Preoperative Nutrition–Inflammation Status and Surgical Factors Predict Delayed Extubation After Oral Cancer Surgery with Free Flap Reconstruction: A Machine Learning Approach

Preoperative Nutrition–Inflammation Status and Surgical Factors Predict Delayed Extubation After Oral Cancer Surgery with Free Flap Reconstruction: A Machine Learning Approach | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Preoperative Nutrition–Inflammation Status and Surgical Factors Predict Delayed Extubation After Oral Cancer Surgery with Free Flap Reconstruction: A Machine Learning Approach Baolin Jia, Chuan Ye, Huan Zhang, Jun Ren, Guixin Li, XianJie Zheng, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8243006/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 Delayed extubation (DE) is a common postoperative challenge after oral cancer resection with free flap reconstruction. Existing risk assessments largely focus on anatomical and surgical factors, with limited consideration of patients’ systemic physiological status. The Advanced Lung Cancer Inflammation Index (ALI), a preoperative composite measure of nutritional and inflammatory status, has shown prognostic value in oncology but its role in perioperative airway management is unclear. Methods We retrospectively analyzed 752 patients undergoing oral cancer resection with free flap reconstruction at a single center. Associations between preoperative ALI and DE were evaluated using multivariable logistic regression. A random forest model integrating ALI with clinical and surgical factors was developed and interpreted using SHapley Additive exPlanations (SHAP) to identify key predictors. Results DE occurred in 32.2% of patients. Higher preoperative ALI was independently associated with lower risk of DE (adjusted OR per 10-point increase = 0.90, 95% CI: 0.84–0.95, p < 0.001). The random forest model achieved an AUC of 0.875 in the validation cohort and demonstrated good calibration. SHAP analysis revealed tumor T stage, extent of resection, bilateral neck dissection, and ALI as the most influential predictors, with higher ALI consistently protective. Conclusions Preoperative ALI is an independent predictor of delayed extubation. An interpretable machine learning model combining ALI with clinical and surgical variables provides a high-performing tool for individualized perioperative airway risk assessment, supporting tailored extubation strategies and postoperative management in oral cancer patients. Advanced Lung Cancer Inflammation Index Delayed Extubation Oral Cancer Free Flap Reconstruction Machine Learning Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1. Introduction Extensive resection of oral malignancies combined with free flap reconstruction remains a fundamental component of contemporary head and neck cancer treatment( 1 ). Although ongoing refinements in microsurgical techniques have improved both tumor control and postoperative functional outcomes, perioperative airway management continues to be a critical concern( 2 ). Altered upper airway anatomy, substantial postoperative edema, and the risk of hematoma formation may rapidly compromise airway patency in the immediate postoperative period. These challenges have been linked to higher complication rates, prolonged ICU stays, and increased postoperative morbidity and mortality( 3 , 4 ). Currently, three primary strategies are commonly employed in clinical practice: immediate extubation in the operating room, delayed extubation (DE) in the ICU, or prophylactic tracheostomy( 5 ). Considerable variation exists in the selection of these approaches across institutions and among individual surgeons, reflecting the lack of robust, objective, and generalizable tools for perioperative risk assessment( 6 ). Existing clinical scoring systems, such as the widely cited Kruse-Lösler score, have demonstrated limited predictive performance, with positive predictive values ranging from 0.08 to 0.18 in validation studies( 7 , 8 ). As a result, conservative strategies may lead to unnecessary tracheostomies, whereas overly aggressive extubation could precipitate critical airway compromise( 9 ), underscoring the ongoing challenge of balancing safety and efficiency in postoperative airway management. Previous studies have predominantly focused on anatomical or surgical factors, including tumor T stage, extent of mandibular resection, bilateral neck dissection, and flap volume( 10 , 11 ). While these variables undoubtedly influence the degree of airway compromise, traditional statistical models may not fully capture a patient’s overall physiological vulnerability, particularly with respect to systemic factors such as inflammation and nutritional status. Evidence suggests that elevated perioperative inflammation is associated with increased tissue edema, and inflammatory indices such as the neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) have been linked not only to flap complications( 12 ) but also to delayed extubation and pulmonary events in other surgical populations( 13 ). Similarly, malnutrition, reflected by conditions such as hypoalbuminemia, has been recognized as an independent risk factor for multiple postoperative pulmonary complications after major surgery( 14 ). Despite these insights, there remains a paucity of research in head and neck surgery that integrates inflammatory and nutritional assessment to inform perioperative airway management strategies. The Advanced Lung Cancer Inflammation Index (ALI) has been widely validated as an integrative measure of patients’ inflammatory burden and nutritional reserves, combining BMI, serum albumin, and NLR, and serving as a prognostic marker across multiple malignancies( 15 ). A lower ALI theoretically reflects heightened systemic inflammation and poorer nutritional status, which may contribute to prolonged postoperative edema and delayed recovery, thereby increasing the likelihood of delayed extubation. However, the application of ALI in postoperative airway management for oral cancer remains largely unexplored. To our knowledge, no prior studies have systematically combined ALI with detailed clinical and surgical factors to assess delayed extubation risk, representing a notable knowledge gap that the present study aims to address. Meanwhile, machine learning (ML) approaches are increasingly being explored for perioperative risk assessment, offering the ability to handle high-dimensional clinical data with complex interactions and nonlinear relationships. In the context of head and neck surgery, ML models have been applied to predict outcomes such as flap necrosis and overall survival, often demonstrating improved predictive performance compared with conventional regression techniques( 16 – 18 ). Nevertheless, to date, no studies have integrated ML with systemic inflammation–nutrition indices to predict delayed extubation following free flap reconstruction in oral cancer patients. Moreover, the use of interpretable ML methods remains limited, leaving a gap in clinically actionable insights that could support individualized airway management decisions. To address current gaps in the literature, this study focuses on three interrelated objectives. We first evaluate the independent association between the inflammation–nutrition composite index ALI and the risk of delayed extubation following oral cancer surgery with free flap reconstruction. We then develop and validate a random forest–based prediction model to improve perioperative risk stratification, incorporating both clinical and surgical variables. Finally, we apply SHAP analysis to enhance the interpretability of the model, providing clinicians with a clearer understanding of the factors driving individual predictions. Collectively, these efforts aim to offer data-driven guidance for postoperative airway management and support the optimization of individualized extubation strategies. 2. Methods 2.1. Data Source and Study Population This retrospective, single-center observational study included patients who underwent oral cancer surgery with free flap reconstruction at the Department of Oral and Maxillofacial Surgery, Suining Central Hospital, China, between August 2017 and August 2025. The study protocol was approved by the Biomedical Research Ethics Committee of Suining Central Hospital (Approval No. KYLLKYS20250144). The requirement for informed consent was waived due to the retrospective nature of the study. All procedures followed the principles of the Declaration of Helsinki, and all patient information was anonymized before analysis. 2.2. Inclusion and Exclusion Criteria Patients were considered eligible if they had a histopathologically confirmed diagnosis of primary oral squamous cell carcinoma and underwent tumor resection followed by immediate free flap reconstruction. We excluded patients who received a prophylactic tracheostomy at the time of surgery, those who required reintubation or subsequent tracheostomy after an initially successful extubation, and individuals with incomplete or missing essential perioperative information. The patient selection process is summarized in Fig. 1 . 2.3. Data Collection and Variable Definitions Clinical data were retrospectively collected from the electronic medical record system using a structured template. The extracted variables encompassed demographic characteristics, comorbidities, tumor- and surgery-related profiles, and intraoperative metrics. Preoperative laboratory values were recorded for all patients, and the ALI was calculated as (body mass index × serum albumin) / neutrophil-to-lymphocyte ratio( 15 ). To ensure data integrity, two trained clinicians independently performed the data extraction. Any discrepancies were resolved through consensus or, when necessary, adjudication by a senior investigator. The primary outcome was delayed extubation, defined as the planned transfer of the patient to the intensive care unit with the endotracheal tube in place for ongoing ventilatory support, rather than proceeding with immediate extubation in the operating room. This endpoint aligns with established definitions in studies of airway management following major head and neck reconstruction( 3 – 5 ). Patients requiring reintubation after a successful initial extubation were excluded from the primary analysis, as this event typically stems from distinct postoperative complications. 2.4 Statistical Analysis 2.4.1 Baseline analysis and correlation analysis The final analysis included 752 patients after applying the predefined exclusion criteria. We summarized baseline characteristics for the entire cohort, comparing them between patients with and without delayed extubation. Categorical variables are reported as counts (percentages), compared using the χ² or Fisher's exact test, while continuous variables, presented as medians (IQRs), were compared with the Mann-Whitney U test. We assessed the association between ALI and delayed extubation using logistic regression in three steps: an unadjusted model, a model adjusted for demographics and lifestyle, and a fully adjusted model that also included comorbidities, tumor stage, and surgical extent. ALI was analyzed as a continuous measure (per 10-point increase) and by tertiles. We explored the dose-response relationship using restricted cubic splines and conducted pre-specified subgroup analyses to test the consistency of the association. 2.4.2 Model development and validation The cohort was randomly split into a training set (70%, n = 527) and a testing set (30%, n = 225) to support model development and internal validation. Feature selection was carried out within the training set using Least Absolute Shrinkage and Selection Operator (LASSO) regression with 10-fold cross-validation, which yielded twelve predictors with non-zero coefficients for subsequent analysis. These predictors were then used to develop seven machine-learning models: logistic regression, decision tree, Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), support vector machine (SVM), and an artificial neural network (ANN). Each model underwent hyperparameter tuning through 10-fold cross-validation within the training set, with parameter grids tailored to the characteristics of each algorithm. The full list of candidate hyperparameters and the final selected values is provided in Supplementary Table S1 . All tuned models were evaluated on the independent testing set. Model discrimination was assessed by the area under the receiver operating characteristic curve (AUC), and calibration was examined using calibration plots and the Brier score. Decision curve analysis (DCA) was performed to assess clinical utility. Additional performance metrics—including accuracy, sensitivity, specificity, and the F1 score—were also calculated. The final model was selected based on a comprehensive assessment of these performance measures, prioritizing discriminative ability and calibration quality. 2.4.3 Model interpretability To improve the clinical interpretability of the final model and mitigate its black-box characteristics, we applied the SHapley Additive exPlanations (SHAP) framework in a post hoc analysis to quantify the contribution of individual predictors. SHAP is grounded in cooperative game theory and provides a unified measure of feature influence on model output( 19 ). The analysis was structured to yield both global and local insights. At the global level, mean absolute SHAP values were used to rank the relative importance of each predictor, and summary plots were generated to illustrate the distribution and directional impact of feature contributions across the cohort. For continuous predictors, dependence plots were constructed to characterize their marginal relationship with the model’s predicted risk. For patient-level interpretation, SHAP values were decomposed for individual predictions to highlight the specific variables driving risk estimates. These contributions were visualized using waterfall or force plots to facilitate case-based reasoning and potential bedside applicability. All statistical analyses and model development were performed in R (version 4.4.3). SHAP computations and visualization were conducted in Python (version 3.12). A two-tailed p value < 0.05 was considered statistically significant. The overall study workflow is presented in Fig. 1 . 3. Results 3.1 Baseline Characteristics of the Oral Cancer Surgery with Free Flap Reconstruction Cohort This study included 752 patients undergoing oral cancer surgery with free flap reconstruction. The incidence of delayed extubation was 32.2% (242/752). Baseline characteristics stratified by extubation outcome are detailed in Table 1 . Table 1 Baseline characteristics stratified by extubation outcome Variables Total (n = 752) Non-delayed extubation (n = 510) Delayed extubation (n = 242) P Age, M (Q₁, Q₃) 56.00 (48.00, 65.00) 55.00 (46.00, 64.00) 60.00 (51.00, 67.00) < 0.001 BMI, M (Q₁, Q₃) 27.70 (24.30, 32.10) 27.40 (24.20, 31.60) 28.30 (24.60, 32.88) 0.040 Surgery Time, M (Q₁, Q₃) 418.50 (365.00, 481.00) 418.00 (366.00, 482.00) 420.50 (364.00, 478.25) 0.844 Blood Loss, M (Q₁, Q₃) 348.00 (281.00, 426.25) 347.00 (282.25, 427.00) 350.00 (280.00, 423.00) 0.852 BP fluctuation, M (Q₁, Q₃) 12.50 (9.00, 16.00) 12.00 (9.00, 16.00) 13.00 (9.00, 17.00) 0.307 HR fluctuation, M (Q₁, Q₃) 9.00 (6.00, 12.00) 9.00 (6.00, 12.00) 9.00 (6.00, 11.75) 0.290 Hypotension events, M (Q₁, Q₃) 2.00 (1.00, 4.00) 2.00 (1.00, 4.00) 2.00 (1.00, 4.00) 0.703 WBC, M (Q₁, Q₃) 6.40 (5.30, 7.60) 6.10 (5.20, 7.20) 6.90 (5.53, 8.60) < 0.001 Lymphocyte, M (Q₁, Q₃) 1.80 (1.40, 2.20) 1.80 (1.40, 2.20) 1.75 (1.40, 2.20) 0.420 Neutrophils, M (Q₁, Q₃) 3.75 (2.90, 4.70) 3.60 (2.80, 4.40) 4.15 (3.20, 5.40) < 0.001 RBC, M (Q₁, Q₃) 4.66 (4.34, 4.97) 4.64 (4.32, 4.97) 4.70 (4.40, 4.98) 0.356 Hemoglobin, M (Q₁, Q₃) 14.10 (13.10, 15.10) 14.10 (13.10, 15.00) 14.20 (13.20, 15.10) 0.618 RDW, M (Q₁, Q₃) 13.50 (13.10, 14.10) 13.40 (13.00, 13.90) 13.80 (13.30, 14.40) < 0.001 Platelet, M (Q₁, Q₃) 225.00 (189.00, 270.25) 228.00 (191.00, 267.00) 219.00 (183.25, 273.00) 0.443 MPV, M (Q₁, Q₃) 8.40 (7.80, 9.00) 8.40 (7.80, 8.90) 8.40 (7.90, 9.10) 0.235 ALT, M (Q₁, Q₃) 21.00 (16.00, 28.00) 22.00 (17.00, 29.00) 21.00 (16.00, 26.00) 0.042 AST, M (Q₁, Q₃) 23.00 (19.00, 27.00) 23.00 (20.00, 27.00) 22.00 (19.00, 27.00) 0.051 Bilirubin Total, M (Q₁, Q₃) 10.26 (8.55, 13.68) 11.97 (8.55, 13.68) 10.26 (8.55, 13.68) 0.004 Albumin, M (Q₁, Q₃) 42.00 (39.00, 45.00) 43.00 (41.00, 46.00) 39.00 (38.00, 41.00) < 0.001 GGT, M (Q₁, Q₃) 20.00 (15.00, 31.00) 19.00 (14.00, 30.00) 22.00 (16.00, 34.00) 0.003 Creatinine, M (Q₁, Q₃) 74.26 (62.76, 89.28) 73.37 (61.88, 87.52) 76.02 (63.65, 90.17) 0.177 Uric Acid, M (Q₁, Q₃) 321.20 (267.70, 380.70) 315.20 (261.70, 368.80) 333.10 (273.60, 398.50) 0.004 Blood Urea Nitrogen, M (Q₁, Q₃) 4.64 (3.57, 5.71) 4.64 (3.57, 5.71) 4.64 (3.57, 6.07) 0.591 Sodium, M (Q₁, Q₃) 140.00 (139.00, 141.00) 140.00 (139.00, 141.00) 140.00 (138.00, 141.00) 0.377 Phosphorus, M (Q₁, Q₃) 1.20 (1.10, 1.32) 1.20 (1.10, 1.32) 1.20 (1.07, 1.32) 0.740 Calcium Total, M (Q₁, Q₃) 2.35 (2.30, 2.40) 2.35 (2.30, 2.40) 2.35 (2.30, 2.40) 0.590 Triglycerides, M (Q₁, Q₃) 1.16 (0.79, 1.73) 1.15 (0.78, 1.72) 1.20 (0.86, 1.76) 0.092 Hdl Cholesterol, M (Q₁, Q₃) 1.37 (1.14, 1.66) 1.39 (1.15, 1.71) 1.29 (1.11, 1.60) 0.006 Ldl Cholesterol, M (Q₁, Q₃) 2.95 (2.35, 3.54) 2.95 (2.38, 3.57) 2.95 (2.33, 3.50) 0.527 ALI, M (Q₁, Q₃) 57.80 (40.65, 74.97) 60.30 (43.47, 80.29) 51.95 (34.62, 67.73) < 0.001 Sex, n(%) 0.419 Male 430 (57.18) 286 (56.08) 144 (59.50) Female 322 (42.82) 224 (43.92) 98 (40.50) Smoking status, n(%) < 0.001 No 423 (56.25) 317 (62.16) 106 (43.80) Yes 329 (43.75) 193 (37.84) 136 (56.20) Alcohol intake, n(%) 0.603 No 536 (71.28) 360 (70.59) 176 (72.73) Yes 216 (28.72) 150 (29.41) 66 (27.27) Hypertension, n(%) < 0.001 No 373 (49.60) 276 (54.12) 97 (40.08) Yes 379 (50.40) 234 (45.88) 145 (59.92) DM, n(%) 0.002 No 581 (77.26) 411 (80.59) 170 (70.25) Yes 171 (22.74) 99 (19.41) 72 (29.75) Hyperlipidemia, n(%) 0.274 No 185 (24.60) 132 (25.88) 53 (21.90) Yes 567 (75.40) 378 (74.12) 189 (78.10) COPD, n(%) 0.166 No 721 (95.88) 493 (96.67) 228 (94.21) Yes 31 (4.12) 17 (3.33) 14 (5.79) CVD, n(%) 0.010 No 673 (89.49) 467 (91.57) 206 (85.12) Yes 79 (10.51) 43 (8.43) 36 (14.88) ASA, n(%) 0.001 I 110 (14.63) 86 (16.86) 24 (9.92) II 242 (32.18) 176 (34.51) 66 (27.27) III 400 (53.19) 248 (48.63) 152 (62.81) Tumor T stage, n(%) < 0.001 T1/T2 414 (55.05) 309 (60.59) 105 (43.39) T3/T4 338 (44.95) 201 (39.41) 137 (56.61) Extensive maxillomandibular resection, n(%) < 0.001 No 545 (72.47) 404 (79.22) 141 (58.26) Yes 207 (27.53) 106 (20.78) 101 (41.74) Bilateral neck dissection, n(%) < 0.001 No 597 (79.39) 442 (86.67) 155 (64.05) Yes 155 (20.61) 68 (13.33) 87 (35.95) Flap size, n(%) < 0.001 Small flap 317 (42.15) 243 (47.65) 74 (30.58) Medium flap 329 (43.75) 211 (41.37) 118 (48.76) Large flap 106 (14.10) 56 (10.98) 50 (20.66) Transfusion, n(%) 0.352 No 519 (69.02) 358 (70.20) 161 (66.53) Yes 233 (30.98) 152 (29.80) 81 (33.47) Tumor site, n(%) < 0.001 Tongue 195 (25.93) 142 (27.84) 53 (21.90) Floor of mouth 113 (15.03) 70 (13.73) 43 (17.77) Buccal mucosa 140 (18.62) 111 (21.76) 29 (11.98) Gingiva 89 (11.84) 65 (12.75) 24 (9.92) Palate 117 (15.56) 67 (13.14) 50 (20.66) Other sites 98 (13.03) 55 (10.78) 43 (17.77) Data are presented as n (%) or median [interquartile range]. P were derived from the Chi-squared test for categorical variables and the Mann-Whitney U test for continuous variables. Univariate comparisons demonstrated that differences in delayed extubation were predominantly driven by pre-operative and oncologic factors rather than intraoperative events. Specifically, there were no significant differences between the two groups in surgery duration, blood loss, hemodynamic fluctuations, or transfusion rate (all p > 0.05). In contrast, significant pre-operative disparities were identified. Patients in the delayed extubation group were older (Median: 60.0 vs. 55.0 years, p < 0.001) and had a higher comorbidity burden, including hypertension (59.9% vs. 45.9%, p < 0.001), diabetes mellitus (29.8% vs. 19.4%, p = 0.002), cardiovascular disease (14.9% vs. 8.4%, p = 0.010), and a higher ASA classification (62.8% vs. 48.6% ASA III, p = 0.001). A history of smoking was also more prevalent in this group (56.2% vs. 37.8%, p < 0.001). The surgical profile indicated more extensive resections in the delayed extubation group. These patients had higher T-stage tumors (T3/T4: 56.6% vs. 39.4%, p < 0.001) and more frequently required mandibular resection (41.7% vs. 20.8%, p < 0.001) and bilateral neck dissection (36.0% vs. 13.3%, p < 0.001). Accordingly, the need for larger flaps was more common (20.7% vs. 11.0%, p < 0.001). The pre-operative laboratory profile was markedly different. The delayed extubation group exhibited signs of systemic inflammation and poorer nutritional status, with significantly higher white blood cell count, neutrophil count, and red cell distribution width, alongside lower albumin, and HDL cholesterol (all p < 0.01). Crucially, the ALI was significantly lower in patients facing delayed extubation (51.95 vs. 60.30, p < 0.001). 3.2. The Association between ALI and Delayed Extubation Risk The association between ALI and delayed extubation was assessed through multivariable logistic regression (Table 2 ). After full adjustment for demographic, comorbidity, surgical, and oncologic confounders (Model 3), ALI remained an independent predictor of delayed extubation. Each 10-point increase in ALI was associated with a 10% reduction in odds (Adjusted OR = 0.90, 95% CI: 0.84–0.95, p < .001). When analyzed by tertiles, a significant inverse trend was observed ( p for trend < 0.001), with patients in the highest tertile (Q3) having a significantly lower risk compared to the lowest tertile (Q1) (Adjusted OR = 0.62, 95% CI: 0.40–0.97, p = 0.035). Table 2 Logistic regression analysis of the association between ALI and delayed extubation Variables Model 1 Model 2 Model 3 OR (95%CI) P OR (95%CI) P OR (95%CI) P ALI (per 10-point increase) 0.87 (0.82 ~ 0.92) < .001 0.87 (0.82 ~ 0.92) < .001 0.90 (0.84 ~ 0.95) < .001 ALI (tertile) Q1 1.00 (Reference) 1.00 (Reference) 1.00 (Reference) Q2 0.76 (0.53 ~ 1.09) 0.141 0.77 (0.53 ~ 1.13) 0.180 0.83 (0.55 ~ 1.26) 0.388 Q3 0.48 (0.32 ~ 0.70) < .001 0.50 (0.33 ~ 0.75) < .001 0.62 (0.40 ~ 0.97) 0.035 P for trend < 0.001 < 0.001 < 0.001 OR: Odds Ratio, CI: Confidence Interval Model 1: Crude; Model 2: Adjust: Sex, Smoking status, Alcohol intake, Hypertension, DM, Hyperlipidemia, COPD, CVD, Age; Model 3: Adjust: Sex, Smoking status, Alcohol intake, Hypertension, DM, Hyperlipidemia, COPD, CVD, ASA, Tumor T stage, Extensive maxillomandibular resection, Bilateral neck dissection, Flap size, Age, Tumor site. To elucidate the dose-response relationship, a restricted cubic spline (RCS) analysis was performed with full adjustment for the same covariates. As shown in Fig. 2 , the analysis confirmed a linear and inverse association between ALI and the log odds of delayed extubation ( P for nonlinearity = 0.277). The overall association was highly significant ( P for overall < 0.001), demonstrating a steady decrease in risk across the entire observed range of ALI values. Subgroup analyses were conducted to evaluate the consistency of this association across various patient characteristics (Fig. 3 ). Subgroup analyses showed no evidence of heterogeneity for most covariates. The protective effect of ALI was remarkably consistent in direction across most subgroups. Significant interactions were observed for several factors. The inverse association was significantly stronger among non-smokers (aOR = 0.78, 95% CI: 0.69–0.88) compared to smokers (aOR = 0.96, 95% CI: 0.90–1.03; P for interaction = 0.014), and in patients aged ≥ 60 years (aOR = 0.81, 95% CI: 0.72–0.92) compared to those aged < 60 years (aOR = 0.94, 95% CI: 0.87–1.02; P for interaction = 0.019). A significant interaction was also noted with COPD status ( P for interaction = 0.007) and flap size ( P for interaction = 0.026). No significant interactions were found for sex, alcohol intake, or other comorbidities and surgical factors. 3.3. Development and Performance of Machine Learning Models 3.3.1. Data Partition and Balance Assessment To develop and validate the machine learning models, the study cohort was randomly divided into a training set (n = 527) for model development and a testing set (n = 225) for independent validation. The balance between these two sets was assessed by comparing their baseline characteristics (Supplementary Table S2). The two sets were well-balanced, as evidenced by the nearly identical incidence of the primary outcome, delayed extubation (32.3% vs. 32.0%, p = 0.913). The distributions of key predictors, including the ALI, age, comorbidities, and critical indicators of surgical complexity, were all comparable (all p > 0.05). A significant difference was observed only in sex distribution between the sets. 3.3.2 Feature Selection Feature selection was conducted on the training set using LASSO regression with 10-fold cross-validation. The optimal penalty parameter (λ) was determined as the value that lay within one standard error of the minimum binomial deviance (λ.1se = 0.042) (Fig. 4 ). At this λ value, twelve predictors demonstrated non-zero coefficients and were selected for the development of all machine learning models. In descending order of the absolute value of their coefficients, they were: ALI, RDW, Bilateral neck dissection, Age; Extensive maxillomandibular resection, WBC; Smoking status, Tumor T stage, Uric acid, ASA, Flap size and Sodium. 3.3.3 Model Evaluation The predictive performance of all candidate models was rigorously evaluated on an independent validation set. The comprehensive metrics are detailed in Table 3 . Table 3 Predictive performance of seven machine learning models in the training and validation sets Model AUC (95% CI) Accuracy Precision Sensitivity Specificity F1 Score Kappa Logistic 0.758 (0.689, 0.821) 0.707 0.552 0.444 0.830 0.492 0.289 DT 0.806 (0.739, 0.860) 0.773 0.627 0.722 0.797 0.671 0.499 RF 0.875 (0.825, 0.914) 0.813 0.759 0.611 0.908 0.677 0.548 XGBoost 0.857 (0.801, 0.905) 0.707 0.800 0.111 0.987 0.195 0.127 LightGBM 0.847 (0.791, 0.894) 0.800 0.690 0.681 0.856 0.685 0.539 SVM 0.757 (0.687, 0.817) 0.716 0.580 0.403 0.863 0.475 0.289 ANN 0.811 (0.746, 0.867) 0.773 0.644 0.653 0.830 0.648 0.481 In the validation set, all models achieved AUCs above 0.75 (Fig. 5 A). The RF model demonstrated the highest discriminative ability, with an AUC of 0.875 (95% CI: 0.825–0.914) and an accuracy of 0.813. Compared to the Logistic Regression baseline (AUC: 0.758), the RF model also showed the highest precision (0.759) and specificity (0.908). The calibration curve of the RF model closely aligned with the ideal diagonal in the multi-model comparison (Fig. 5 B), which was supported by a Brier score of 0.141. DCA indicated that the RF model provided higher net clinical benefit than the treat-all and treat-none strategies across most threshold probabilities (Fig. 5 C). The selected RF model was further evaluated and demonstrated strong standalone performance, as evidenced by the ROC curve in Fig. 5 D, the calibration curve in Fig. 5 E (Brier score: 0.141), and its DCA in Fig. 5 F. Based on its superior discrimination, calibration, and clinical utility, the RF model was selected as the final predictive model. 3.3.4 Model interpretation SHAP analysis of the final RF model identified key predictors for delayed extubation and elucidated their effects. The global feature importance is summarized in Fig. 6 A. Tumor T stage was the most influential predictor, with the highest mean absolute SHAP value of 0.06, followed by Extensive maxillomandibular resection and Bilateral neck dissection (both 0.04). Among the moderately important features (0.02), the inflammatory-nutritional index ALI was notable. The directional impact of these top features on the prediction is visualized in the beeswarm plot (Fig. 6 B). This analysis confirmed that the presence of major surgical procedures (e.g., Bilateral neck dissection) and higher values of Tumor T stage and RDW were associated with an increased risk. In contrast, higher ALI values were consistently associated with a reduced risk, reinforcing its role as a protective factor. To illustrate how these factors converge in an individual prediction, the SHAP waterfall plot for a representative high-risk patient is shown in Fig. 6 C. Here, a low ALI value was the strongest individual driver increasing the predicted risk, followed by elevated RDW and Age. Finally, the SHAP dependence plots in Fig. 6 D detailed the nature of these associations. For key continuous variables, a near-linear increase in risk was observed with rising RDW, whereas a dose-dependent protective effect was seen with increasing ALI. Binary surgical factors confirmed a distinct risk elevation when the procedure was performed. 4. Discussion This study presents a machine learning–based framework for predicting delayed extubation (DE) following oral cancer surgery with free flap reconstruction. In a cohort of 752 patients, we observed that both established surgical risk factors and the Advanced Lung Cancer Inflammation Index (ALI) were associated with DE risk. By integrating ALI with clinical and surgical variables into a Random Forest model, we achieved high predictive accuracy (AUC 0.875), and SHAP analysis provided insight into the relative contribution of each factor. These findings suggest a potential avenue for more individualized, data-informed perioperative airway management, moving beyond reliance on subjective assessment alone( 20 ). Overall, approximately one-third of patients experienced delayed extubation, underscoring the ongoing clinical challenge of balancing airway safety with minimizing unnecessary prolonged ventilation. The associations observed between preoperative patient characteristics and DE are broadly consistent with the epidemiology of perioperative airway vulnerability. Older age, higher ASA classification, and greater comorbidity burden were more frequent among patients requiring prolonged ventilatory support, reflecting potentially reduced physiological reserve, impaired airway reflexes, and baseline cardiorespiratory compromise( 21 , 22 ). In this cohort, intraoperative variables such as operative duration, blood loss, and transfusion requirement did not differ significantly between groups. This lack of variability may reflect the relatively standardized surgical and anesthetic practices at our single center. These findings highlight the importance of thorough preoperative assessment to identify patients who might benefit from closer airway monitoring postoperatively. Among the factors examined, ALI demonstrated a consistent inverse relationship with DE risk. After adjustment for demographic, comorbidity, surgical, and oncologic confounders, each 10-point increase in ALI was associated with approximately 10% lower odds of delayed extubation. Dose–response analysis indicated a linear trend, and subgroup analyses suggested that the protective association of ALI was generally consistent across age, smoking status, and comorbidity subgroups, with somewhat stronger effects observed among older adults and non-smokers. These findings are consistent with previous studies showing that lower ALI, reflecting systemic inflammation and poor nutritional status, is associated with higher postoperative complication rates and worse outcomes in both oncologic and non-oncologic surgical populations( 23 – 26 ). While causality cannot be inferred, it is plausible that patients with higher ALI experience less postoperative edema and more rapid respiratory muscle recovery, facilitating timely extubation. The components of ALI—body mass index, albumin, and neutrophil-to-lymphocyte ratio—have been individually implicated in perioperative outcomes( 27 ). Low albumin has been associated with impaired wound healing and reduced respiratory muscle strength( 28 , 29 ), while elevated NLR reflects systemic inflammation that can exacerbate tissue edema( 30 ). The present findings suggest that combining these markers into a single index may provide a more holistic assessment of patient vulnerability, offering additional information beyond traditional demographic or comorbidity measures. Clinically, preoperative evaluation of ALI could potentially inform risk stratification and guide perioperative optimization strategies, such as nutritional support or targeted anti-inflammatory interventions, though further prospective studies would be needed to confirm such applications. Surgical factors, particularly extensive maxillomandibular resection and bilateral neck dissection, were strongly associated with DE( 31 – 33 ). These procedures can alter airway geometry, disrupt fascial planes, and increase postoperative soft tissue swelling, all of which may compromise airway patency. SHAP analysis in the Random Forest model highlighted these factors as dominant predictors, consistent with clinical experience. Tumor T stage and flap size also contributed meaningfully to risk prediction, underscoring the impact of surgical complexity on postoperative airway management. Taken together, these results reinforce the need for close coordination between surgical and anesthesia teams when planning extubation strategies, particularly in patients undergoing extensive resections or bilateral neck procedures. Inflammatory hematologic markers, including RDW and white blood cell count, were independently associated with the risk of delayed extubation. A higher RDW may indicate disturbances in erythrocyte homeostasis and systemic inflammation, which could compromise oxygen delivery and contribute to postoperative tissue hypoxia. Elevated WBC similarly suggests ongoing inflammatory stress. These observations are consistent with findings in other surgical populations, where elevated preoperative RDW or WBC have been linked to greater postoperative morbidity and mortality, including increased risk of complications, organ dysfunction, and prolonged recovery periods( 34 – 36 ). Routine preoperative hematologic markers may therefore offer useful complementary information when combined with clinical and surgical data to improve risk stratification. The machine learning framework developed in this study offers several advantages over conventional regression approaches. The Random Forest model was able to capture nonlinear interactions between predictors, such as how a moderately low ALI may have a greater impact in older patients undergoing complex surgery. SHAP analysis provided interpretable visualizations of the contributions of each factor, addressing the common criticism of black-box models and enhancing potential clinical utility. Compared with prior perioperative models that often rely solely on demographic or anatomical predictors, our approach incorporated both inflammatory-nutritional indices and surgical complexity, representing a more comprehensive risk assessment. While promising, the model should be regarded as a decision-support tool rather than a deterministic predictor, and clinical judgment remains essential. Several limitations warrant consideration. First, the retrospective, single-center design may limit generalizability, as institutional practices and surgical techniques vary. Second, some potentially relevant variables, including intraoperative fluid balance, detailed flap design, and dynamic postoperative airway assessments, were not available. Third, ALI and other laboratory markers were assessed preoperatively at a single time point, precluding evaluation of perioperative changes that may influence airway outcomes. Finally, while internal validation demonstrated robust model performance, external validation in independent cohorts is necessary before broad clinical adoption. 