Artificial intelligence in the GPs office: a retrospective study on diagnostic accuracy

Type 2 diabetes (T2D) is a disease with high morbidity and mortality and a disproportionate impact on minority groups. Machine learning (ML) is increasingly used to characterize T2D risk factors; however, it has not been used to study T2D health disparities. Our objective was to use explainable ML methods to discover and characterize T2D health disparity risk factors. We applied SHapley Additive exPlanations (SHAP), a new class of explainable ML methods that provide interpretability to ML classifiers, to this end. ML classifiers were used to model T2D risk within and between self-identified race and ethnicity (SIRE) groups, and SHAP values were calculated to quantify the effect of T2D risk factors. We then stratified SHAP values by SIRE to quantify the effect of T2D risk factors on prevalence differences between groups. We found that ML classifiers (random forest, lightGBM, and XGBoost) accurately modeled T2D risk and recaptured the observed prevalence differences between SIRE groups. SHAP analysis showed the top seven most important T2D risk factors for all SIRE groups were the same, with the order of importance for features differing between groups. SHAP values stratified by SIRE showed that income, waist circumference, and education best explain the higher prevalence of T2D in the Black or African American group, compared to the White group, whereas income, education and triglycerides best explain the higher prevalence of T2D in the Hispanic or Latino group. This study demonstrates that explainable ML can be used to elucidate health disparity risk factors and quantify their group-specific effects.

Dissolved organic matter (DOM) assemblages in freshwater rivers are formed from mixtures of simple to complex compounds that are highly variable across time and space. These mixtures largely form due to the environmental heterogeneity of river networks and the contribution of diverse allochthonous and autochthonous DOM sources. Most studies are, however, confined to local and regional scales, which precludes an understanding of how these mixtures arise at large, e.g., continental, spatial scales. The processes contributing to these mixtures are also difficult to study because of the complex interactions between various environmental factors and DOM. Here we propose the use of machine learning (ML) approaches to identify ecological processes contributing toward mixtures of DOM at a continental-scale. We related a dataset that characterized the molecular composition of DOM from river water and sediment with Fourier-transform ion cyclotron resonance mass spectrometry to explanatory physicochemical variables such as nutrient concentrations and stable water isotopes (2H and 18O). Using unsupervised ML, distinctive clusters for sediment and water samples were identified, with unique molecular compositions influenced by environmental factors like terrestrial input and microbial activity. Sediment clusters showed a higher proportion of protein-like and unclassified compounds than water clusters, while water clusters exhibited a more diversified chemical composition. We then applied a supervised ML approach, involving a two-stage use of SHapley Additive exPlanations (SHAP) values. In the first stage, SHAP values were obtained and used to identify key physicochemical variables. These parameters were employed to train models using both the default and subsequently tuned hyperparameters of the Histogram-based Gradient Boosting (HGB) algorithm. The supervised ML approach, using HGB and SHAP values, highlighted complex relationships between environmental factors and DOM diversity, in particular the existence of dams upstream, precipitation events, and other watershed characteristics were important in predicting higher chemical diversity in DOM. Our data-driven approach can now be used more generally to reveal the interplay between physical, chemical, and biological factors in determining the diversity of DOM in other ecosystems.

The application of Artificial Intelligence (AI) and Machine Learning (ML) models is increasingly leveraged to automate and optimize Data Centre (DC) operations. However, the interpretability and transparency of these complex models pose critical challenges. Hence, this paper explores the Shapley Additive exPlanations (SHAP) values model explainability method for addressing and enhancing the critical interpretability and transparency challenges of predictive maintenance models. This method computes and assigns Shapley values for each feature, then quantifies and assesses their impact on the model’s output. By quantifying the contribution of each feature, SHAP values can assist DC operators in understanding the underlying reasoning behind the model’s output in order to make proactive decisions. As DC operations are dynamically changing, we additionally investigate how SHAP can capture the temporal behaviors of feature importance in the dynamic DC environment over time. We validate our approach with selected predictive models using an actual dataset from a High-Performance Computing (HPC) DC sourced from the Enea CRESCO6 cluster in Italy. The experimental analyses are formalized using summary, waterfall, force, and dependency explanations. We delve into temporal feature importance analysis to capture the features’ impact on model output over time. The results demonstrate that model explainability can improve model transparency and facilitate collaboration between DC operators and AI systems, which can enhance the operational efficiency and reliability of DCs by providing a quantitative assessment of each feature’s impact on the model’s output.

