Development and external validation of interpretable machine learning models for personalized multiple treatment recommendations in non-small cell lung cancer.

Development and validation of an interpretable machine learning model for cerebral small vessel disease risk assessment.

ObjectiveDistant organ metastasis remains the primary factor affecting long-term survival following radical surgery for colorectal cancer (CRC). This study aimed to develop and validate an interpretable machine learning (ML) model to predict the 5-year cumulative risk of distant metastasis after radical CRC surgery.MethodsA retrospective observational cohort study was conducted using clinical and follow-up data from 341 CRC patients who underwent radical surgery. The cohort was randomly divided into a training set n=239 and a validation set n=102 at a 7:3 ratio. Feature selection was performed using least absolute shrinkage and selection operator (LASSO) regression, identifying variables associated with the 5-year cumulative occurrence of metastasis. Prediction models were constructed using seven algorithms. Model performance was evaluated through multiple metrics: area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity, F1 score, calibration plots, and decision curve analysis. The SHapley Additive exPlanations (SHAP) method was applied to improve model interpretability.ResultsLASSO combined with tenfold cross-validation selected 11 key features for model development. Among the models tested, the SVM model demonstrated superior performance, achieving a Brier score of 0.144 and an AUC of 0.865 in the validation set. Calibration and clinical decision curves confirmed the SVM model’s strong calibration and clinical applicability. The SHAP dependence plots and force analysis provided explanations at both feature and individual patient levels for the model’s 5-year risk predictions.ConclusionThis study established a high-accuracy and interpretable ML model capable of effectively predicting the 5-year cumulative risk of distant organ metastasis after radical colorectal cancer surgery, while further external validation is necessary to confirm its clinical utility.

To develop an interpretable machine learning (ML) model for predicting surgical outcomes in renal hilar tumors and propose a hilar-specific anatomical nephrometry scoring system. A total of 414 patients with renal hilar tumors who underwent robot-assisted partial nephrectomy (RAPN) were included in this study, comprising 304 patients from the First Affiliated Hospital of Nanchang University and 110 patients from the Second Affiliated Hospital of Nanchang University, which served as the external validation cohort. To identify predictors of trifecta achievement, we used least absolute shrinkage and selection operator regression and the Boruta algorithm, followed by multivariate logistic regression to identify independent factors. Five ML models were developed and evaluated using receiver operating characteristic curves, calibration plots, decision curve analysis, and precision-recall curves. The generalizability of the model was further validated in external cohort. Finally, we used SHapley Additive exPlanations (SHAP) to interpret the contribution of each predictor and enhance the model's explainability. Furthermore, based on anatomical features identified through logistic regression, we developed a modified nephrometry scoring system and compared its risk stratification performance with the traditional R.E.N.A.L. (i.e., Radius, Exophytic or endophytic, Nearness, Anterior or posterior, and Location) scoring system. Among the 304 patients in the primary cohort, 168 achieved trifecta outcomes. Eight variables were incorporated into the predictive model, with logistic regression model ultimately being selected as the optimal predictive model. It showed robust predictive performance in internal and external validation. SHAP methods identified surgeon, classification of hilar tumor, and radius as the three most significant predictive variables. Compared with the traditional R.E.N.A.L. score, the modified R.E.N.A.L. score demonstrated superior stratification ability for operation time, change in serum creatinine, change in estimated glomerular filtration rate, and trifecta achievement. The interpretable ML model accurately predicts trifecta in RAPN for hilar tumors. The modified R.E.N.A.L. score provides refined anatomical stratification and facilitates individualized surgical planning.