5. Conclusion In summary, this study highlights the multifactorial nature of delayed extubation following oral cancer surgery with free flap reconstruction, with contributions from preoperative inflammatory-nutritional status, surgical complexity, and systemic physiologic markers. ALI emerged as a promising, independent marker of protective effect, and the Random Forest model, complemented by SHAP interpretability, offers a nuanced approach to perioperative risk assessment. These findings may inform individualized airway management strategies and encourage further exploration of inflammatory-nutritional optimization as part of preoperative preparation. Future research should aim to validate these results across multiple centers, evaluate dynamic perioperative changes in ALI, and investigate whether interventions targeting nutritional and inflammatory status can meaningfully influence airway outcomes. Abbreviations DE Delayed Extubation ALI Advanced Lung Cancer Inflammation Index ICU Intensive Care Unit ML Machine Learning SHAP SHapley Additive exPlanations RF Random Forest NLR Neutrophil-to-Lymphocyte Ratio PLR Platelet-to-Lymphocyte Ratio BMI Body Mass Index ASA American Society of Anesthesiologists RDW Red Cell Distribution Width WBC White Blood Cell DCA Decision Curve Analysis RCS Restricted Cubic Spline ANN Artificial Neural Network SVM Support Vector Machine XGBoost eXtreme Gradient Boosting LightGBM Light Gradient Boosting Machine OR Odds Ratio CI Confidence Interval IQR Interquartile Range. Declarations Ethics approval and consent to participate This retrospective study was approved by the Biomedical Research Ethics Committee of Suining Central Hospital (Approval No. KYLLKYS20250144). The requirement for informed consent was waived due to the retrospective nature of the study and anonymized data handling. All procedures were conducted in accordance with the principles of the Declaration of Helsinki. Consent for publication Not applicable. Data Availability statement The datasets generated and/or analyzed during the current study are not publicly available due to patient confidentiality and privacy restrictions but are available from the corresponding author on reasonable request. Competing interests The authors declare no competing interests. Funding Statement This research was funded by the Sichuan Provincial Department of Science and Technology [grant number 2022SNZY001]; the Sichuan Provincial Health and Health Commission [grant number 2022JDXM021]; and the Suining Municipal Health and Science Technology Program – Guiding Project [grant number 25ZDJB08]. Author Contributions B.J. and S.Y. conceived and designed the study. B.J. and X.W. performed the data analysis and wrote the first draft of the manuscript. Q.W., H.Z., and J.R. contributed to data collection and interpretation. G.L. and X.Z. provided administrative and technical support. C.Y. prepared the figures and tables. X.W. and S.Y. critically revised the manuscript. All authors read and approved the final version of the manuscript. Acknowledgments We appreciate the assistance in data collection, statistical analysis, and manuscript preparation, which contributed to the completion of this study. References Li MM, Miller LE, Old M. State of Head and Neck Microvascular Reconstruction: Current and Future Directions. Surg Oncol Clin N Am. 2024;33(4):711-21. Myatra SN, Kornerup SC, Parotto M. Extubation in head and neck surgery. Curr Opin Anaesthesiol. 2025;38(6):869-76. Riekert M, Rempel V, Keilwerth S, Zöller JE, Kreppel M, Schick VC. Airway-Associated Complications With and Without Primary Tracheotomy in Oral Squamous Cell Carcinoma Surgery. J Craniofac Surg. 2023;34(1):279-83. Madgar O, Livneh N, Dobriyan A, Dagan E, Alon EE. Airway management following head and neck microvascular reconstruction: is tracheostomy mandatory? Braz J Otorhinolaryngol. 2022;88 Suppl 4(Suppl 4):S44-s9. Coyle MJ, Tyrrell R, Godden A, Hughes CW, Perkins C, Thomas S, Godden D. Replacing tracheostomy with overnight intubation to manage the airway in head and neck oncology patients: towards an improved recovery. Br J Oral Maxillofac Surg. 2013;51(6):493-6. Lee HJ, Kim JW, Choi SY, Kim CS, Kwon TG, Paeng JY. The evaluation of a scoring system in airway management after oral cancer surgery. Maxillofac Plast Reconstr Surg. 2015;37(1):19. Schmutz A, Dieterich R, Kalbhenn J, Voss P, Loop T, Heinrich S. Protocol based evaluation for feasibility of extubation compared to clinical scoring systems after major oral cancer surgery safely reduces the need for tracheostomy: a retrospective cohort study. BMC Anesthesiol. 2018;18(1):43. Janik S, Brkic FF, Grasl S, Königswieser M, Franz P, Erovic BM. Tracheostomy in bilateral neck dissection: Comparison of three tracheostomy scoring systems. Laryngoscope. 2020;130(11):E580-e6. Kwon MA, Song J, Kim S, Oh PW, Kang M. Airway Management Failure after Delayed Extubation in a Patient with Oral Malignant Melanoma Who Underwent Partial Mandibulectomy and Reconstruction with a Free Flap. Case Rep Dent. 2021;2021:7792843. Mossinelli C, Pietrobon G, Zorzi S, Tagliabue M, Chu F, Tomarchio E, et al. Airway assessment and management in head and neck cancer surgery. Acta Otorhinolaryngol Ital. 2025;45(3):173-81. Xu S, Wang K, Liu K, Liu Y, Huang Y, Zhang Y, et al. Predictive Nomogram for the Necessity of Tracheotomy During Oral and Oropharyngeal Cancer Surgery. Laryngoscope. 2021;131(5):E1489-e95. Liu F, Liu Z, Han Z, Cao M, Liufu N, Fu G. Development and Validation of a Nomogram Incorporating Changes in Inflammatory Markers for Predicting Complications After Free Flap Reconstruction of Oral and Maxillofacial Defects. J Craniofac Surg. 2025;36(6):e727-e31. Liu Q, Zhou Y, Cao X, Wang W, Pan C, YichenXu, et al. The impact of systemic inflammation index on prolonged mechanical ventilation after cardiac surgery: a retrospective study. J Cardiothorac Surg. 2025;20(1):293. Chen Y, Wu G, Wang R, Chen J. Preoperative Albumin Level Serves as a Predictor for Postoperative Pulmonary Complications Following Elective Laparoscopic Gastrectomy. Curr Pharm Des. 2018;24(27):3250-5. Jafri SH, Shi R, Mills G. Advance lung cancer inflammation index (ALI) at diagnosis is a prognostic marker in patients with metastatic non-small cell lung cancer (NSCLC): a retrospective review. BMC Cancer. 2013;13:158. Deo RC. Machine Learning in Medicine. Circulation. 2015;132(20):1920-30. Moharrami M, Azimian Zavareh P, Watson E, Singhal S, Johnson AEW, Hosni A, et al. Prognosing post-treatment outcomes of head and neck cancer using structured data and machine learning: A systematic review. PLoS One. 2024;19(7):e0307531. Asaad M, Lu SC, Hassan AM, Kambhampati P, Mitchell D, Chang EI, et al. The Use of Machine Learning for Predicting Complications of Free-Flap Head and Neck Reconstruction. Ann Surg Oncol. 2023;30(4):2343-52. Singh R, Lanchantin J, Sekhon A, Qi Y. Attend and Predict: Understanding Gene Regulation by Selective Attention on Chromatin. Adv Neural Inf Process Syst. 2017;30:6785-95. Singh T, Sankla P, Smith G. Tracheostomy or delayed extubation after maxillofacial free-flap reconstruction? Br J Oral Maxillofac Surg. 2016;54(8):878-82. Boruk M, Chernobilsky B, Rosenfeld RM, Har-El G. Age as a prognostic factor for complications of major head and neck surgery. Arch Otolaryngol Head Neck Surg. 2005;131(7):605-9. Wu L, Li Q, Zhu Z, Li S, Zhu Z, Zhang T. Impact of preoperative comorbidities on postoperative complication rates and survival outcome in patients with head and neck cancer undergoing surgical treatment. Sci Rep. 2025;15(1):38746. Kobayashi S, Karube Y, Inoue T, Araki O, Maeda S, Matsumura Y, Chida M. Advanced Lung Cancer Inflammation Index Predicts Outcomes of Patients with Pathological Stage IA Lung Adenocarcinoma Following Surgical Resection. Ann Thorac Cardiovasc Surg. 2019;25(2):87-94. Gu Y, Zhang C, Miao JB, Wang H, Hu B, Li X. The association of the advanced lung cancer inflammation index with postoperative complications in patients undergoing lung resection for bronchiectasis. J Thorac Dis. 2025;17(6):3577-89. Hua X, Chen J, Wu Y, Sha J, Han S, Zhu X. Prognostic role of the advanced lung cancer inflammation index in cancer patients: a meta-analysis. World J Surg Oncol. 2019;17(1):177. Yin C, Toiyama Y, Okugawa Y, Omura Y, Kusunoki Y, Kusunoki K, et al. Clinical significance of advanced lung cancer inflammation index, a nutritional and inflammation index, in gastric cancer patients after surgical resection: A propensity score matching analysis. Clin Nutr. 2021;40(3):1130-6. Qiu X, Shen S, Jiang N, Feng Y, Yang G, Lu D. Association of advanced lung cancer inflammation index with all-cause and cardiovascular mortality in metabolic dysfunction associated steatotic liver disease. Sci Rep. 2025;15(1):15121. Zhang Y, Tan S, Wu G. ESPEN practical guideline: Clinical nutrition in surgery. Clin Nutr. 2021;40(9):5071. Zheng M, Zhang X, Wang H, Yuan P, Yu Q. Interpretable machine learning model for identification and risk factor of premature rupture of membranes (PROM) and its association with nutritional inflammatory index: a retrospective study. Front Med (Lausanne). 2025;12:1557919. Feier CVI, Motoc A, Muntean C, Vonica RC, Gaborean V, Olariu S, Murariu MS. Systemic inflammatory indices and age-dependent severity in acute appendicitis: a retrospective cohort study. Front Immunol. 2025;16:1620459. Hasegawa T, Yatagai N, Furukawa T, Wakui E, Saito I, Takeda D, et al. The prospective evaluation and risk factors of dysphagia after surgery in patients with oral cancer. J Otolaryngol Head Neck Surg. 2021;50(1):4. Ou M, Wang G, Yan Y, Chen H, Xu X. Perioperative symptom burden and its influencing factors in patients with oral cancer: A longitudinal study. Asia Pac J Oncol Nurs. 2022;9(8):100073. Kuo PJ, Lin PC, Hsieh CH. Airway Management Following Head and Neck Microvascular Reconstruction: When is a Tracheostomy Necessary? Risk Manag Healthc Policy. 2025;18:2551-63. Wei W, Feng B, Chen Z, Liu X, Xiao M, Hu H. Association of preoperative red blood cell width and postoperative 30-day mortality in patients undergoing non-cardiac surgery: a retrospective cohort study using propensity-score matching. Perioper Med (Lond). 2024;13(1):95. Frentiu AA, Mao K, Caruana CB, Raveendran D, Perry LA, Penny-Dimri JC, et al. The Prognostic Significance of Red Cell Distribution Width in Cardiac Surgery: A Systematic Review and Meta-Analysis. J Cardiothorac Vasc Anesth. 2023;37(3):471-9. van Koeverden ID, den Ruijter HM, Scholtes VPW, M GEHL, Haitjema S, Buijsrogge MP, et al. A single preoperative blood test predicts postoperative sepsis and pneumonia after coronary bypass or open aneurysm surgery. Eur J Clin Invest. 2019;49(3):e13055. Additional Declarations No competing interests reported. Supplementary Files SupplementaryMaterials.docx Supplementary Material Table S1. Optimal hyperparameters for different machine learning models. Table S2.Comparison of Baseline Characteristics Between the Training and Testing Sets. 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-8243006","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":557546880,"identity":"2c9d8570-1370-45a6-9fad-1366e770781b","order_by":0,"name":"Baolin Jia","email":"","orcid":"","institution":"Suining Central Hospital","correspondingAuthor":false,"prefix":"","firstName":"Baolin","middleName":"","lastName":"Jia","suffix":""},{"id":557546881,"identity":"21b50c69-18f1-499d-8e0e-53172cd83bab","order_by":1,"name":"Chuan Ye","email":"","orcid":"","institution":"Suining Central Hospital","correspondingAuthor":false,"prefix":"","firstName":"Chuan","middleName":"","lastName":"Ye","suffix":""},{"id":557546882,"identity":"3055e7ff-6afe-4fcd-8c8c-88125b13867e","order_by":2,"name":"Huan Zhang","email":"","orcid":"","institution":"Suining Central Hospital","correspondingAuthor":false,"prefix":"","firstName":"Huan","middleName":"","lastName":"Zhang","suffix":""},{"id":557546883,"identity":"a3f1fb5c-fab1-46a1-b336-25fba7351ee0","order_by":3,"name":"Jun Ren","email":"","orcid":"","institution":"Suining Central Hospital","correspondingAuthor":false,"prefix":"","firstName":"Jun","middleName":"","lastName":"Ren","suffix":""},{"id":557546884,"identity":"188c0270-6230-4c4d-b271-f9d633de88f2","order_by":4,"name":"Guixin Li","email":"","orcid":"","institution":"Suining Central Hospital","correspondingAuthor":false,"prefix":"","firstName":"Guixin","middleName":"","lastName":"Li","suffix":""},{"id":557546885,"identity":"1ea55cf6-c4a1-4c46-98ca-94d7f4fe81ca","order_by":5,"name":"XianJie Zheng","email":"","orcid":"","institution":"Suining Central Hospital","correspondingAuthor":false,"prefix":"","firstName":"XianJie","middleName":"","lastName":"Zheng","suffix":""},{"id":557546886,"identity":"22b039c5-723a-46a8-9ac4-3df23328ed22","order_by":6,"name":"Qiang Wang","email":"","orcid":"","institution":"Suining Central Hospital","correspondingAuthor":false,"prefix":"","firstName":"Qiang","middleName":"","lastName":"Wang","suffix":""},{"id":557546887,"identity":"fee09745-b5ec-4bf7-b15e-fee021dfdb63","order_by":7,"name":"Xiaojuan Wu","email":"","orcid":"","institution":"Suining Central Hospital","correspondingAuthor":false,"prefix":"","firstName":"Xiaojuan","middleName":"","lastName":"Wu","suffix":""},{"id":557546888,"identity":"cb96f401-6d65-44c0-a3ad-0244c59eba98","order_by":8,"name":"Sen Yang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9ElEQVRIie2RMWoDMRBFJQRWk41bmQ1WjhChJkUOI2FwZWJCGhchGVjYygcwJOQYrscInMbgVqU3e4H1DSKvS4PkMoVeJcQ8vv6IkEzmX0KrQ7cQTEo0m3AYy7TCarXaPXG1Io0Tu0etIKlwKIt6OiSetljUC0swITzsNxBSXEk/K4Ojb2EosObXxxRvIXRxmt9tDb6sxTMnA61nUYX2KRMKs5CyFq8UbgZlVAkPC13cB+C8w+JLWMCUgravz4g3Bgu4Qhl52y+ZqSUaJ7ZCqyrR5Xb/056/ksPk2L29jyWvmjam3OPFFYuMn5CQGMhkMpkM+QPfu1qWYw413AAAAABJRU5ErkJggg==","orcid":"","institution":"Suining Central Hospital","correspondingAuthor":true,"prefix":"","firstName":"Sen","middleName":"","lastName":"Yang","suffix":""}],"badges":[],"createdAt":"2025-11-30 15:23:16","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8243006/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8243006/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":98048135,"identity":"5f5fc78d-8ede-41b4-aec1-aca63a798158","added_by":"auto","created_at":"2025-12-12 08:31:59","extension":"tif","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4739924,"visible":true,"origin":"","legend":"","description":"","filename":"Fig1.tif","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/ac3cc76d6b48c73c7bfcf8b2.tif"},{"id":98425922,"identity":"ee6c9b8e-586b-43da-a5a2-c09a36723a48","added_by":"auto","created_at":"2025-12-17 16:35:21","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":102611,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/003929d24f3d99c5cfea3f14.docx"},{"id":98048134,"identity":"7f70cde8-a48a-4a86-9f44-98beba1ff1a3","added_by":"auto","created_at":"2025-12-12 08:31:59","extension":"tiff","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":342069,"visible":true,"origin":"","legend":"","description":"","filename":"Fig2.tiff","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/b09b8b0a03ee5f98e90a696c.tiff"},{"id":98048111,"identity":"e95a57b3-beae-4574-98db-6fe693910a1a","added_by":"auto","created_at":"2025-12-12 08:31:58","extension":"tif","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9263764,"visible":true,"origin":"","legend":"","description":"","filename":"Fig3.tif","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/e525c04588a7018f4dd9f912.tif"},{"id":98425940,"identity":"9764cd80-2744-4f2e-8bd8-648357a257c2","added_by":"auto","created_at":"2025-12-17 16:35:23","extension":"tif","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2484272,"visible":true,"origin":"","legend":"","description":"","filename":"Fig4.tif","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/da104c041330106ac9b5716e.tif"},{"id":98048113,"identity":"1621cd6a-72dc-4763-88d9-9068f513438c","added_by":"auto","created_at":"2025-12-12 08:31:58","extension":"tif","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7611696,"visible":true,"origin":"","legend":"","description":"","filename":"Fig5.tif","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/3241fe66d1935e75d005a508.tif"},{"id":98427491,"identity":"007c516c-d285-49fb-9ea5-86c6550db388","added_by":"auto","created_at":"2025-12-17 