Prevalence and Factors Associated with Insomnia Among Chronic Disease Patients in Bangladesh: A Machine Learning...

Insomnia significantly impairs both mental and physical health, and its bidirectional relationship with chronic diseases exacerbates outcomes for both conditions. While insomnia risk factors are well-studied in general populations, little is known about its prevalence and determinants among chronic disease patients in Bangladesh. Using machine learning (ML) alongside traditional analyses may improve prediction and early identification of insomnia risk in this high-vulnerability group.

This cross-sectional study recruited 1,222 adult chronic disease patients from healthcare facilities in Dhaka and Chattogram between May and November 2024. Insomnia was assessed using the Insomnia Severity Index (ISI-7). Multivariable logistic regression identified significant risk and protective factors. Six ML classifiers, K-Nearest Neighbors (KNN), Random Forest (RF), Support Vector Machine (SVM), Gradient Boosting Machine (GBM), Extreme Gradient Boosting (XGBoost), and Categorical Boosting (CatBoost), were trained and tested (with Synthetic Minority Over-sampling Technique for class imbalance), and model performance was evaluated using accuracy, precision, F1 score, log loss, and the area under the receiver operating characteristic curve (AUC-ROC). Feature importance was determined via SHapley Additive exPlanations (SHAP) and gain values.

Insomnia affected 41.3% of patients. Risk factors included female gender, joint family, urban residence, smokeless tobacco and substance use, prolonged daytime napping, late disease onset, presence of other chronic diseases, and unmet mental healthcare needs. Protective factors were physical activity, 7-9 hours of nighttime sleep, met mental healthcare needs, and notably, presence of urinary disease. Among ML models, CatBoost outperformed others (accuracy 71.67%, AUC 77.27%, F1 score 71.23%), followed closely by RF and SVM. Feature importance analysis consistently identified mental healthcare need fulfillment and nighttime sleep duration as the strongest predictors of insomnia.

Insomnia was common among Bangladeshi chronic disease patients and linked to sociodemographic, behavioral, clinical, and mental health factors. CatBoost and other ML models showed strong predictive ability, supporting their use in early screening. Prospective studies are needed to validate these findings and guide targeted interventions.