The applicability of four major traditional in-hospital mortality models in the Chinese setting is unclear due to disease spectrum and population heterogeneity. This study aimed to test the performance of these models in the Chinese setting and to construct and externally validate a novel model. 21,855 consecutive pediatric patients who underwent congenital heart surgery from January 2015 to December 2021 in Shanghai Children's Medical Center were enrolled. For external validation, we additionally pooled 5,221 consecutive pediatric patients who underwent this surgical treatment from January 2020 to December 2021 in Beijing Fuwai Hospital. The performance of Aristotle Basis Complexity (ABC) score, Risk Adjustment for Congenital Heart Surgery (RACHS)-1 categories, Society of Thoracic Surgeons (STS)-European Association for Cardiothoracic Surgery (STAT) score, and STAT categories was tested. Independent predictors were used to develop a model. The area under the receiver operating characteristic curves (AUROCs) and Brier score were used to examine the model performance. The AUROCs were 0.778 for ABC score, 0.685 for RACHS-1 categories, 0.808 for STAT score, and 0.784 for STAT categories. When preoperative covariates were added to four models, the AUROCs improved: ABC score (AUROC=0.860), RACHS-1 categories (AUROC=0.844), STAT score (AUROC= 0.856), and STAT categories (AUROC=0.864). The best-performing model incorporated 6 variables, including age, height, oxygen support, previous cardiac operation, emergency surgery, and STAT categories. The AUROCs and Brier score were 0.864 and 0.00977 in the development cohort, and 0.860 and 0.00654 in the external validation cohort. Four major traditional models were only moderately effective in predicting in-hospital mortality after congenital heart surgery in the Chinese setting. The novel model founded on the STAT categories in combination with preoperative covariates can serve as a useful and effective tool for predicting the risk of in-hospital mortality after congenital heart surgery in the Chinese setting.

ObjectivePostoperative complications remain a major concern in laryngeal cancer surgery, often requiring invasive interventions or intensive care. This study aimed to develop and validate an interpretable machine learning (ML) model to preoperatively predict Clavien-Dindo Grade ≥ III complications and support risk-informed perioperative decision-making.MethodsWe conducted a retrospective study using a temporally split cohort of laryngeal cancer patients. Postoperative complications were graded using the Clavien-Dindo (CD) classification. Eight ML algorithms were trained and evaluated using receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis (DCA). Model interpretability was assessed using SHapley Additive exPlanations (SHAP). A web-based calculator was deployed for clinical use.ResultThe random forest (RF) model achieved the best performance, with an area under the curve (AUC) of 0.935 in the training set and 0.842 in the test set. The model demonstrated robust sensitivity and specificity for both surgical and medical complications. Calibration curves indicated strong agreement between predicted and actual outcomes. SHAP analysis identified eight key predictors—such as vocal cord mobility, tumor subsite, and nutritional status—that contributed most to risk estimation. A user-friendly web calculator was developed and is accessible at: https://qilushiny.shinyapps.io/qilupredicate/.ConclusionWe developed a clinically interpretable ML model that accurately predicts major postoperative complications in patients undergoing laryngeal cancer surgery. This tool provides individualized risk assessments that can guide surgical planning, optimize perioperative strategies, and enhance shared decision-making. Prospective multicenter validation is needed to confirm its utility in routine practice.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12893-025-03213-z.

Pediatric adrenocortical tumors (pACTs) are rare and clinically heterogeneous. Existing risk stratification systems rely on fixed thresholds and linear assumptions, which may limit their prognostic accuracy-particularly for nonmetastatic, locally advanced cases. We aimed to develop an interpretable machine learning (ML) model for individualized survival prediction using only routine clinical features. We retrospectively analyzed 97 patients with pACT from the German Pediatric Oncology Hematology-Malignant Endocrine Tumors Registry (1997-2024). An Extreme Gradient Boosting Cox proportional hazards model was trained using 4 features-tumor volume, distant metastases, pathologic T stage, and resection status-identified via systematic feature evaluation across 11 737 model combinations. Performance was assessed using a stratified 80/20 train-test split, 500 bootstrap iterations, and Harrell's concordance index (C-index). SHapley Additive exPlanations (SHAP) were used for interpretability. The model achieved strong prognostic performance (test-set C-index: 0.925; bootstrap mean: 0.891, 95% confidence interval: 0.817-0.946). SHAP analysis confirmed the dominant influence of metastatic status, followed by tumor volume, T stage, and resection status. The model uncovered nonlinear and additive effects, including a SHAP- and bootstrap-guided tumor volume cut-off (190 mL, 95% confidence interval 127-910 mL) that only slightly differed from conventional thresholds. Stratification remained robust in subgroups, including nonmetastatic patients with advanced local disease. This interpretable ML model enables individualized survival prediction in pACT using only routine clinical data. It offers a clinically accessible and clinically meaningful complement to existing scoring systems, particularly in patients with ambiguous risk profiles who may benefit from more personalized management.

