JMIR Publications

ABSTRACT

Background:

Schizophrenia is a severe mental disorder characterized by high disability and relapse rates, posing significant challenges to public safety and community-based mental health management.

Objective:

Developing an accurate and interpretable relapse risk prediction model based on real-world community rehabilitation data is essential for implementing targeted and effective interventions.

Methods:

This study aimed to evaluate the predictive capability of structured variables extracted from information management system (IMS) in identifying relapse among patients with schizophrenia undergoing community-based rehabilitation during a clinically stable phase, and to identify key influencing factors. A total of 25,803 electronic records of community-managed patients with schizophrenia collected between Jun 1, 2023, and December 31, 2023, were analyzed. The dataset included demographic information, clinical records, interview data, policy support, and management indicators to construct a comprehensive feature system. A LightGBM machine learning algorithm was employed for model development, with a hybrid resampling strategy applied to address class imbalance. Model interpretability was enhanced using SHapley Additive exPlanations (SHAP) to identify the most influential predictors of relapse risk.

Results:

The dataset was stratified and randomly divided into a training set (80%) and an independent test set (20%). The model demonstrated robust predictive performance across multiple evaluation metrics, achieving an accuracy of 0.94, an F1 score of 0.92, and a recall of 0.92, indicating strong discriminative ability for individual relapse risk. SHAP analysis revealed that medication adherence pattern (11.0%), participation in medication-related rehabilitation training (10.8%), age (6.28%), social participation (6.19%), and participation in daily living skills training (5.66%) were the principal determinants of relapse. The model also uncovered potential relapse risks hidden beneath seemingly stable social functioning, underscoring the importance of psychological regulation and sustained system-level support.

Conclusions:

Using large-scale community information management system data, this study established a relapse risk prediction model that integrates multidimensional factors capturing the clinical complexity of schizophrenia. The findings provide an evidence-based foundation for designing personalized intervention strategies and optimizing community-based mental health services to prevent relapse and promote long-term stability.