therapy. \u003cem\u003eHIV Med.\u003c/em\u003e \u003cb\u003e7\u003c/b\u003e (5), 323\u0026ndash;330. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/j.1468-1293.2006.00383.x\u003c/span\u003e\u003cspan address=\"10.1111/j.1468-1293.2006.00383.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2006).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang, D. et al. Plasma hemoglobin and the risk of death in HIV/AIDS patients treated with antiretroviral therapy. \u003cem\u003eAging (Albany NY)\u003c/em\u003e. \u003cb\u003e13\u003c/b\u003e (9), 13061\u0026ndash;13072. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.18632/aging.202987\u003c/span\u003e\u003cspan address=\"10.18632/aging.202987\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCostello, C. et al. HIV-1 subtype E progression among northern Thai couples: traditional and non-traditional predictors of survival. \u003cem\u003eInt. J. Epidemiol.\u003c/em\u003e \u003cb\u003e34\u003c/b\u003e (3), 577\u0026ndash;584. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/ije/dyi023\u003c/span\u003e\u003cspan address=\"10.1093/ije/dyi023\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2005).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWHO. \u003cem\u003eGuideline on when to start antiretroviral therapy and on pre-exposure prophylaxis for HIV\u003c/em\u003e (Geneva, 2015).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCollini, P. et al. Sustained immunological responses to highly active antiretroviral therapy at 36 months in a Ghanaian HIV cohort. \u003cem\u003eClin. Infect. Dis.\u003c/em\u003e \u003cb\u003e48\u003c/b\u003e (7), 988\u0026ndash;991. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1086/597353\u003c/span\u003e\u003cspan address=\"10.1086/597353\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2009).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"HIV/AIDS, hematological markers, machine learning, SHAP","lastPublishedDoi":"10.21203/rs.3.rs-8226880/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8226880/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e\u003cp\u003eA large number of people living with HIV are not diagnosed until the advanced stage, and they face a high risk of death even after initiating antiretroviral therapy (ART). This study aimed to identify high-risk patients for advanced HIV disease using easily accessible hematological markers, explore effective predictive indicators in resource-limited settings, and provide a basis for early clinical intervention.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eData were collected from HIV/AIDS patients receiving ART in Linhai, Zhejiang Province, China, from 2010 to June 2025. Patients were classified into advanced infection (CD4\u0026thinsp;+\u0026thinsp;T cell count\u0026thinsp;\u0026lt;\u0026thinsp;200 cells/mm\u0026sup3;) and non-advanced infection according to WHO criteria. Feature selection was performed using Lasso regression combined with the Boruta algorithm, and eight machine learning models were developed. Model performance was evaluated by discrimination, calibration, and clinical practicability. The optimal model was subjected to SHAP (SHapley Additive exPlanations) analysis to assess variable importance.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eA total of 709 patients were included. Among them, 260 individuals (accounting for 36.6%) had progressed to the advanced stage of HIV disease before starting ART.Seven variables were selected to construct the machine learning models. The ENET model demonstrated the highest AUC (0.801) in the validation set, along with satisfactory calibration and clinical utility. SHAP analysis revealed that CD8\u0026thinsp;+\u0026thinsp;T cells had the highest average SHAP value, contributing the most to model prediction. Among the easily accessible hematological markers, total cholesterol had the greatest contribution.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e\u003cp\u003eThe ENET model exhibited optimal performance for predicting advanced HIV disease, serving as an effective tool for identifying high-risk patients. CD8\u0026thinsp;+\u0026thinsp;T cells are the core immune indicator for predicting disease progression, while total cholesterol is the most influential among easily accessible hematological markers. Combining these markers with others such as hemoglobin provides a convenient and reliable approach for assessing HIV disease progression risk in resource-limited settings.\u003c/p\u003e","manuscriptTitle":"Explainable Machine Learning Predicts Advanced HIV Disease Progression Using Easily Accessible Hematological Markers","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-08 11:16:34","doi":"10.21203/rs.3.rs-8226880/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"eab096c9-3d56-446c-a7b1-88cd308ea7f8","owner":[],"postedDate":"December 8th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":59085516,"name":"Health sciences/Biomarkers"},{"id":59085517,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":59085518,"name":"Health sciences/Diseases"},{"id":59085519,"name":"Health sciences/Medical research"}],"tags":[],"updatedAt":"2026-01-12T09:54:59+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-08 11:16:34","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8226880","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8226880","identity":"rs-8226880","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}