Interpretable vs. noninterpretable machine learning models for data-driven hydro-climatological process modeling

BackgroundAccurate individual risk assessment is crucial for guiding and improving the prevention of atherosclerotic cardiovascular disease (ASCVD). Existing prediction models are primarily derived from Western Caucasian and Chinese Han populations. Our objective is to develop and validate an interpretable machine learning (ML) model based on biomarkers for predicting coronary artery disease (CAD) risk among multi-ethnic patients in Xinjiang, China.MethodsThis retrospective cohort study enrolled patients who underwent coronary angiography or coronary computed tomography angiography (CCTA) at the First Affiliated Hospital of Xinjiang Medical University. The cohort was divided into training, validation, and test sets. Feature selection was performed using logistic regression and LASSO, followed by prediction model development with six machine learning algorithms (XGBoost, RF, MLP, SVM, KNN, AdaBoost). Predictive performance was evaluated using the area under the receiver operating characteristic curve (AUROC) as the primary metric to identify the optimal algorithm. The selected algorithm was further validated on both the validation and testing sets. Shapley Additive Explanations (SHAP) were applied to quantify each feature’s contribution to CAD risk prediction, generating individualized risk explanations. Furthermore, the model's calibration was assessed using calibration curves and the Brier score. Its clinical utility was evaluated through decision curve analysis and was benchmarked against the established SCORE2 Asia Pacific risk model.ResultsThis study enrolled 7655 male and 4461 female participants, divided into training, validation, and test sets in a 7:1.5:1.5 ratio. XGBoost demonstrated optimal performance in both cohorts: the male model achieved AUROCs of 0.845 (95% CI: 0.834–0.855), 0.814 (0.789–0.839), and 0.826 (0.802–0.850) in the training, validation, and test sets, respectively, while the female model attained values of 0.817 (0.802–0.832), 0.759 (0.721–0.796), and 0.786 (0.751–0.821). Male CAD risk was significantly associated with advanced age, multiple abnormal clinical indicators (elevated creatinine, total cholesterol, lipoprotein(a), etc., and decreased HDL-C), hypertension, and diabetes, with higher risk observed in Kazakh and Hui ethnicities, whereas higher education and married status served as protective factors. In females, hypertension was the strongest predictor, while elevated uric acid, systolic blood pressure, fasting blood glucose, along with histories of hypertension and diabetes increased risk; married status and higher education similarly exhibited protective effects. The prediction model demonstrated favorable clinical utility and accuracy in both cohorts, with calibration significantly enhancing predictive performance. Compared to the SCORE2 Asia Pacific risk model, our model exhibited superior discriminatory ability (male: 0.826 vs. 0.662; female: 0.786 vs. 0.720) and improved calibration.ConclusionsMachine learning models can provide personalized and highly accurate predictions of CAD risk. The interpretability of these models facilitates the identification of modifiable risk factors in individual patients, offering valuable insights to enhance primary prevention and management of cardiovascular disease in the Xinjiang region.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40001-025-03763-1.

XAI is critical for establishing trust and enabling the appropriate development of machine learning models. By offering transparency into how these models make judgements, XAI enables researchers and users to uncover potential biases, admit limits, and eventually enhance the fairness and dependability of AI systems. In this paper, we demonstrates two techniques, LIME and SHAP, used to improve the interpretability of machine learning models. Assessing Explainable AI (XAI) approaches is critical in searching for transparent and interpretable artificial intelligence (AI) models. Explainable AI (XAI) approaches are designed to provide insight into how complex models make decisions. This paper thoroughly analyzes two prominent XAI methods: Shapley Additive explanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME). This study aims to understand the decision made by a machine learning model and how the model came to that decision. We discuss the approaches and framework of both LIME and SHAP and assess their behavior in predicting the model's outcome.

