JMIR Publications

ABSTRACT

Background:

Chronic diseases remain a major contributor to global morbidity and mortality, often progressing silently before clinical detection. Conventional risk assessment models are limited in their ability to incorporate diverse health determinants and capture complex interactions among risk factors.

Objective:

This study aims to develop a Machine Learning–Driven Health Risk Index (ML-HRI) that integrates multi-source health data to predict individual susceptibility to chronic diseases and enable effective risk stratification.

Methods:

A data-driven framework was designed using demographic, clinical, and lifestyle variables. Data preprocessing techniques, including cleaning, normalization, and imputation, were applied to ensure data quality. Multiple machine learning algorithms—logistic regression, decision tree, random forest, and gradient boosting—were implemented and evaluated using performance metrics such as accuracy, precision, recall, F1-score, and area under the receiver operating characteristic curve (AUC-ROC). The optimal model was used to generate a normalized composite risk score, which was categorized into low-, moderate-, and high-risk levels.

Results:

Ensemble models, particularly random forest and gradient boosting, outperformed baseline approaches in predictive accuracy and overall performance. Key predictors included age, body mass index, blood pressure, and glucose levels. The ML-HRI effectively distinguished between different levels of chronic disease risk, demonstrating strong capability for population stratification.

Conclusions:

The proposed ML-HRI provides a robust and interpretable framework for chronic disease risk prediction. By combining machine learning with comprehensive health data, the model supports early identification of at-risk individuals and facilitates data-driven decision-making in clinical and public health contexts. This approach contributes to advancing personalized healthcare and preventive strategies.