Related Experiment Video
Updated: May 28, 2026
12:18
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Development and Validation of a Machine Learning Model for Predicting Medication Adherence Among Home-Dwelling
Yujie Zhang1,2, Yongli Han2, Xuemei Yin3
1Department of Pharmacy, Beijing Tiantan Hospital, Capital Medical University, Beijing, People's Republic of China.
Patient Preference and Adherence
|May 27, 2026
Summary
This study developed an interpretable machine learning model to predict medication adherence in elderly Chinese patients with chronic diseases. The model highlights self-efficacy and social support as key factors for improving adherence and enabling targeted interventions.
Area of Science:
- Gerontology
- Pharmacology
- Artificial Intelligence
Background:
- Elderly chronic disease management is complex due to multimorbidity and polypharmacy, with poor medication adherence hindering progress.
- Existing predictive models often lack interpretability and fail to integrate crucial psychological and social support factors for community-dwelling elderly populations.
- Predictive models identifying adherence risks are vital for enabling proactive interventions in elderly patient care.
Purpose of the Study:
- To develop an interpretable machine learning prediction model for medication adherence in elderly Chinese patients with chronic diseases.
- To address the research gap in interpretable models that integrate psychological and social support factors.
- To provide a tool for identifying elderly patients at risk of medication non-adherence.
Main Methods:
- Data collected via face-to-face interviews from 1722 elderly patients (≥60 years) with chronic diseases on home medication therapy.
- Variables included demographics, comorbidities, medication details, adherence, self-efficacy, beliefs, social support, and literacy.
- Six machine learning algorithms were applied, with the Gradient Boosting Machine (GBM) selected as optimal and interpreted using Shapley Additive Explanations (SHAP).
Main Results:
- The Gradient Boosting Machine (GBM) model achieved the highest predictive performance (AUC = 0.811).
- Key predictors identified were self-efficacy in rational drug use, medication practice, concern beliefs, and social support availability.
- SHAP analysis enhanced model interpretability, offering a clear basis for clinical decision-making.
Conclusions:
- An interpretable prediction model for home medication adherence in elderly chronic disease patients was successfully constructed.
- The model incorporates significant social and psychological factors influencing patient adherence.
- This provides robust evidence for developing targeted interventions to improve medication adherence in this population.
