An Explainable Prediction Model for Cerebral Small Vessel Disease Based on Motor Function and Feature Interactions

Abstract

Cerebral small vessel disease (CSVD) is a major condition with systemic and whole-brain effects, significantly impacting the daily lives of middle-aged and elderly people. Currently, predictions for CSVD are limited to traditional statistical methods and lack consideration of interactions between risk factors. Based on this, this study proposes an interpretable risk prediction method for CSVD based on combined motor function and interaction features. First, the proposed multidimensional feature selection method is applied for feature selection. Next, the RS+GWO+Voting method is proposed for model construction. Finally, the model constructed using SHapley Additive exPlanations (SHAP) is applied for interpretation, enhancing the clinical applicability of the model. The results show that the proposed model achieved a training set Area Under the Curve (AUC) of 0.829 and a validation set AUC of 0.824, demonstrating its strong predictive performance. The method provides theoretical support and technical backing for CSVD risk prediction under small-sample conditions and holds promise for integration with wearable devices to enable early warning and health management of CSVD.



Received: 8 January 2026 | Revised: 31 March 2026 | Accepted: 15 April 2026



Conflicts of Interest

Congsi Wang is an editorial board member for Smart Wearable Technology and was not involved in the editorial review or the decision to publish this article. The authors declare that they have no conflicts of interest to this work.



Data Availability Statement

Data are available from the corresponding author upon reasonable request.



Author Contribution Statement

Benben Wang: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Writing – original draft, Visualization. Yuefei Yan: Methodology, Writing – review & editing, Project administration. Huizhen Lu: Validation, Investigation, Data curation. Baoqing Han: Writing – review & editing, Project administration. Yuangen Deng: Software. Chengliang Zhang: Data curation. Jianwei Xu: Data curation. Jingtan Chen: Supervision. Chuanliu Wang: Resources, Supervision. Congsi Wang: Conceptualization, Formal analysis, Resources, Funding acquisition.