BackgroundSurgery is an important approach to treat non-small cell lung cancer (NSCLC) brain metastases (BM). Here, we analyzed the survival outcome and prognostic factors for patients with NSCLC after BM resection.MethodsThe Surveillance Epidemiology and End Results (SEER) database was employed to address the incidence of BM from NSCLC and the current prognosis at population level. 674 contemporaneous NSCLC patients received BM resection at Sun Yat-sen University Cancer Center (SYSUCC) were used for survival comparison and Cox proportional hazards model was applied for identifying prognostic factors.Results60,436 NSCLC patients diagnosed between 2010 to 2017 were enrolled from SEER database. Among them, 8,708 (14.4%) BM were identified at primary NSCLC diagnosis (synchronous BM, SBM). Median overall survival (OS) of SBM was 6 months with 1-, and 3-year survival percentages of 30.3% and 9.8%, respectively. Furthermore, the survival of BM patients without extracranial metastasis is significantly longer than those with extracranial metastases (median OS: 10 versus 5 months, P<0.001). 225 SBM (cohort A) and 449 BM with treatment history on primary NSCLC (cohort B) were collected from SYSUCC. In cohort A, 86 BM with extracranial metastases were found (38.2%) and the median OS was significantly shorter than those without extracranial metastases (15.2 versus 23.7 months, P<0.001). In cohort B, 255 cases with extracranial metastases were found (56.8%) and their prognosis was also worse than cases without extracranial metastases (median OS: 18.3 versus 22.1 months, P=0.002). Multivariate analyses revealed that younger age (HR=0.71, P=0.003), without extracranial metastases (HR=0.65, P<0.001) and radiation for BM (HR=0.78, P=0.005) were independent factors for better OS.ConclusionImproved survival of patients received BM resection was observed in SYSUCC cohort as comparison with SEER patients with NSCLC and BM. Aggressive local treatment including surgery and radiation is still important in Modern management of BM from NSCLC.

Non-small cell lung cancer (NSCLC) patients with central airway obstruction (CAO) may have better survival on systemic therapy if the airway patency is successfully restored by bronchoscopic interventions. It remains unclear which therapeutic bronchoscopic modality [laser, stenting, external beam radiation, brachytherapy and photodynamic therapy (PDT)] used for restoring airway patency positively affects outcomes in these patients. We analyzed the effectiveness of PDT in terms of mortality, and time to subsequent treatments in patients with stage III and IV NSCLC. Study used Surveillance, Epidemiology, and End Results (SEER) Medicare linked data. We categorized NSCLC patients diagnosed between 2000 and 2011 and with stage III and IV, into three treatment groups: PDT + radiation ± chemotherapy, non-PDT ablation therapy + radiation ± chemotherapy, and radiation + chemotherapy. We analyzed all-cause and cause-specific mortality using Cox proportional hazard models with an inverse probability weighted propensity score adjustment. Time to subsequent treatment was analyzed using GLM model. For the PDT group, hazard for all-cause and cause-specific mortality was comparable to the radiation + chemotherapy group (HR =1.03, 95% CI: 0.73-1.45; and HR =1.04, 95% CI: 0.71-1.51, respectively). The non-PDT ablation group had higher hazard for all-cause (HR =1.22, 95% CI: 1.13-1.33) and cause-specific mortality (HR =1.10, 95% CI: 1.01-1.20), compared to the radiation + chemotherapy group. The PDT group had longer time to follow-up treatment, compared to non-PDT ablation group. In our exploratory study of stage III and IV NSCLC patients with CAO, addition of PDT demonstrated hazard of mortality comparable to radiation + chemotherapy group. However, addition of non-PDT ablation showed higher mortality compared to the radiation + chemotherapy group. Future studies should investigate the efficacy and effectiveness of multimodal therapy including radiation, chemo, immunotherapy and bronchoscopic interventions.

