Machine Learning and Multivariate Analysis of Depression Prevalence and Predictors in Acute Coronary Syndrome

Depression is a pervasive yet under-recognized comorbidity of acute coronary syndrome (ACS), which further degrades cardiac outcomes, particularly in low-resource environments. A cross-sectional case–control study was performed at the National Institute of Cardiovascular Diseases, Karachi, to quantify the burden of depressive symptoms and model key determinants. Two hundred eighty-six adults (150 ACS cases and 136 age-matched controls) completed the Patient Health Questionnaire-9 (PHQ-9). Conventional statistics and five machine-learning classifiers—logistic regression, support vector machine, random forest, XGBoost, and a deep neural network—were benchmarked for detection and risk stratification. Depressive symptoms, as measured by the PHQ-9, were identified in 93.3% of ACS patients versus 18.4% of controls, and 66.7% of cases reported moderate-to-severe symptoms. Age ≥ 55 years (adjusted OR = 2.1, p < 0.01) and diabetes mellitus (adjusted OR = 1.9, p < 0.05) emerged as independent associated factors. Logistic regression achieved the best overall discrimination (accuracy = 0.93, AUC = 0.95), while the neural network yielded the highest F1-score (0.878). These findings highlight an overwhelming but addressable mental health gap in cardiology in Pakistan. Routine PHQ-9 screening coupled with algorithm-guided referral pathways could enable timely psychosocial interventions and ultimately improve both psychological well-being and cardiovascular prognosis in ACS patients.