Breast cancer continues to be a leading cause of death among women in the world. The prediction of survival outcomes based on treatment modalities, i.e., chemotherapy, hormone therapy, surgery, and radiation therapy is an essential step towards personalization in treatment planning. However, Machine Learning (ML) models may improve these predictions by investigating intricate relationships between clinical variables and survival. This study investigates the performance of several ML models to predict survival rate in patients undergoing diverse breast cancer treatments i.e., chemotherapy, hormone therapy, surgery and radiation using multiple clinical parameters. The dataset consisted of 5000 samples and turned into downloaded from Kaggle. The models assessed blanketed Support Vector Machines (SVM), K-Nearest Neighbor (KNN), AdaBoost, Gradient Boosting, Random Forest, Gaussian Naive Bayes, Logistic Regression, Extreme Gradient Boosting (XG boost), and Decision tree. Performance of the models was assessed using parameters such as Accuracy, Precision, Recall, F1-Score and Area under the Receiver Operating Characteristic Curve (AUC-ROC). SHAP (SHapley Additive exPlanations) values analysis was done to provide an explanation for the impact of a feature on model predictions using Waterfall and Beeswarm plots. Anticipated baseline (E(f(x))) were in comparison to the predictions (f(x)) for each therapy group. Performance of Gradient Boosting was shown to be the best with an Accuracy: 0.972, Precision: 0.973, Recall: 0.972, F1-Score: 0.973, and AUC-ROC Score: 0.997. Chemotherapy had a notably bad impact on survival, with an f(x) of -0.274 and an E(f(x)) of -0.025. Hormone therapy showed the maximum detrimental effect on survival, with an f(x) of -0.408. Surgical operation had an extraordinarily impartial impact (f(x) = -0.041), even as radiation therapy positively impacted survival results with an f(x) of 0.22. Gradient Boosting was the most predictive algorithm for breast cancer survival outcomes. This SHAP-primarily based analysis provides a complete knowledge of ways one-of-a-kind treatments have an effect on survival predictions in breast cancer patients. Radiation therapy indicates the most tremendous effect on survival, whilst hormone therapy reveals the maximum poor effect. Future studies need to explore personalized treatment strategies that comprise these insights to enhance patient effects.

Current prediction tools for venous thromboembolism (VTE) after colorectal cancer (CRC) surgery lack accuracy for individualized care. To address this critical gap, we developed machine learning (ML) models to dynamically predict postoperative VTE in Chinese CRC surgery patients. We developed ML models using data from 1,836 CRC surgery patients across 46 Chinese centers in the CRC-VTE trial. The cohort was divided into a modeling set (n=1,515) and an external test set (n=321), with the modeling set further split into training and validation subsets (4:1 ratio). Preoperative and postoperative clinical features were selected using recursive feature elimination and Boruta. A total of 162 ML models were developed and evaluated across both preoperative and postoperative datasets, with model performance primarily assessed using the area under the receiver operating characteristic curve (AUROC). The final model underwent rigorous validation in an additional multicenter retrospective cohort and incorporated SHapley Additive exPlanations (SHAP) for risk calculation and decision interpretation. The CatBoost model achieved optimal performance, with AUROCs of 0.950 (preoperative data) and 0.971 (postoperative data) in the validation set, and 0.686±0.037 (preoperative) and 0.715±0.036 (postoperative) in the test set. Models incorporating postoperative data consistently outperformed preoperative-only models. Simplified models based on key predictors (7 preoperative and 9 postoperative features selected by SHAP values) maintained comparable performance, with AUROCs of 0.933 (preoperative data) and 0.949 (postoperative data) in the validation set, and 0.640±0.030 (preoperative data) and 0.695±0.030 (postoperative data) in the test set, respectively. Our study demonstrates the feasibility of a ML-based approach for predicting VTE following CRC surgery. The integration of ML with SHAP methodology provides a clinically actionable tool for individualized risk assessment and optimized VTE prevention strategies.

Prostate cancer is still one of the most common and fatal types of cancer in men worldwide. Early detection is the key to enhancing patient survival and reducing the use of invasive treatments. This research examines the use of machine learning as a tool for assisting the classification of prostate cancer cases based on standard clinical biomarkers. A public dataset of 100 patient samples was utilized, containing clinical features like PSA levels, perimeter, radius, texture, and more. A predictive model was created based on the XGBoost algorithm and its performance was tested using a 5-fold cross-validation method. The model resulted in a mean F1 score of 0.90 and a mean ROC–AUC of 0.93, indicating superior predictive capability. To improve interpretability, SHAP (Shapley Additive exPlanations) values were calculated, which identified PSA, perimeter, and texture as the most predictive features. These results indicate that machine learning algorithms, when combined with publicly available clinical biomarkers, can be used to develop trustworthy, interpretable, and cost-efficient tools for prostate cancer detection at an early stage. The method has the potential to be of use in supporting clinical decision-making and minimizing the necessity for invasive diagnostic testing. KEYWORDS: Prostate cancer, Machine learning, XGBoost algorithm, SHAP values, Clinical biomarkers