Repeat transurethral resection (ReTUR) is essential for reducing residual and recurrent non-muscle-invasive bladder cancer (NMIBC). Pathological upstaging after ReTUR significantly influences prognosis. This study aimed to develop an interpretable machine learning model using CT radiomics to predict the risk of pathological upstaging following ReTUR in NMIBC patients. We retrospectively analyzed 104 NMIBC patients who underwent ReTUR at the Second Affiliated Hospital of Nanchang University from March 2019 to July 2022. Data were split 7:3 into training and internal validation sets. An external validation set included 40 patients from two other hospitals. Radiomic features were extracted from preoperative CT scans. Least Absolute Shrinkage and Selection Operator (LASSO) and multivariate logistic regression were used to identify predictors of pathological upstaging. Four machine learning models, including Extreme Gradient Boosting (XGBoost), Gradient Boosting Decision Tree (GBDT), Random Forest (RF), and Linear Discriminant Analysis (LDA), were constructed and evaluated using AUC, accuracy, precision, F1 score, calibration curves, and decision curve analysis (DCA). The best model was interpreted via SHapley Additive exPlanations (SHAP) to identify key predictive features. umor grade (OR = 7.02, 95% CI: 1.17-42.21), tumor size (OR = 5.83, 95% CI: 1.21-28.15), and tumor number (OR = 6.83, 95% CI: 1.18-39.52) were independent risk factors. From 4,738 radiomic features, nine were selected. The XGBoost model outperformed others, with an AUC of 0.804 (95% CI: 0.756-0.862), accuracy of 77.4%, precision of 82.7%, and F1 score of 0.701 in internal validation. External validation confirmed its robustness. SHAP analysis highlighted Wavelet_LLH_firstorder_Maximum.1, Gradient_ngtdm_Complexity, and tumor grade as top predictors. The model showed good calibration and clinical utility on DCA. An interpretable CT radiomics-based machine learning model integrating clinical and imaging features was developed to accurately predict pathological upstaging risk after ReTUR in NMIBC patients. This tool may support clinical decision-making for individualized treatment after multicenter validation.

ObjectiveRobot-assisted partial nephrectomy (RAPN) is an established, minimally invasive technique to treat patients with renal masses. The incidence of acute kidney injury (AKI) after RAPN is high and is associated with poor prognosis. This study aims to develop and validate an interpretable machine-learning model based on clinical features for individualized risk assessment of RAPN-AKI.MethodsWe retrospectively reviewed 325 patients undergoing RAPN at the Third Medical Center of PLA General Hospital (May 2022–Oct 2023) as the training dataset, and 146 from the Fifth Medical Center of PLA General Hospital (Nov 2023–Dec 2024) for external validation. Models were constructed using Boruta-selected features and eight machine learning algorithms. Performance was assessed by the area under the receiver operating characteristic curve (AUC), F1-score, accuracy, precision, calibration, and decision curve analysis (DCA). Shapley additive explanations (SHAP) interpreted feature contributions.ResultsThe incidence of AKI in internal training and external validation datasets was 24.6% and 26%, respectively. The Boruta algorithm identified duration of renal artery blockade, preoperative serum creatinine (Scr), gender, body mass index (BMI), and age as important features. Among the eight machine learning models, the Gradient Boosting Machine (GBM) model demonstrated the best and most stable predictive outcomes in the internal training dataset (AUC = 0.889) and external validation dataset (AUC = 0.779). Both the calibration curve and DCA indicated better calibration and greater net benefit. SHAP analysis revealed the contribution of important features in the following order: duration of renal artery blockade, Scr, BMI, age, and gender. Dependency plots showed that duration of renal artery blockade > 22 min, Scr > 80 µmol/L, BMI > 25 kg/m², age > 60 years, and male were significantly associated with an increased risk of AKI.ConclusionThe GBM model exhibited strong predictive performance in both internal training dataset and external validation dataset and has the potential to assist clinicians to identify the high-risk patients early, enabling timely interventions that may reduce the incidence of RAPN-AKI and improving clinical outcomes. While, the interpretable machine learning model is currently applicable only to patients with low-risk or normal preoperative renal function.

Optimisation and interpretation of machine and deep learning models for improved water quality management in Lake Loktak

Machine learning-driven prediction of readmission risk in heart failure patients with diabetes: synergistic assessment of inflammatory and metabolic biomarkers.