16:40:33","extension":"tif","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4269976,"visible":true,"origin":"","legend":"","description":"","filename":"Fig6.tif","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/e43597a75bbaa18b4f08b5b9.tif"},{"id":98048107,"identity":"f578b43e-1a26-4781-ace5-124e53b30a5b","added_by":"auto","created_at":"2025-12-12 08:31:58","extension":"json","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10006,"visible":true,"origin":"","legend":"","description":"","filename":"e566cf4693594c088b11819cc0e6dff0.json","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/c4b8d3da07b61323c3576ac8.json"},{"id":98048108,"identity":"3f465345-9677-4d3d-9fbe-6fa7cd7117b6","added_by":"auto","created_at":"2025-12-12 08:31:58","extension":"docx","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":22268,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterials.docx","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/076aedbd1a623eda6ac8e0f5.docx"},{"id":98425937,"identity":"485e3294-64f8-4a1d-96c9-dc98333d7373","added_by":"auto","created_at":"2025-12-17 16:35:23","extension":"xml","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":153314,"visible":true,"origin":"","legend":"","description":"","filename":"e566cf4693594c088b11819cc0e6dff01enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/7772c413442cca9a119669de.xml"},{"id":98427498,"identity":"07af3ac0-6037-43b8-8822-89680b347e7e","added_by":"auto","created_at":"2025-12-17 16:40:33","extension":"tif","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4739924,"visible":true,"origin":"","legend":"","description":"","filename":"Fig1.tif","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/ed36cfba78a00624d654894a.tif"},{"id":98427514,"identity":"fba68c4c-db2e-4f3c-bf8f-542e91d252a3","added_by":"auto","created_at":"2025-12-17 16:40:36","extension":"tiff","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":342069,"visible":true,"origin":"","legend":"","description":"","filename":"Fig2.tiff","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/0290ba6563fc8871d62359c5.tiff"},{"id":98048126,"identity":"23153ef2-7eb3-4d62-b8a9-ef3d92651c47","added_by":"auto","created_at":"2025-12-12 08:31:58","extension":"tif","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9263764,"visible":true,"origin":"","legend":"","description":"","filename":"Fig3.tif","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/0f97f2bb4f32cff7775d6a28.tif"},{"id":98427465,"identity":"c8640dad-0d0d-4ab8-b1a5-33776a76514d","added_by":"auto","created_at":"2025-12-17 16:40:29","extension":"tif","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2484272,"visible":true,"origin":"","legend":"","description":"","filename":"Fig4.tif","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/cb6b4c7ff83bd802f81e613c.tif"},{"id":98048131,"identity":"aaae6bde-41d4-473f-9941-4b6b36e7ac1a","added_by":"auto","created_at":"2025-12-12 08:31:58","extension":"tif","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7611696,"visible":true,"origin":"","legend":"","description":"","filename":"Fig5.tif","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/a3a48dad131117c3228559fd.tif"},{"id":98048124,"identity":"0101ab59-d5dc-4797-aa85-b7026efa7c2a","added_by":"auto","created_at":"2025-12-12 08:31:58","extension":"tif","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4269976,"visible":true,"origin":"","legend":"","description":"","filename":"Fig6.tif","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/66d88c17760cfa7c429b8c0f.tif"},{"id":98048130,"identity":"4b962c2c-2659-4c6a-8baf-813ac3f22462","added_by":"auto","created_at":"2025-12-12 08:31:58","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1768299,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig1.png","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/ca6d2e882a85336493e9a050.png"},{"id":98426086,"identity":"7d2da582-7d03-40e3-834a-df719e9480f7","added_by":"auto","created_at":"2025-12-17 16:35:36","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":106657,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig2.png","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/0d4295eb539e856e77992c5a.png"},{"id":98425965,"identity":"2d3c6907-47c5-4cbc-baf4-6561add17d0a","added_by":"auto","created_at":"2025-12-17 16:35:25","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4153657,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig3.png","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/e40f5acd3394643f16c65f3a.png"},{"id":98048122,"identity":"35719d70-8072-4ede-b1e5-fe2ee1df2428","added_by":"auto","created_at":"2025-12-12 08:31:58","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":559014,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig4.png","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/5a6ca28fb4c3164e2622c7b9.png"},{"id":98048133,"identity":"f07475de-b668-4146-869e-d2d60e229398","added_by":"auto","created_at":"2025-12-12 08:31:58","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2288751,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig5.png","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/84b167b882c3bea43710f21a.png"},{"id":98048128,"identity":"bc382a38-57c5-41c3-b142-03ef25290dd0","added_by":"auto","created_at":"2025-12-12 08:31:58","extension":"png","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1467836,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFig6.png","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/44db64a57d7382f21f195dca.png"},{"id":98427627,"identity":"9262adb9-199f-4fb4-a2f1-52e71ab91ef2","added_by":"auto","created_at":"2025-12-17 16:40:53","extension":"xml","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":150366,"visible":true,"origin":"","legend":"","description":"","filename":"e566cf4693594c088b11819cc0e6dff01structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/3f7fd7d966e7a88552b96deb.xml"},{"id":98427804,"identity":"44d3eac9-5a3f-4c2a-99b9-ecff8fa0b1e7","added_by":"auto","created_at":"2025-12-17 16:41:14","extension":"html","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":161298,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/a23b0647940db08566d7daf9.html"},{"id":98426020,"identity":"b8288ed9-0692-4ef1-9877-3e186d43f30a","added_by":"auto","created_at":"2025-12-17 16:35:32","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":3167516,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of study design and methodology.\u003c/p\u003e","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/9b2c04fab7198b1a62d31567.png"},{"id":98428051,"identity":"acf06ed2-3990-4a8f-acf2-25faa8231b55","added_by":"auto","created_at":"2025-12-17 16:41:32","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":2966865,"visible":true,"origin":"","legend":"\u003cp\u003eRestricted cubic spline (RCS) analysis of the association between ALI and DE. The model was adjusted for Sex, Smoking status, Alcohol intake, Hypertension, DM, Hyperlipidemia, COPD, CVD, ASA, Tumor T stage, Extensive maxillomandibular resection, Bilateral neck dissection, Flap size, Age, Tumor site.\u003c/p\u003e","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/f428da59be4af31255b45a27.png"},{"id":98427717,"identity":"77ab681f-e013-4e77-a82b-2ce10609952a","added_by":"auto","created_at":"2025-12-17 16:41:00","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":6704795,"visible":true,"origin":"","legend":"\u003cp\u003eSubgroup analysis of the association between ALI and DE. Adjusted for Sex, Smoking status, Alcohol intake, Hypertension, DM, Hyperlipidemia, COPD, CVD, ASA, Tumor T stage, Extensive maxillomandibular resection, Bilateral neck dissection, Flap size, Age, Tumor site.\u003c/p\u003e","description":"","filename":"Fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/774be9187d907a3f70c443be.png"},{"id":98427406,"identity":"dbffbcde-01f9-4bab-8f9e-f2ebce90ca7f","added_by":"auto","created_at":"2025-12-17 16:40:18","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":1953261,"visible":true,"origin":"","legend":"\u003cp\u003eMultidimensional analyses of variables related to DE. (A) LASSO regression coefficients plotted across different penalty parameter (lambda) values. (B) Tenfold cross-validation of the LASSO model to select the optimal penalty parameter (lambda.1se), identifying the most relevant features associated with DE.\u003c/p\u003e","description":"","filename":"Fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/5605d12718dfe87070fd1ef4.png"},{"id":98048117,"identity":"f524a478-da79-4948-9b9f-4942b498a5b9","added_by":"auto","created_at":"2025-12-12 08:31:58","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":13259105,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance evaluation of the models in the validation set. (A) Receiver operating characteristic (ROC) curves in the validation set. (B) Calibration curves of seven models in the validation set. (C) Decision curve analysis (DCA) of seven models in the validation set. (D) ROC curve of the selected RF model. (E) Calibration curve of the selected RF model (Brier score: 0.141). (F) DCA of the selected RF model demonstrating net clinical benefit across threshold probabilities.\u003c/p\u003e","description":"","filename":"Fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/92a0efe249786aba1d66c50f.png"},{"id":98048136,"identity":"cc709e1d-c226-496e-8f4a-b905587583f4","added_by":"auto","created_at":"2025-12-12 08:31:59","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":4994734,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP analysis of the final random forest (RF) model for delayed extubation. (A) SHAP significance analysis of feature importance ranking based on the mean value. (B) SHAP Beeswarm plot of the RF model. (C) SHAP waterfall plot for a representative high-risk patient, illustrating individual feature contributions. (D) SHAP dependence plots for selected continuous and binary features.\u003c/p\u003e","description":"","filename":"Fig6.png","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/fe187e526dbe3772c37123af.png"},{"id":106994314,"identity":"bd1277e3-091c-416a-86f1-245c0ac83be1","added_by":"auto","created_at":"2026-04-15 15:07:43","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":33168008,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/d46b956a-f6c3-407e-b6b8-3e20a73dbcd8.pdf"},{"id":98048105,"identity":"3f8bd496-e838-47ef-afaa-c888de69d662","added_by":"auto","created_at":"2025-12-12 08:31:58","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":22268,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSupplementary Material\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable S1. \u003c/strong\u003eOptimal hyperparameters for different machine learning models.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable S2.\u003c/strong\u003eComparison of Baseline Characteristics Between the Training and Testing Sets.\u003c/p\u003e","description":"","filename":"SupplementaryMaterials.docx","url":"https://assets-eu.researchsquare.com/files/rs-8243006/v1/30d3de9a904c80f47c329d36.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Preoperative Nutrition–Inflammation Status and Surgical Factors Predict Delayed Extubation After Oral Cancer Surgery with Free Flap Reconstruction: A Machine Learning Approach","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eExtensive resection of oral malignancies combined with free flap reconstruction remains a fundamental component of contemporary head and neck cancer treatment(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). Although ongoing refinements in microsurgical techniques have improved both tumor control and postoperative functional outcomes, perioperative airway management continues to be a critical concern(\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Altered upper airway anatomy, substantial postoperative edema, and the risk of hematoma formation may rapidly compromise airway patency in the immediate postoperative period. These challenges have been linked to higher complication rates, prolonged ICU stays, and increased postoperative morbidity and mortality(\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eCurrently, three primary strategies are commonly employed in clinical practice: immediate extubation in the operating room, delayed extubation (DE) in the ICU, or prophylactic tracheostomy(\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). Considerable variation exists in the selection of these approaches across institutions and among individual surgeons, reflecting the lack of robust, objective, and generalizable tools for perioperative risk assessment(\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). Existing clinical scoring systems, such as the widely cited Kruse-L\u0026ouml;sler score, have demonstrated limited predictive performance, with positive predictive values ranging from 0.08 to 0.18 in validation studies(\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e). As a result, conservative strategies may lead to unnecessary tracheostomies, whereas overly aggressive extubation could precipitate critical airway compromise(\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e), underscoring the ongoing challenge of balancing safety and efficiency in postoperative airway management.\u003c/p\u003e\u003cp\u003ePrevious studies have predominantly focused on anatomical or surgical factors, including tumor T stage, extent of mandibular resection, bilateral neck dissection, and flap volume(\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). While these variables undoubtedly influence the degree of airway compromise, traditional statistical models may not fully capture a patient\u0026rsquo;s overall physiological vulnerability, particularly with respect to systemic factors such as inflammation and nutritional status. Evidence suggests that elevated perioperative inflammation is associated with increased tissue edema, and inflammatory indices such as the neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) have been linked not only to flap complications(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e) but also to delayed extubation and pulmonary events in other surgical populations(\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). Similarly, malnutrition, reflected by conditions such as hypoalbuminemia, has been recognized as an independent risk factor for multiple postoperative pulmonary complications after major surgery(\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e). Despite these insights, there remains a paucity of research in head and neck surgery that integrates inflammatory and nutritional assessment to inform perioperative airway management strategies.\u003c/p\u003e\u003cp\u003eThe Advanced Lung Cancer Inflammation Index (ALI) has been widely validated as an integrative measure of patients\u0026rsquo; inflammatory burden and nutritional reserves, combining BMI, serum albumin, and NLR, and serving as a prognostic marker across multiple malignancies(\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). A lower ALI theoretically reflects heightened systemic inflammation and poorer nutritional status, which may contribute to prolonged postoperative edema and delayed recovery, thereby increasing the likelihood of delayed extubation. However, the application of ALI in postoperative airway management for oral cancer remains largely unexplored. To our knowledge, no prior studies have systematically combined ALI with detailed clinical and surgical factors to assess delayed extubation risk, representing a notable knowledge gap that the present study aims to address.\u003c/p\u003e\u003cp\u003eMeanwhile, machine learning (ML) approaches are increasingly being explored for perioperative risk assessment, offering the ability to handle high-dimensional clinical data with complex interactions and nonlinear relationships. In the context of head and neck surgery, ML models have been applied to predict outcomes such as flap necrosis and overall survival, often demonstrating improved predictive performance compared with conventional regression techniques(\u003cspan additionalcitationids=\"CR17\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). Nevertheless, to date, no studies have integrated ML with systemic inflammation\u0026ndash;nutrition indices to predict delayed extubation following free flap reconstruction in oral cancer patients. Moreover, the use of interpretable ML methods remains limited, leaving a gap in clinically actionable insights that could support individualized airway management decisions.\u003c/p\u003e\u003cp\u003eTo address current gaps in the literature, this study focuses on three interrelated objectives. We first evaluate the independent association between the inflammation\u0026ndash;nutrition composite index ALI and the risk of delayed extubation following oral cancer surgery with free flap reconstruction. We then develop and validate a random forest\u0026ndash;based prediction model to improve perioperative risk stratification, incorporating both clinical and surgical variables. Finally, we apply SHAP analysis to enhance the interpretability of the model, providing clinicians with a clearer understanding of the factors driving individual predictions. Collectively, these efforts aim to offer data-driven guidance for postoperative airway management and support the optimization of individualized extubation strategies.