Supervised learning applied to classifying fallers versus non-fallers among older adults with cancer

BackgroundGout is a chronic inflammatory condition increasingly recognized as a risk factor for cardiovascular events (CVE). Early identification of high-risk individuals is crucial for targeted prevention and management. However, conventional risk stratification approaches often fall short in accuracy and clinical utility. This study aimed to develop and validate a robust, interpretable machine learning (ML)-based model for predicting CVE in patients with gout.MethodsThis retrospective cohort study included 686 hospitalized gout patients at Xiyuan Hospital (Beijing, China) between January 1, 2013, and December 31, 2023. We applied Synthetic Minority Oversampling Technique (SMOTE) combined with random undersampling of the majority class. Then, patients were randomly divided into training (70%) and testing (30%) sets. A comprehensive set of clinical and biochemical variables (n = 39) was collected. Feature selection was performed using Boruta algorithms and Lasso to identify the most predictive variables. Multiple ML algorithms—including Decision Tree Learner, LightGBM Learner, K Nearest Neighbors Learner, CatBoost Learner, Gradient Boosting Desicion Tree Learner—were implemented to construct predictive models. SHAP values were used to assess model interpretability, and robustness was evaluated through 10-fold bootstrap resampling with enhanced standard error estimation.ResultsOf the 686 patients, 263 experienced cardiovascular events during follow-up (incidence rate: 38.3%). A logistic regression model was constructed based on eight variables selected using the Boruta feature selection algorithm: sex, age, PLT, EOS, LYM, CO2, GLU and APO-B. Among the five models evaluated, the CatBoost classifier achieved the best performance, with the highest area under the ROC curve (AUC) of 0.976 and the recall of 0.971. Furthermore, SHAP (SHapley Additive exPlanations) values were employed to provide both global and individual-level interpretability of the CatBoost model. To assess the model’s generalization performance, bootstrap resampling was performed 10 times. Based on these results, the standard error was improved using machine learning-based enhancement methods, thereby optimizing the model’s robustness and predictive stability.ConclusionThe logistic regression analysis revealed that age (OR=1.351, p<0.001), CO2 (OR=0.603, p=0.004), eosinophil count (OR=2.128, p=0.001), and platelet count (OR=0.961, p<0.001) were significantly associated with the outcome, indicating their potential roles as independent predictors. Notably, while APO_B (p=0.138) and sex (p=0.132) showed no significant association, glucose levels (OR=2.1, p=0.066) exhibited a marginal trend toward significance, warranting further investigation. This tool may support clinicians in identifying high-risk individuals, enabling early interventions and optimized management strategies.LimitationsThis study has several limitations. First, the analysis was based on a single-center dataset, which may limit the generalizability of the findings. External validation in multi-center and prospective cohorts, along with an expanded sample size, is warranted to confirm these results. Second, key confounding factors such as medication use, lifestyle habits, and gout flare frequency were not included in the analysis; future studies should incorporate these variables to provide a more comprehensive assessment.

Employee-Friendly Corporate Culture and Firm Performance: Evidence from a Machine Learning Approach

Tree-based machine learning (ML) algorithms, such as Extra Trees (ET), Random Forest (RF), Gradient Boosting Machine (GBM), and XGBoost (XGB) are among the most widely used in early drug discovery, given their versatility and performance. However, models based on these algorithms often suffer from misclassification and reduced interpretability issues, which limit their applicability in practice. To address these challenges, several approaches have been proposed, including the use of SHapley Additive Explanations (SHAP). While SHAP values are commonly used to elucidate the importance of features driving models' predictions, they can also be employed in strategies to improve their prediction performance. Building on these premises, we propose a novel approach that integrates SHAP and features value analyses to reduce misclassification in model predictions. Specifically, we benchmarked classifiers based on ET, RF, GBM, and XGB algorithms using data sets of compounds with known antiproliferative activity against three prostate cancer (PC) cell lines (i.e., PC3, LNCaP, and DU-145). The best-performing models, based on RDKit and ECFP4 descriptors with GBM and XGB algorithms, achieved MCC values above 0.58 and F1-score above 0.8 across all data sets, demonstrating satisfactory accuracy and precision. Analyses of SHAP values revealed that many misclassified compounds possess feature values that fall within the range typically associated with the opposite class. Based on these findings, we developed a misclassification-detection framework using four filtering rules, which we termed "RAW", SHAP, "RAW OR SHAP", and "RAW AND SHAP". These filtering rules successfully identified several potentially misclassified predictions, with the "RAW OR SHAP" rule retrieving up to 21%, 23%, and 63% of misclassified compounds in the PC3, DU-145, and LNCaP test sets, respectively. The developed flagging rules enable the systematic exclusion of likely misclassified compounds, even across progressively higher prediction confidence levels, thus providing a valuable approach to improve classifier performance in virtual screening applications.

