Mohammed, Omer

Abstract INTRODUCTION Alzheimer's disease (AD) is a progressive neurodegenerative disorder that needs better predictive tools. Using the National Alzheimer's Coordinating Center Uniform Data Set, this study developed machine learning (ML) models and a practical clinical tool for AD prediction. METHODS Data from 52,537 individuals (22,371 with AD) and more than 200 variables were processed with MissForest imputation and genetic algorithm‐based selection. Multiple ML models were trained, and interpretability was performed using SHAP and permutation importance. A LightGBM model was refined through iterative backward feature elimination (IBFE) followed by manual refinement. RESULTS LightGBM performed best (receiver operating characteristic‐area under the curve [ROC‐AUC] 0.91, accuracy 82.0%). Key predictors included arthritis, age, body mass index, and heart rate. A 19‐feature model retained accuracy (81.2%) and ROC‐AUC (0.90). DISCUSSION This lightweight tool predicts AD using mostly routine variables. Limitations include its cross‐sectional nature, and would need external validation. An interactive web app and GitHub resource are available. Highlights Developed a lightweight ML based tool using 19 routinely available features. The lightweight model achieved an ROC-AUC of 0.90 for Alzheimer's disease prediction on NACC multicenter data. Genetic algorithm, IBFE, and manual refinement enabled optimal feature selection. Tool hosted on an open‐access platform for clinical and research use. SHAP analysis provided model interpretability and feature‐level insights.

Background: Stroke is a devastating complication of transcatheter aortic valve replacement (TAVR). Cerebral embolic protection devices (CEPD) have been introduced to reduce TAVR-related stroke by capturing or deflecting debris during the procedure. While these devices can capture embolic material, their clinical benefit in preventing stroke remains uncertain. We aimed to determine whether the use of CEPD during TAVR reduces the incidence of periprocedural stroke. Methods: We performed a systematic review and meta-analysis of studies comparing TAVR with versus without CEPD. We searched PubMed, Embase, and ClinicalTrials.gov for RCTs (randomized controlled trials) comparing CEPDs to no protection in TAVR patients from 2011 to May 2025. Nine RCTs involving a total of 11,608 patients (CEPD group n=5,953; control n=5,655) were included. Stroke outcomes were defined according to Valve Academic Research Consortium-2 (VARC-2) criteria and categorized as- all stroke, disabling stroke, or non-disabling stroke. Risk ratios (RR) with 95% confidence intervals (CI) were calculated using a fixed-effects model. Heterogeneity also was measured. Results: Stroke rates were low and did not differ significantly between the CEPD (n=343) and control (n=319) groups. The incidence of all strokes was 1.96% with CEPD vs 1.88% without CEPD (RR 0.97, 95% CI 0.75–1.15; p=0.50). Disabling stroke occurred in 0.79% of CEPD patients vs 0.95% of controls (RR 0.79, 95% CI 0.57-1.10; p=0.16). Non-disabling stroke occurred in 1.26% of patients in both groups (RR 1.06, 95% CI 0.8 -1.40; p=0.64). There was no significant heterogeneity across trials is (0) for all endpoints. Notably, the largest trial in the analysis (BHF PROTECT-TAVI, n=7,635) showed no difference in stroke rates with CEPD, aligning with the overall results. Subgroup analyses revealed a trend toward fewer disabling strokes with CEPD use, but this trend was not statistically significant. Conclusion: We found no significant stroke risk reduction with the use of CEPDs during TAVR. While a trend toward fewer disabling strokes was observed, it was not statistically significant. Routine use of CEPDs is not supported, and selective use in high-risk patients warrants further large RCTs.