\u003c/p\u003e"},{"header":"2. Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1. Data Source and Study Population\u003c/h2\u003e\u003cp\u003e This retrospective, single-center observational study included patients who underwent oral cancer surgery with free flap reconstruction at the Department of Oral and Maxillofacial Surgery, Suining Central Hospital, China, between August 2017 and August 2025. The study protocol was approved by the Biomedical Research Ethics Committee of Suining Central Hospital (Approval No. KYLLKYS20250144). The requirement for informed consent was waived due to the retrospective nature of the study. All procedures followed the principles of the Declaration of Helsinki, and all patient information was anonymized before analysis.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2. Inclusion and Exclusion Criteria\u003c/h2\u003e\u003cp\u003ePatients were considered eligible if they had a histopathologically confirmed diagnosis of primary oral squamous cell carcinoma and underwent tumor resection followed by immediate free flap reconstruction. We excluded patients who received a prophylactic tracheostomy at the time of surgery, those who required reintubation or subsequent tracheostomy after an initially successful extubation, and individuals with incomplete or missing essential perioperative information. The patient selection process is summarized in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.3. Data Collection and Variable Definitions\u003c/h2\u003e\u003cp\u003eClinical data were retrospectively collected from the electronic medical record system using a structured template. The extracted variables encompassed demographic characteristics, comorbidities, tumor- and surgery-related profiles, and intraoperative metrics. Preoperative laboratory values were recorded for all patients, and the ALI was calculated as (body mass index \u0026times; serum albumin) / neutrophil-to-lymphocyte ratio(\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eTo ensure data integrity, two trained clinicians independently performed the data extraction. Any discrepancies were resolved through consensus or, when necessary, adjudication by a senior investigator.\u003c/p\u003e\u003cp\u003eThe primary outcome was delayed extubation, defined as the planned transfer of the patient to the intensive care unit with the endotracheal tube in place for ongoing ventilatory support, rather than proceeding with immediate extubation in the operating room. This endpoint aligns with established definitions in studies of airway management following major head and neck reconstruction(\u003cspan additionalcitationids=\"CR4\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). Patients requiring reintubation after a successful initial extubation were excluded from the primary analysis, as this event typically stems from distinct postoperative complications.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.4 Statistical Analysis\u003c/h2\u003e\u003cdiv id=\"Sec7\" class=\"Section3\"\u003e\u003ch2\u003e2.4.1 Baseline analysis and correlation analysis\u003c/h2\u003e\u003cp\u003eThe final analysis included 752 patients after applying the predefined exclusion criteria. We summarized baseline characteristics for the entire cohort, comparing them between patients with and without delayed extubation. Categorical variables are reported as counts (percentages), compared using the χ\u0026sup2; or Fisher's exact test, while continuous variables, presented as medians (IQRs), were compared with the Mann-Whitney U test.\u003c/p\u003e\u003cp\u003eWe assessed the association between ALI and delayed extubation using logistic regression in three steps: an unadjusted model, a model adjusted for demographics and lifestyle, and a fully adjusted model that also included comorbidities, tumor stage, and surgical extent. ALI was analyzed as a continuous measure (per 10-point increase) and by tertiles. We explored the dose-response relationship using restricted cubic splines and conducted pre-specified subgroup analyses to test the consistency of the association.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section3\"\u003e\u003ch2\u003e2.4.2 Model development and validation\u003c/h2\u003e\u003cp\u003eThe cohort was randomly split into a training set (70%, n\u0026thinsp;=\u0026thinsp;527) and a testing set (30%, n\u0026thinsp;=\u0026thinsp;225) to support model development and internal validation. Feature selection was carried out within the training set using Least Absolute Shrinkage and Selection Operator (LASSO) regression with 10-fold cross-validation, which yielded twelve predictors with non-zero coefficients for subsequent analysis.\u003c/p\u003e\u003cp\u003eThese predictors were then used to develop seven machine-learning models: logistic regression, decision tree, Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), support vector machine (SVM), and an artificial neural network (ANN). Each model underwent hyperparameter tuning through 10-fold cross-validation within the training set, with parameter grids tailored to the characteristics of each algorithm. The full list of candidate hyperparameters and the final selected values is provided in Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e.\u003c/p\u003e\u003cp\u003eAll tuned models were evaluated on the independent testing set. Model discrimination was assessed by the area under the receiver operating characteristic curve (AUC), and calibration was examined using calibration plots and the Brier score. Decision curve analysis (DCA) was performed to assess clinical utility. Additional performance metrics\u0026mdash;including accuracy, sensitivity, specificity, and the F1 score\u0026mdash;were also calculated. The final model was selected based on a comprehensive assessment of these performance measures, prioritizing discriminative ability and calibration quality.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec9\" class=\"Section3\"\u003e\u003ch2\u003e2.4.3 Model interpretability\u003c/h2\u003e\u003cp\u003eTo improve the clinical interpretability of the final model and mitigate its black-box characteristics, we applied the SHapley Additive exPlanations (SHAP) framework in a post hoc analysis to quantify the contribution of individual predictors. SHAP is grounded in cooperative game theory and provides a unified measure of feature influence on model output(\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThe analysis was structured to yield both global and local insights. At the global level, mean absolute SHAP values were used to rank the relative importance of each predictor, and summary plots were generated to illustrate the distribution and directional impact of feature contributions across the cohort. For continuous predictors, dependence plots were constructed to characterize their marginal relationship with the model\u0026rsquo;s predicted risk. For patient-level interpretation, SHAP values were decomposed for individual predictions to highlight the specific variables driving risk estimates. These contributions were visualized using waterfall or force plots to facilitate case-based reasoning and potential bedside applicability.\u003c/p\u003e\u003cp\u003eAll statistical analyses and model development were performed in R (version 4.4.3). SHAP computations and visualization were conducted in Python (version 3.12). A two-tailed \u003cem\u003ep\u003c/em\u003e value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically significant. The overall study workflow is presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\n \u003ch2\u003e3.1 Baseline Characteristics of the Oral Cancer Surgery with Free Flap Reconstruction Cohort\u003c/h2\u003e\n \u003cp\u003eThis study included 752 patients undergoing oral cancer surgery with free flap reconstruction. The incidence of delayed extubation was 32.2% (242/752). Baseline characteristics stratified by extubation outcome are detailed in Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eBaseline characteristics stratified by extubation outcome\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eVariables\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eTotal\u003c/p\u003e\n \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;752)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eNon-delayed extubation (n\u0026thinsp;=\u0026thinsp;510)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eDelayed extubation\u003c/p\u003e\n \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;242)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e56.00 (48.00, 65.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e55.00 (46.00, 64.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e60.00 (51.00, 67.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBMI, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27.70 (24.30, 32.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27.40 (24.20, 31.60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e28.30 (24.60, 32.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.040\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSurgery Time, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e418.50 (365.00, 481.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e418.00 (366.00, 482.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e420.50 (364.00, 478.25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.844\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBlood Loss, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e348.00 (281.00, 426.25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e347.00 (282.25, 427.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e350.00 (280.00, 423.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.852\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBP fluctuation, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12.50 (9.00, 16.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12.00 (9.00, 16.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13.00 (9.00, 17.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.307\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHR fluctuation, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e9.00 (6.00, 12.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e9.00 (6.00, 12.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e9.00 (6.00, 11.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.290\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHypotension events, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.00 (1.00, 4.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.00 (1.00, 4.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.00 (1.00, 4.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.703\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWBC, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6.40 (5.30, 7.60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6.10 (5.20, 7.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6.90 (5.53, 8.60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLymphocyte, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.80 (1.40, 2.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.80 (1.40, 2.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.75 (1.40, 2.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.420\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNeutrophils, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.75 (2.90, 4.70)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.60 (2.80, 4.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.15 (3.20, 5.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRBC, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.66 (4.34, 4.97)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.64 (4.32, 4.97)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.70 (4.40, 4.98)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.356\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHemoglobin, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14.10 (13.10, 15.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14.10 (13.10, 15.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14.20 (13.20, 15.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.618\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRDW, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13.50 (13.10, 14.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13.40 (13.00, 13.90)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13.80 (13.30, 14.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePlatelet, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e225.00 (189.00, 270.25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e228.00 (191.00, 267.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e219.00 (183.25, 273.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.443\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMPV, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.40 (7.80, 9.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.40 (7.80, 8.90)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.40 (7.90, 9.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.235\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eALT, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e21.00 (16.00, 28.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e22.00 (17.00, 29.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e21.00 (16.00, 26.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.042\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAST, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23.00 (19.00, 27.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23.00 (20.00, 27.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e22.00 (19.00, 27.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.051\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBilirubin Total, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e10.26 (8.55, 13.68)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11.97 (8.55, 13.68)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e10.26 (8.55, 13.68)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.004\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAlbumin, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e42.00 (39.00, 45.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e43.00 (41.00, 46.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e39.00 (38.00, 41.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGGT, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e20.00 (15.00, 31.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e19.00 (14.00, 30.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e22.00 (16.00, 34.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCreatinine, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e74.26 (62.76, 89.28)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e73.37 (61.88, 87.52)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e76.02 (63.65, 90.17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.177\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUric Acid, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e321.20 (267.70, 380.70)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e315.20 (261.70, 368.80)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e333.10 (273.60, 398.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.004\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBlood Urea Nitrogen, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.64 (3.57, 5.71)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.64 (3.57, 5.71)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.64 (3.57, 6.07)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.591\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSodium, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e140.00 (139.00, 141.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e140.00 (139.00, 141.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e140.00 (138.00, 141.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.377\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePhosphorus, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.20 (1.10, 1.32)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.20 (1.10, 1.32)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.20 (1.07, 1.32)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.740\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCalcium Total, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.35 (2.30, 2.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.35 (2.30, 2.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.35 (2.30, 2.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.590\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTriglycerides, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.16 (0.79, 1.73)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.15 (0.78, 1.72)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.20 (0.86, 1.76)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.092\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHdl Cholesterol, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.37 (1.14, 1.66)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.39 (1.15, 1.71)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.29 (1.11, 1.60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.006\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLdl Cholesterol, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.95 (2.35, 3.54)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.95 (2.38, 3.57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.95 (2.33, 3.