Artificial intelligence-based analysis of behavior and brain images in cocaine-self-administered marmosets

Forecasting Hematological Activity in Antiphospholipid Syndrome Using Predictive Models

Institutional delivery dropout (IDD), defined as delivery outside a health facility despite attending antenatal care (ANC), remains a significant barrier to reducing maternal mortality in Nigeria. Traditional statistical models often fall short of capturing the complex, non-linear interactions among the socio-demographic factors that drive this critical health behavior. Using a comprehensive dataset of 16,100 women from the 2018 Nigeria Demographic and Health Survey (NDHS), we applied and compared seven diverse machine learning (ML) algorithms, including models such as Support Vector Machine (SVM), Gradient Boosting (GB), and Extreme Gradient Boosting (XGBoost). The model performance was systematically evaluated using metrics such as accuracy, Area Under the Receiver Operating Characteristic curve (AUROC), F1-score, and detailed confusion matrices. Furthermore, SHapley Additive explanations (SHAP) were used to provide transparent interpretations of feature importance and predictive contributions. Gradient Boosting was the best-performing model, achieving the highest F1-score (0.755) and AUROC (0.82). SVM achieved the highest accuracy (0.740) and recall (0.780). SHAP identified education level, household wealth, and religion as strong predictors of IDD. The performance metrics reported with confidence intervals showed modest variability across the models. Machine learning approaches were effective in identifying women at an increased risk of institutional delivery dropout. SHAP analysis provides insights into the key sociodemographic predictors of IDD, highlighting the value of interpretable ML methods in maternal health research.

In this work, we aim to show that machine learning algorithms are promising tools for the identification of nuclear data that contribute to increased errors in transport simulations. We demonstrate this through an application of a machine learning algorithm (Random Forest) to the Whisper/MCNP6 criticality validation library to identify nuclear data that are associated with an increase of the bias (simulated-experimental ) in the calculations. Specifically, the sensitivity profiles (with respect to nuclear data) of solution benchmarks are used to predict the bias, and SHapley Additive exPlanations (SHAP) are used to explain how the sensitivities are related to the predicted bias. The SHAP values can be interpreted as sensitivity coefficients of the machine learning model to the sensitivities that are used to make predictions of bias. Using the SHAP values, we can identify specific subsets of nuclear data that have the highest probability of influencing bias. We demonstrate the utility of this method by showing how SHAP values were used to identify an inconsistency in the inelastic scattering nuclear data. The methodology presented here is not limited to transport problems and can be applied to other simulations if there are experimental measurements to compare against, simulations of those experimental measurements, and the ability to calculate sensitivities of the model output with respect to the data inputs.

Ransomware is malware that encrypts a victim's files and demands payment, typically in cryptocurrencies, to restore access. The global spread of ransomware attacks underscores the need for automatic detection systems. Signature‐based antivirus tools cannot identify novel ransomware strains, while machine learning (ML) and deep learning (DL) models offer promise because they learn data patterns and generalize to new unseen samples. However, the choice of model architecture remains an open question: do classical ML methods suffice or do DL methods bring a considerable performance increase for static malware analysis? In this paper, we benchmark three classic algorithms (Logistic Regression, Random Forest and XGBoost) and two deep architectures (Multi‑Layer Perceptron and TabNet) on the Windows Portable Executable (PE) ransomware detection dataset from Kaggle. We perform comprehensive data preprocessing, 5‑fold stratified cross‑validation, and compute performance metrics like accuracy, precision, recall, F1 score, area under the ROC curve (AUC), and Matthews correlation coefficient (MCC). We further examine the feature importance, plot receiver operating characteristic (ROC) curves, and discuss the computational expense and interpretability of each method. Our experiments show that tree ensemble methods (Random Forest and XGBoost) achieve near‑perfect detection with very low false positives and negatives, outperforming Logistic Regression and MLP. Deep models offer no significant advantages for this tabular data set but with higher computational expense. We discuss the implications of these results for ransomware defense in practice and propose directions for future research.