Predicting the Trajectory of Type 2 Diabetes Using a Hybrid Cellular Learning Automata and SIR Model: A Real-World Data Approach

Mostafa  Kashani; Sedigheh  Barzehkar

doi:10.18502/ijdl.v25i5.20339

Mostafa Kashani Department of Health Information Technology, Sirjan School of Medical Sciences, Sirjan, Iran
Sedigheh Barzehkar Department of Medicine, Sirjan School of Medical Sciences, Sirjan, Iran

DOI: https://doi.org/10.18502/ijdl.v25i5.20339

Keywords: Type 2 diabetes, Predictive modeling, Cellular Learning Automata (CLA), Susceptible–Infected–Recovered (SIR) model, health data mining

Abstract

Background: Type 2 diabetes is a major public-health threat of the present century, imposing substantial clinical and economic burdens on health systems. Accurate forecasting of disease incidence can support resource allocation and the design of targeted interventions.

Methods: In this study, we developed a hybrid model that integrates Cellular Learning Automata (CLA) with a Susceptible–Infected–Recovered (SIR) framework to predict the 20-year spread of type 2 diabetes using real patient data from Kerman province. The dataset comprised demographic and laboratory features of patients with diabetes collected during the Persian calendar years 2005– 2013. After preprocessing and imputation of missing values, the proposed model was implemented in MATLAB.

Results: Results indicate that the CLA–SIR combination models the disease trajectory with high accuracy. Moreover, factors such as blood pressure, cholesterol, and body mass index were identified as key drivers influencing the activation states of model cells.

Conclusion: These findings suggest that intelligent hybrid approaches can be effective for health-data analysis and long- term prediction of chronic diseases.