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.527\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eALI, M (Q₁, Q₃)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e57.80 (40.65, 74.97)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e60.30 (43.47, 80.29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e51.95 (34.62, 67.73)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSex, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.419\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e430 (57.18)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e286 (56.08)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e144 (59.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e322 (42.82)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e224 (43.92)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e98 (40.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSmoking status, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e423 (56.25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e317 (62.16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e106 (43.80)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e329 (43.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e193 (37.84)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e136 (56.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAlcohol intake, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.603\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e536 (71.28)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e360 (70.59)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e176 (72.73)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e216 (28.72)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e150 (29.41)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e66 (27.27)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHypertension, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e373 (49.60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e276 (54.12)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e97 (40.08)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e379 (50.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e234 (45.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e145 (59.92)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDM, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e581 (77.26)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e411 (80.59)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e170 (70.25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e171 (22.74)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e99 (19.41)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e72 (29.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHyperlipidemia, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.274\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e185 (24.60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e132 (25.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e53 (21.90)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e567 (75.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e378 (74.12)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e189 (78.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCOPD, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.166\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e721 (95.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e493 (96.67)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e228 (94.21)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e31 (4.12)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e17 (3.33)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14 (5.79)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCVD, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e673 (89.49)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e467 (91.57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e206 (85.12)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e79 (10.51)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e43 (8.43)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e36 (14.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eASA, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e110 (14.63)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e86 (16.86)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e24 (9.92)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eII\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e242 (32.18)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e176 (34.51)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e66 (27.27)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIII\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e400 (53.19)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e248 (48.63)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e152 (62.81)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTumor T stage, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eT1/T2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e414 (55.05)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e309 (60.59)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e105 (43.39)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eT3/T4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e338 (44.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e201 (39.41)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e137 (56.61)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eExtensive maxillomandibular resection, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e545 (72.47)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e404 (79.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e141 (58.26)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e207 (27.53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e106 (20.78)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e101 (41.74)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBilateral neck dissection, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e597 (79.39)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e442 (86.67)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e155 (64.05)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e155 (20.61)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e68 (13.33)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e87 (35.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFlap size, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSmall flap\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e317 (42.15)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e243 (47.65)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e74 (30.58)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMedium flap\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e329 (43.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e211 (41.37)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e118 (48.76)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLarge flap\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e106 (14.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e56 (10.98)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e50 (20.66)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTransfusion, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.352\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e519 (69.02)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e358 (70.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e161 (66.53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e233 (30.98)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e152 (29.80)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e81 (33.47)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTumor site, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTongue\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e195 (25.93)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e142 (27.84)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e53 (21.90)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFloor of mouth\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e113 (15.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e70 (13.73)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e43 (17.77)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBuccal mucosa\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e140 (18.62)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e111 (21.76)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e29 (11.98)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGingiva\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e89 (11.84)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e65 (12.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e24 (9.92)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePalate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e117 (15.56)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e67 (13.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e50 (20.66)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eOther sites\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e98 (13.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e55 (10.78)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e43 (17.77)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"5\"\u003e\n \u003cp\u003eData are presented as n (%) or median [interquartile range]. \u003cem\u003eP\u003c/em\u003e were derived from the Chi-squared test for categorical variables and the Mann-Whitney U test for continuous variables.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eUnivariate comparisons demonstrated that differences in delayed extubation were predominantly driven by pre-operative and oncologic factors rather than intraoperative events. Specifically, there were no significant differences between the two groups in surgery duration, blood loss, hemodynamic fluctuations, or transfusion rate (all \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05).\u003c/p\u003e\n \u003cp\u003eIn contrast, significant pre-operative disparities were identified. Patients in the delayed extubation group were older (Median: 60.0 vs. 55.0 years, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and had a higher comorbidity burden, including hypertension (59.9% vs. 45.9%, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), diabetes mellitus (29.8% vs. 19.4%, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.002), cardiovascular disease (14.9% vs. 8.4%, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.010), and a higher ASA classification (62.8% vs. 48.6% ASA III, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.001). A history of smoking was also more prevalent in this group (56.2% vs. 37.8%, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e\n \u003cp\u003eThe surgical profile indicated more extensive resections in the delayed extubation group. These patients had higher T-stage tumors (T3/T4: 56.6% vs. 39.4%, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and more frequently required mandibular resection (41.7% vs. 20.8%, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and bilateral neck dissection (36.0% vs. 13.3%, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Accordingly, the need for larger flaps was more common (20.7% vs. 11.0%, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e\n \u003cp\u003eThe pre-operative laboratory profile was markedly different. The delayed extubation group exhibited signs of systemic inflammation and poorer nutritional status, with significantly higher white blood cell count, neutrophil count, and red cell distribution width, alongside lower albumin, and HDL cholesterol (all \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01). Crucially, the ALI was significantly lower in patients facing delayed extubation (51.95 vs. 60.30, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\n \u003ch2\u003e3.2. The Association between ALI and Delayed Extubation Risk\u003c/h2\u003e\n \u003cp\u003eThe association between ALI and delayed extubation was assessed through multivariable logistic regression (Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e). After full adjustment for demographic, comorbidity, surgical, and oncologic confounders (Model 3), ALI remained an independent predictor of delayed extubation. Each 10-point increase in ALI was associated with a 10% reduction in odds (Adjusted OR\u0026thinsp;=\u0026thinsp;0.90, 95% CI: 0.84\u0026ndash;0.95, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001). When analyzed by tertiles, a significant inverse trend was observed (\u003cem\u003ep\u003c/em\u003e for trend\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with patients in the highest tertile (Q3) having a significantly lower risk compared to the lowest tertile (Q1) (Adjusted OR\u0026thinsp;=\u0026thinsp;0.62, 95% CI: 0.40\u0026ndash;0.97, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.035).\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eLogistic regression analysis of the association between ALI and delayed extubation\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003eVariables\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eModel 1\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" rowspan=\"2\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eModel 2\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" rowspan=\"2\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eModel 3\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eOR (95%CI)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eOR (95%CI)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eOR (95%CI)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003eP\u003c/em\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eALI (per 10-point increase)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.87 (0.82\u0026thinsp;~\u0026thinsp;0.92)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.87 (0.82\u0026thinsp;~\u0026thinsp;0.92)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.90 (0.84\u0026thinsp;~\u0026thinsp;0.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eALI (tertile)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eQ1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.00 (Reference)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.00 (Reference)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.00 (Reference)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eQ2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.76 (0.53\u0026thinsp;~\u0026thinsp;1.09)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.141\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.77 (0.53\u0026thinsp;~\u0026thinsp;1.13)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.180\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.83 (0.55\u0026thinsp;~\u0026thinsp;1.26)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.388\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eQ3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.48 (0.32\u0026thinsp;~\u0026thinsp;0.70)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.50 (0.33\u0026thinsp;~\u0026thinsp;0.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.62 (0.40\u0026thinsp;~\u0026thinsp;0.97)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.035\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003eP\u003c/em\u003e for trend\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"9\"\u003e\n \u003cp\u003eOR: Odds Ratio, CI: Confidence Interval\u003c/p\u003e\n \u003cp\u003eModel 1: Crude; Model 2: Adjust: Sex, Smoking status, Alcohol intake, Hypertension, DM, Hyperlipidemia, COPD, CVD, Age; Model 3: Adjust: Sex, Smoking status, Alcohol intake, Hypertension, DM, Hyperlipidemia, COPD, CVD, ASA, Tumor T stage, Extensive maxillomandibular resection, Bilateral neck dissection, Flap size, Age, Tumor site.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eTo elucidate the dose-response relationship, a restricted cubic spline (RCS) analysis was performed with full adjustment for the same covariates. As shown in Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e, the analysis confirmed a linear and inverse association between ALI and the log odds of delayed extubation (\u003cem\u003eP\u003c/em\u003e for nonlinearity\u0026thinsp;=\u0026thinsp;0.277). The overall association was highly significant (\u003cem\u003eP\u003c/em\u003e for overall\u0026thinsp;\u0026lt;\u0026thinsp;0.001), demonstrating a steady decrease in risk across the entire observed range of ALI values.