Predictors of medication adherence in elderly patients with chronic diseases using support vector machine models

Area of Science:

• Gerontology
• Health Informatics
• Pharmacology

Background:

• Medication adherence is critical for managing chronic diseases in elderly populations.
• Predictive modeling can improve adherence rates and patient outcomes.
• Traditional statistical methods may not fully capture complex adherence patterns.

Purpose of the Study:

• To develop a prediction model for medication adherence in elderly patients with chronic diseases.
• To identify key variables influencing medication adherence using machine learning and statistical approaches.
• To compare the classification accuracy of Support Vector Machine (SVM) and Logistic Regression (LR).

Main Methods:

• A cohort of 293 elderly patients (age > 65) with chronic diseases was studied.
• Medication adherence was assessed using Morisky's self-report questionnaire via face-to-face interviews.
• Logistic Regression (LR) and Support Vector Machine (SVM) models were employed for classification.

Main Results:

• SVM achieved a significantly higher classification accuracy (97.3%) compared to LR (71.1%).
• Self-efficacy emerged as a single predictor with 72.4% accuracy, highlighting its importance.
• SVM identified self-efficacy, depression, health literacy, and medication knowledge as key factors.

Conclusions:

• Self-efficacy is strongly associated with medication adherence in elderly Korean patients.
• Modifiable factors like depression, health literacy, and medication knowledge impact adherence.
• SVM offers a powerful tool for accurate medication adherence prediction with fewer variables.