Develop an ADR prediction system of Chinese herbal injections containing Panax notoginseng saponin: a nested case-control study using machine learning

A personalized prediction model for urinary tract infections in type 2 diabetes mellitus using machine learning

Patients with type 2 diabetes mellitus (T2DM) are at higher risk for urinary tract infections (UTIs), which greatly impacts their quality of life. Developing a risk prediction model to identify high-risk patients for UTIs in those with T2DM and assisting clinical decision-making can help reduce the incidence of UTIs in T2DM patients. To construct the predictive model, potential relevant variables were first selected from the reference literature, and then data was extracted from the Hospital Information System (HIS) of the Sichuan Academy of Medical Sciences and Sichuan Provincial People's Hospital for analysis. The data set was split into a training set and a test set in an 8:2 ratio. To handle the data and establish risk warning models, four imputation methods, four balancing methods, three feature screening methods, and eighteen machine learning algorithms were employed. A 10-fold cross-validation technique was applied to internally validate the training set, while the bootstrap method was used for external validation in the test set. The area under the receiver operating characteristic curve (AUC) and decision curve analysis (DCA) were used to evaluate the performance of the models. The contributions of features were interpreted using the SHapley Additive ExPlanation (SHAP) approach. And a web-based prediction platform for UTIs in T2DM was constructed by Flask framework. Finally, 106 variables were identified for analysis from a total of 119 literature sources, and 1340 patients were included in the study. After comprehensive data preprocessing, a total of 48 datasets were generated, and 864 risk warning models were constructed based on various balancing methods, feature selection techniques, and a range of machine learning algorithms. The receiver operating characteristic (ROC) curves were used to assess the performances of these models, and the best model achieved an impressive AUC of 0.9789 upon external validation. Notably, the most critical factors contributing to UTIs in T2DM patients were found to be UTIs-related inflammatory markers, medication use, mainly SGLT2 inhibitors, severity of comorbidities, blood routine indicators, as well as other factors such as length of hospital stay and estimated glomerular filtration rate (eGFR). Furthermore, the SHAP method was utilized to interpret the contribution of each feature to the model. And based on the optimal predictive model a user-friendly prediction platform for UTIs in T2DM was built to assist clinicians in making clinical decisions. The machine learning model-based prediction system developed in this study exhibited favorable predictive ability and promising clinical utility. The web-based prediction platform, combined with the professional judgment of clinicians, can assist to make better clinical decisions.

Development and assessment of novel machine learning models to predict the probability of postoperative nausea and vomiting for patient-controlled analgesia

Postoperative nausea and vomiting (PONV) can lead to various postoperative complications. The risk assessment model of PONV is helpful in guiding treatment and reducing the incidence of PONV, whereas the published models of PONV do not have a high accuracy rate. This study aimed to collect data from patients in Sichuan Provincial People's Hospital to develop models for predicting PONV based on machine learning algorithms, and to evaluate the predictive performance of the models using the area under the receiver characteristic curve (AUC), accuracy, precision, recall rate, F1 value and area under the precision-recall curve (AUPRC). The AUC (0.947) of our best machine learning model was significantly higher than that of the past models. The best of these models was used for external validation on patients from Chengdu First People's Hospital, and the AUC was 0.821. The contributions of variables were also interpreted using SHapley Additive ExPlanation (SHAP). A history of motion sickness and/or PONV, sex, weight, history of surgery, infusion volume, intraoperative urine volume, age, BMI, height, and PCA_3.0 were the top ten most important variables for the model. The machine learning models of PONV provided a good preoperative prediction of PONV for intravenous patient-controlled analgesia.