\u003c/p\u003e\n \u003cp\u003eSubgroup analyses were conducted to evaluate the consistency of this association across various patient characteristics (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e). Subgroup analyses showed no evidence of heterogeneity for most covariates. The protective effect of ALI was remarkably consistent in direction across most subgroups. Significant interactions were observed for several factors. The inverse association was significantly stronger among non-smokers (aOR\u0026thinsp;=\u0026thinsp;0.78, 95% CI: 0.69\u0026ndash;0.88) compared to smokers (aOR\u0026thinsp;=\u0026thinsp;0.96, 95% CI: 0.90\u0026ndash;1.03; \u003cem\u003eP\u003c/em\u003e for interaction\u0026thinsp;=\u0026thinsp;0.014), and in patients aged\u0026thinsp;\u0026ge;\u0026thinsp;60 years (aOR\u0026thinsp;=\u0026thinsp;0.81, 95% CI: 0.72\u0026ndash;0.92) compared to those aged\u0026thinsp;\u0026lt;\u0026thinsp;60 years (aOR\u0026thinsp;=\u0026thinsp;0.94, 95% CI: 0.87\u0026ndash;1.02; \u003cem\u003eP\u003c/em\u003e for interaction\u0026thinsp;=\u0026thinsp;0.019). A significant interaction was also noted with COPD status (\u003cem\u003eP\u003c/em\u003e for interaction\u0026thinsp;=\u0026thinsp;0.007) and flap size (\u003cem\u003eP\u003c/em\u003e for interaction\u0026thinsp;=\u0026thinsp;0.026). No significant interactions were found for sex, alcohol intake, or other comorbidities and surgical factors.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\n \u003ch2\u003e3.3. Development and Performance of Machine Learning Models\u003c/h2\u003e\n \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e\n \u003ch2\u003e3.3.1. Data Partition and Balance Assessment\u003c/h2\u003e\n \u003cp\u003eTo develop and validate the machine learning models, the study cohort was randomly divided into a training set (n\u0026thinsp;=\u0026thinsp;527) for model development and a testing set (n\u0026thinsp;=\u0026thinsp;225) for independent validation. The balance between these two sets was assessed by comparing their baseline characteristics (Supplementary Table S2). The two sets were well-balanced, as evidenced by the nearly identical incidence of the primary outcome, delayed extubation (32.3% vs. 32.0%, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.913). The distributions of key predictors, including the ALI, age, comorbidities, and critical indicators of surgical complexity, were all comparable (all \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05). A significant difference was observed only in sex distribution between the sets.\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec15\" class=\"Section3\"\u003e\n \u003ch2\u003e3.3.2 Feature Selection\u003c/h2\u003e\n \u003cp\u003eFeature selection was conducted on the training set using LASSO regression with 10-fold cross-validation. The optimal penalty parameter (\u0026lambda;) was determined as the value that lay within one standard error of the minimum binomial deviance (\u0026lambda;.1se\u0026thinsp;=\u0026thinsp;0.042) (Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\n \u003cp\u003eAt this \u0026lambda; value, twelve predictors demonstrated non-zero coefficients and were selected for the development of all machine learning models. In descending order of the absolute value of their coefficients, they were: ALI, RDW, Bilateral neck dissection, Age; Extensive maxillomandibular resection, WBC; Smoking status, Tumor T stage, Uric acid, ASA, Flap size and Sodium.\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec16\" class=\"Section3\"\u003e\n \u003ch2\u003e3.3.3 Model Evaluation\u003c/h2\u003e\n \u003cp\u003eThe predictive performance of all candidate models was rigorously evaluated on an independent validation set. The comprehensive metrics are detailed in Table \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003ePredictive performance of seven machine learning models in the training and validation sets\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eModel\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eAUC (95% CI)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePrecision\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSensitivity\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eF1 Score\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eKappa\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLogistic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.758 (0.689, 0.821)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.707\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.552\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.444\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.830\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.492\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.289\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.806 (0.739, 0.860)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.773\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.627\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.722\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.797\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.671\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.499\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.875 (0.825, 0.914)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.813\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.759\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.611\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.908\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.677\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.548\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.857 (0.801, 0.905)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.707\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.800\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.111\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.987\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.195\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.127\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLightGBM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.847 (0.791, 0.894)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.800\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.690\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.681\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.856\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.685\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.539\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSVM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.757 (0.687, 0.817)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.716\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.580\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.403\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.863\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.475\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.289\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eANN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.811 (0.746, 0.867)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.773\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.644\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.653\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.830\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.648\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.481\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eIn the validation set, all models achieved AUCs above 0.75 (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eA). The RF model demonstrated the highest discriminative ability, with an AUC of 0.875 (95% CI: 0.825\u0026ndash;0.914) and an accuracy of 0.813. Compared to the Logistic Regression baseline (AUC: 0.758), the RF model also showed the highest precision (0.759) and specificity (0.908). The calibration curve of the RF model closely aligned with the ideal diagonal in the multi-model comparison (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eB), which was supported by a Brier score of 0.141. DCA indicated that the RF model provided higher net clinical benefit than the treat-all and treat-none strategies across most threshold probabilities (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eC).\u003c/p\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003cp\u003eThe selected RF model was further evaluated and demonstrated strong standalone performance, as evidenced by the ROC curve in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eD, the calibration curve in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eE (Brier score: 0.141), and its DCA in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eF.\u003c/p\u003e\n \u003cp\u003eBased on its superior discrimination, calibration, and clinical utility, the RF model was selected as the final predictive model.\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec17\" class=\"Section3\"\u003e\n \u003ch2\u003e3.3.4 Model interpretation\u003c/h2\u003e\n \u003cp\u003eSHAP analysis of the final RF model identified key predictors for delayed extubation and elucidated their effects. The global feature importance is summarized in Fig. \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003eA. Tumor T stage was the most influential predictor, with the highest mean absolute SHAP value of 0.06, followed by Extensive maxillomandibular resection and Bilateral neck dissection (both 0.04). Among the moderately important features (0.02), the inflammatory-nutritional index ALI was notable.\u003c/p\u003e\n \u003cp\u003eThe directional impact of these top features on the prediction is visualized in the beeswarm plot (Fig. \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003eB). This analysis confirmed that the presence of major surgical procedures (e.g., Bilateral neck dissection) and higher values of Tumor T stage and RDW were associated with an increased risk. In contrast, higher ALI values were consistently associated with a reduced risk, reinforcing its role as a protective factor.\u003c/p\u003e\n \u003cp\u003eTo illustrate how these factors converge in an individual prediction, the SHAP waterfall plot for a representative high-risk patient is shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003eC. Here, a low ALI value was the strongest individual driver increasing the predicted risk, followed by elevated RDW and Age.\u003c/p\u003e\n \u003cp\u003eFinally, the SHAP dependence plots in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003eD detailed the nature of these associations. For key continuous variables, a near-linear increase in risk was observed with rising RDW, whereas a dose-dependent protective effect was seen with increasing ALI. Binary surgical factors confirmed a distinct risk elevation when the procedure was performed.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eThis study presents a machine learning\u0026ndash;based framework for predicting delayed extubation (DE) following oral cancer surgery with free flap reconstruction. In a cohort of 752 patients, we observed that both established surgical risk factors and the Advanced Lung Cancer Inflammation Index (ALI) were associated with DE risk. By integrating ALI with clinical and surgical variables into a Random Forest model, we achieved high predictive accuracy (AUC 0.875), and SHAP analysis provided insight into the relative contribution of each factor. These findings suggest a potential avenue for more individualized, data-informed perioperative airway management, moving beyond reliance on subjective assessment alone(\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e). Overall, approximately one-third of patients experienced delayed extubation, underscoring the ongoing clinical challenge of balancing airway safety with minimizing unnecessary prolonged ventilation.\u003c/p\u003e\u003cp\u003eThe associations observed between preoperative patient characteristics and DE are broadly consistent with the epidemiology of perioperative airway vulnerability. Older age, higher ASA classification, and greater comorbidity burden were more frequent among patients requiring prolonged ventilatory support, reflecting potentially reduced physiological reserve, impaired airway reflexes, and baseline cardiorespiratory compromise(\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). In this cohort, intraoperative variables such as operative duration, blood loss, and transfusion requirement did not differ significantly between groups. This lack of variability may reflect the relatively standardized surgical and anesthetic practices at our single center. These findings highlight the importance of thorough preoperative assessment to identify patients who might benefit from closer airway monitoring postoperatively.\u003c/p\u003e\u003cp\u003eAmong the factors examined, ALI demonstrated a consistent inverse relationship with DE risk. After adjustment for demographic, comorbidity, surgical, and oncologic confounders, each 10-point increase in ALI was associated with approximately 10% lower odds of delayed extubation. Dose\u0026ndash;response analysis indicated a linear trend, and subgroup analyses suggested that the protective association of ALI was generally consistent across age, smoking status, and comorbidity subgroups, with somewhat stronger effects observed among older adults and non-smokers. These findings are consistent with previous studies showing that lower ALI, reflecting systemic inflammation and poor nutritional status, is associated with higher postoperative complication rates and worse outcomes in both oncologic and non-oncologic surgical populations(\u003cspan additionalcitationids=\"CR24 CR25\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e). While causality cannot be inferred, it is plausible that patients with higher ALI experience less postoperative edema and more rapid respiratory muscle recovery, facilitating timely extubation.\u003c/p\u003e\u003cp\u003eThe components of ALI\u0026mdash;body mass index, albumin, and neutrophil-to-lymphocyte ratio\u0026mdash;have been individually implicated in perioperative outcomes(\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e). Low albumin has been associated with impaired wound healing and reduced respiratory muscle strength(\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e), while elevated NLR reflects systemic inflammation that can exacerbate tissue edema(\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e). The present findings suggest that combining these markers into a single index may provide a more holistic assessment of patient vulnerability, offering additional information beyond traditional demographic or comorbidity measures. Clinically, preoperative evaluation of ALI could potentially inform risk stratification and guide perioperative optimization strategies, such as nutritional support or targeted anti-inflammatory interventions, though further prospective studies would be needed to confirm such applications.\u003c/p\u003e\u003cp\u003eSurgical factors, particularly extensive maxillomandibular resection and bilateral neck dissection, were strongly associated with DE(\u003cspan additionalcitationids=\"CR32\" citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e). These procedures can alter airway geometry, disrupt fascial planes, and increase postoperative soft tissue swelling, all of which may compromise airway patency. SHAP analysis in the Random Forest model highlighted these factors as dominant predictors, consistent with clinical experience. Tumor T stage and flap size also contributed meaningfully to risk prediction, underscoring the impact of surgical complexity on postoperative airway management. Taken together, these results reinforce the need for close coordination between surgical and anesthesia teams when planning extubation strategies, particularly in patients undergoing extensive resections or bilateral neck procedures.\u003c/p\u003e\u003cp\u003eInflammatory hematologic markers, including RDW and white blood cell count, were independently associated with the risk of delayed extubation. A higher RDW may indicate disturbances in erythrocyte homeostasis and systemic inflammation, which could compromise oxygen delivery and contribute to postoperative tissue hypoxia. Elevated WBC similarly suggests ongoing inflammatory stress. These observations are consistent with findings in other surgical populations, where elevated preoperative RDW or WBC have been linked to greater postoperative morbidity and mortality, including increased risk of complications, organ dysfunction, and prolonged recovery periods(\u003cspan additionalcitationids=\"CR35\" citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e). Routine preoperative hematologic markers may therefore offer useful complementary information when combined with clinical and surgical data to improve risk stratification.\u003c/p\u003e\u003cp\u003eThe machine learning framework developed in this study offers several advantages over conventional regression approaches. The Random Forest model was able to capture nonlinear interactions between predictors, such as how a moderately low ALI may have a greater impact in older patients undergoing complex surgery. SHAP analysis provided interpretable visualizations of the contributions of each factor, addressing the common criticism of black-box models and enhancing potential clinical utility. Compared with prior perioperative models that often rely solely on demographic or anatomical predictors, our approach incorporated both inflammatory-nutritional indices and surgical complexity, representing a more comprehensive risk assessment. While promising, the model should be regarded as a decision-support tool rather than a deterministic predictor, and clinical judgment remains essential.