Machine Learning-Based Prediction Model for Health-Related Quality of Life in Diabetic Patients

Area of Science:

• Medical Informatics
• Public Health
• Machine Learning in Healthcare

Background:

• Diabetes mellitus (DM) prevalence is rising, impacting patients' health-related quality of life (HRQoL).
• Limited research exists on generalized models and risk factors for HRQoL decline in DM patients.
• Effective prediction models are needed to identify at-risk individuals for timely intervention.

Purpose of the Study:

• To develop and validate a machine learning (ML) model for predicting HRQoL in adult DM patients.
• To identify and analyze key factors influencing HRQoL among this population.
• To assess the potential of ML as a decision-support tool in diabetes care.

Main Methods:

• Utilized data from 2,501 adult DM patients from the Korea National Health and Nutrition Examination Survey (2016-2020).
• Developed and compared five ML classifiers: logistic regression, naïve Bayes, random forest, support vector machine, and extreme gradient boosting (XGBoost).
• Evaluated model performance using metrics like accuracy, recall, precision, F1-score, and AUC; feature importance was determined using SHAP values.

Main Results:

• The XGBoost model demonstrated superior performance with high accuracy (0.940), recall (0.943), precision (0.940), and AUC (0.984).
• Top predictors of HRQoL included self-rated health, employment status, triglyceride levels, education level, and AST/ALT ratio.
• The ML model achieved over 90% accuracy in distinguishing between stable and at-risk HRQoL groups.

Conclusions:

• Machine learning, particularly XGBoost, offers a highly accurate method for predicting HRQoL in adult diabetic patients.
• Identifying key influencing factors like self-rated health and socioeconomic status is crucial for understanding HRQoL.
• The developed ML model shows promise for integration into routine diabetes care to support clinical decision-making and improve patient outcomes.