\u003c/p\u003e\u003cp\u003eSeveral limitations warrant consideration. First, the retrospective, single-center design may limit generalizability, as institutional practices and surgical techniques vary. Second, some potentially relevant variables, including intraoperative fluid balance, detailed flap design, and dynamic postoperative airway assessments, were not available. Third, ALI and other laboratory markers were assessed preoperatively at a single time point, precluding evaluation of perioperative changes that may influence airway outcomes. Finally, while internal validation demonstrated robust model performance, external validation in independent cohorts is necessary before broad clinical adoption.\u003c/p\u003e"},{"header":"5. Conclusion","content":"\u003cp\u003eIn summary, this study highlights the multifactorial nature of delayed extubation following oral cancer surgery with free flap reconstruction, with contributions from preoperative inflammatory-nutritional status, surgical complexity, and systemic physiologic markers. ALI emerged as a promising, independent marker of protective effect, and the Random Forest model, complemented by SHAP interpretability, offers a nuanced approach to perioperative risk assessment. These findings may inform individualized airway management strategies and encourage further exploration of inflammatory-nutritional optimization as part of preoperative preparation. Future research should aim to validate these results across multiple centers, evaluate dynamic perioperative changes in ALI, and investigate whether interventions targeting nutritional and inflammatory status can meaningfully influence airway outcomes.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eDE\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eDelayed Extubation\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eALI\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eAdvanced Lung Cancer Inflammation Index\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eICU\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eIntensive Care Unit\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eML\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eMachine Learning\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\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eRF\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eRandom Forest\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eNLR\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eNeutrophil-to-Lymphocyte Ratio\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003ePLR\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003ePlatelet-to-Lymphocyte Ratio\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eBMI\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eBody Mass Index\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eASA\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eAmerican Society of Anesthesiologists\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eRDW\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eRed Cell Distribution Width\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\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\"\u003eRCS\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eRestricted Cubic Spline\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eANN\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eArtificial Neural Network\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\"\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\"\u003eLightGBM\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eLight Gradient Boosting Machine\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eOR\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eOdds Ratio\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eCI\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eConfidence Interval\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eIQR\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eInterquartile Range.\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\u003eThis retrospective study was approved by the Biomedical Research Ethics Committee of Suining Central Hospital (Approval No. KYLLKYS20250144). The requirement for informed consent was waived due to the retrospective nature of the study and anonymized data handling. All procedures were conducted in accordance with the principles of the Declaration of Helsinki.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated and/or analyzed during the current study are not publicly available due to patient confidentiality and privacy restrictions but are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research was funded by the Sichuan Provincial Department of Science and Technology [grant number 2022SNZY001]; the Sichuan Provincial Health and Health Commission [grant number 2022JDXM021]; and the Suining Municipal Health and Science Technology Program – Guiding Project [grant number 25ZDJB08].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eB.J. and S.Y. conceived and designed the study. B.J. and X.W. performed the data analysis and wrote the first draft of the manuscript. Q.W., H.Z., and J.R. contributed to data collection and interpretation. G.L. and X.Z. provided administrative and technical support. C.Y. prepared the figures and tables. X.W. and S.Y. critically revised the manuscript. All authors read and approved the final version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe appreciate the assistance in data collection, statistical analysis, and manuscript preparation, which contributed to the completion of this study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eLi MM, Miller LE, Old M. State of Head and Neck Microvascular Reconstruction: Current and Future Directions. Surg Oncol Clin N Am. 2024;33(4):711-21.\u003c/li\u003e\n\u003cli\u003eMyatra SN, Kornerup SC, Parotto M. Extubation in head and neck surgery. Curr Opin Anaesthesiol. 2025;38(6):869-76.\u003c/li\u003e\n\u003cli\u003eRiekert M, Rempel V, Keilwerth S, Z\u0026ouml;ller JE, Kreppel M, Schick VC. Airway-Associated Complications With and Without Primary Tracheotomy in Oral Squamous Cell Carcinoma Surgery. J Craniofac Surg. 2023;34(1):279-83.\u003c/li\u003e\n\u003cli\u003eMadgar O, Livneh N, Dobriyan A, Dagan E, Alon EE. Airway management following head and neck microvascular reconstruction: is tracheostomy mandatory? Braz J Otorhinolaryngol. 2022;88 Suppl 4(Suppl 4):S44-s9.\u003c/li\u003e\n\u003cli\u003eCoyle MJ, Tyrrell R, Godden A, Hughes CW, Perkins C, Thomas S, Godden D. Replacing tracheostomy with overnight intubation to manage the airway in head and neck oncology patients: towards an improved recovery. Br J Oral Maxillofac Surg. 2013;51(6):493-6.\u003c/li\u003e\n\u003cli\u003eLee HJ, Kim JW, Choi SY, Kim CS, Kwon TG, Paeng JY. The evaluation of a scoring system in airway management after oral cancer surgery. Maxillofac Plast Reconstr Surg. 2015;37(1):19.\u003c/li\u003e\n\u003cli\u003eSchmutz A, Dieterich R, Kalbhenn J, Voss P, Loop T, Heinrich S. Protocol based evaluation for feasibility of extubation compared to clinical scoring systems after major oral cancer surgery safely reduces the need for tracheostomy: a retrospective cohort study. BMC Anesthesiol. 2018;18(1):43.\u003c/li\u003e\n\u003cli\u003eJanik S, Brkic FF, Grasl S, K\u0026ouml;nigswieser M, Franz P, Erovic BM. Tracheostomy in bilateral neck dissection: Comparison of three tracheostomy scoring systems. Laryngoscope. 2020;130(11):E580-e6.\u003c/li\u003e\n\u003cli\u003eKwon MA, Song J, Kim S, Oh PW, Kang M. Airway Management Failure after Delayed Extubation in a Patient with Oral Malignant Melanoma Who Underwent Partial Mandibulectomy and Reconstruction with a Free Flap. Case Rep Dent. 2021;2021:7792843.\u003c/li\u003e\n\u003cli\u003eMossinelli C, Pietrobon G, Zorzi S, Tagliabue M, Chu F, Tomarchio E, et al. Airway assessment and management in head and neck cancer surgery. Acta Otorhinolaryngol Ital. 2025;45(3):173-81.\u003c/li\u003e\n\u003cli\u003eXu S, Wang K, Liu K, Liu Y, Huang Y, Zhang Y, et al. Predictive Nomogram for the Necessity of Tracheotomy During Oral and Oropharyngeal Cancer Surgery. Laryngoscope. 2021;131(5):E1489-e95.\u003c/li\u003e\n\u003cli\u003eLiu F, Liu Z, Han Z, Cao M, Liufu N, Fu G. Development and Validation of a Nomogram Incorporating Changes in Inflammatory Markers for Predicting Complications After Free Flap Reconstruction of Oral and Maxillofacial Defects. J Craniofac Surg. 2025;36(6):e727-e31.\u003c/li\u003e\n\u003cli\u003eLiu Q, Zhou Y, Cao X, Wang W, Pan C, YichenXu, et al. The impact of systemic inflammation index on prolonged mechanical ventilation after cardiac surgery: a retrospective study. J Cardiothorac Surg. 2025;20(1):293.\u003c/li\u003e\n\u003cli\u003eChen Y, Wu G, Wang R, Chen J. Preoperative Albumin Level Serves as a Predictor for Postoperative Pulmonary Complications Following Elective Laparoscopic Gastrectomy. Curr Pharm Des. 2018;24(27):3250-5.\u003c/li\u003e\n\u003cli\u003eJafri SH, Shi R, Mills G. Advance lung cancer inflammation index (ALI) at diagnosis is a prognostic marker in patients with metastatic non-small cell lung cancer (NSCLC): a retrospective review. BMC Cancer. 2013;13:158.\u003c/li\u003e\n\u003cli\u003eDeo RC. Machine Learning in Medicine. Circulation. 2015;132(20):1920-30.\u003c/li\u003e\n\u003cli\u003eMoharrami M, Azimian Zavareh P, Watson E, Singhal S, Johnson AEW, Hosni A, et al. Prognosing post-treatment outcomes of head and neck cancer using structured data and machine learning: A systematic review. PLoS One. 2024;19(7):e0307531.\u003c/li\u003e\n\u003cli\u003eAsaad M, Lu SC, Hassan AM, Kambhampati P, Mitchell D, Chang EI, et al. The Use of Machine Learning for Predicting Complications of Free-Flap Head and Neck Reconstruction. Ann Surg Oncol. 2023;30(4):2343-52.\u003c/li\u003e\n\u003cli\u003eSingh R, Lanchantin J, Sekhon A, Qi Y. Attend and Predict: Understanding Gene Regulation by Selective Attention on Chromatin. Adv Neural Inf Process Syst. 2017;30:6785-95.\u003c/li\u003e\n\u003cli\u003eSingh T, Sankla P, Smith G. Tracheostomy or delayed extubation after maxillofacial free-flap reconstruction? Br J Oral Maxillofac Surg. 2016;54(8):878-82.\u003c/li\u003e\n\u003cli\u003eBoruk M, Chernobilsky B, Rosenfeld RM, Har-El G. Age as a prognostic factor for complications of major head and neck surgery. Arch Otolaryngol Head Neck Surg. 2005;131(7):605-9.\u003c/li\u003e\n\u003cli\u003eWu L, Li Q, Zhu Z, Li S, Zhu Z, Zhang T. Impact of preoperative comorbidities on postoperative complication rates and survival outcome in patients with head and neck cancer undergoing surgical treatment. Sci Rep. 2025;15(1):38746.\u003c/li\u003e\n\u003cli\u003eKobayashi S, Karube Y, Inoue T, Araki O, Maeda S, Matsumura Y, Chida M. Advanced Lung Cancer Inflammation Index Predicts Outcomes of Patients with Pathological Stage IA Lung Adenocarcinoma Following Surgical Resection. Ann Thorac Cardiovasc Surg. 2019;25(2):87-94.\u003c/li\u003e\n\u003cli\u003eGu Y, Zhang C, Miao JB, Wang H, Hu B, Li X. The association of the advanced lung cancer inflammation index with postoperative complications in patients undergoing lung resection for bronchiectasis. J Thorac Dis. 2025;17(6):3577-89.\u003c/li\u003e\n\u003cli\u003eHua X, Chen J, Wu Y, Sha J, Han S, Zhu X. Prognostic role of the advanced lung cancer inflammation index in cancer patients: a meta-analysis. World J Surg Oncol. 2019;17(1):177.\u003c/li\u003e\n\u003cli\u003eYin C, Toiyama Y, Okugawa Y, Omura Y, Kusunoki Y, Kusunoki K, et al. Clinical significance of advanced lung cancer inflammation index, a nutritional and inflammation index, in gastric cancer patients after surgical resection: A propensity score matching analysis. Clin Nutr. 2021;40(3):1130-6.\u003c/li\u003e\n\u003cli\u003eQiu X, Shen S, Jiang N, Feng Y, Yang G, Lu D. Association of advanced lung cancer inflammation index with all-cause and cardiovascular mortality in metabolic dysfunction associated steatotic liver disease. Sci Rep. 2025;15(1):15121.\u003c/li\u003e\n\u003cli\u003eZhang Y, Tan S, Wu G. ESPEN practical guideline: Clinical nutrition in surgery. Clin Nutr. 2021;40(9):5071.\u003c/li\u003e\n\u003cli\u003eZheng M, Zhang X, Wang H, Yuan P, Yu Q. Interpretable machine learning model for identification and risk factor of premature rupture of membranes (PROM) and its association with nutritional inflammatory index: a retrospective study. Front Med (Lausanne). 2025;12:1557919.\u003c/li\u003e\n\u003cli\u003eFeier CVI, Motoc A, Muntean C, Vonica RC, Gaborean V, Olariu S, Murariu MS. Systemic inflammatory indices and age-dependent severity in acute appendicitis: a retrospective cohort study. Front Immunol. 2025;16:1620459.\u003c/li\u003e\n\u003cli\u003eHasegawa T, Yatagai N, Furukawa T, Wakui E, Saito I, Takeda D, et al. The prospective evaluation and risk factors of dysphagia after surgery in patients with oral cancer. J Otolaryngol Head Neck Surg. 2021;50(1):4.\u003c/li\u003e\n\u003cli\u003eOu M, Wang G, Yan Y, Chen H, Xu X. Perioperative symptom burden and its influencing factors in patients with oral cancer: A longitudinal study. Asia Pac J Oncol Nurs. 2022;9(8):100073.\u003c/li\u003e\n\u003cli\u003eKuo PJ, Lin PC, Hsieh CH. Airway Management Following Head and Neck Microvascular Reconstruction: When is a Tracheostomy Necessary? Risk Manag Healthc Policy. 2025;18:2551-63.\u003c/li\u003e\n\u003cli\u003eWei W, Feng B, Chen Z, Liu X, Xiao M, Hu H. Association of preoperative red blood cell width and postoperative 30-day mortality in patients undergoing non-cardiac surgery: a retrospective cohort study using propensity-score matching. Perioper Med (Lond). 2024;13(1):95.\u003c/li\u003e\n\u003cli\u003eFrentiu AA, Mao K, Caruana CB, Raveendran D, Perry LA, Penny-Dimri JC, et al. The Prognostic Significance of Red Cell Distribution Width in Cardiac Surgery: A Systematic Review and Meta-Analysis. J Cardiothorac Vasc Anesth. 2023;37(3):471-9.\u003c/li\u003e\n\u003cli\u003evan Koeverden ID, den Ruijter HM, Scholtes VPW, M GEHL, Haitjema S, Buijsrogge MP, et al. A single preoperative blood test predicts postoperative sepsis and pneumonia after coronary bypass or open aneurysm surgery. Eur J Clin Invest. 2019;49(3):e13055.\u003c/li\u003e\n\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":"Advanced Lung Cancer Inflammation Index, Delayed Extubation, Oral Cancer, Free Flap Reconstruction, Machine Learning","lastPublishedDoi":"10.21203/rs.3.rs-8243006/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8243006/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e\u003cp\u003eDelayed extubation (DE) is a common postoperative challenge after oral cancer resection with free flap reconstruction. Existing risk assessments largely focus on anatomical and surgical factors, with limited consideration of patients\u0026rsquo; systemic physiological status. The Advanced Lung Cancer Inflammation Index (ALI), a preoperative composite measure of nutritional and inflammatory status, has shown prognostic value in oncology but its role in perioperative airway management is unclear.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eWe retrospectively analyzed 752 patients undergoing oral cancer resection with free flap reconstruction at a single center. Associations between preoperative ALI and DE were evaluated using multivariable logistic regression. A random forest model integrating ALI with clinical and surgical factors was developed and interpreted using SHapley Additive exPlanations (SHAP) to identify key predictors.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eDE occurred in 32.2% of patients. Higher preoperative ALI was independently associated with lower risk of DE (adjusted OR per 10-point increase\u0026thinsp;=\u0026thinsp;0.90, 95% CI: 0.84\u0026ndash;0.95, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The random forest model achieved an AUC of 0.875 in the validation cohort and demonstrated good calibration. SHAP analysis revealed tumor T stage, extent of resection, bilateral neck dissection, and ALI as the most influential predictors, with higher ALI consistently protective.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e\u003cp\u003ePreoperative ALI is an independent predictor of delayed extubation. An interpretable machine learning model combining ALI with clinical and surgical variables provides a high-performing tool for individualized perioperative airway risk assessment, supporting tailored extubation strategies and postoperative management in oral cancer patients.\u003c/p\u003e","manuscriptTitle":"Preoperative Nutrition–Inflammation Status and Surgical Factors Predict Delayed Extubation After Oral Cancer Surgery with Free Flap Reconstruction: A Machine Learning Approach","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-12 08:31:53","doi":"10.21203/rs.3.rs-8243006/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":"57239b1e-a4b4-443c-ba77-a4e6b00ff54a","owner":[],"postedDate":"December 12th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-15T10:33:33+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-12 08:31:53","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8243006","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8243006","identity":"rs-8243006","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}