AI-BASED PREDICTION OF CARDIOVASCULAR RISK FACTORS AMONG MIDDLE-AGED MEN IN PAKISTAN

Abstract

Background: Cardiovascular diseases remain a major public health concern in Pakistan, particularly among middle-aged men. Conventional risk prediction tools may not adequately reflect the population-specific patterns observed in this demographic. The use of artificial intelligence (AI) offers an opportunity to enhance predictive accuracy through data-driven risk stratification models.

Objective: To assess the effectiveness of AI tools in predicting cardiovascular risk among adult male populations in South Punjab, Pakistan.

Methods: A cross-sectional study was conducted over eight months, enrolling 320 adult males aged 40–60 years through multistage sampling. Clinical, anthropometric, and biochemical data were collected, including blood pressure, lipid profile, fasting glucose, BMI, and lifestyle factors. An AI model utilizing ensemble machine learning techniques was developed using Python libraries. Model performance was evaluated using sensitivity, specificity, precision, recall, and AUC-ROC. Comparisons were made with standard tools including the Framingham Risk Score and WHO/ISH risk charts.

Results: The AI model achieved an AUC of 0.87, with a sensitivity of 81%, specificity of 79%, precision of 76%, and recall of 81%. It accurately identified 84.1% of individuals at high cardiovascular risk, surpassing the predictive accuracy of both the Framingham Score (72.3%) and WHO/ISH charts (68.7%). Key risk factors among the study population included elevated LDL (128 ± 32 mg/dL), high triglycerides (176 ± 47 mg/dL), and a smoking prevalence of 30%.

Conclusion: AI-based tools demonstrated superior accuracy in predicting cardiovascular risk compared to traditional methods. Their integration into primary care settings may significantly improve early detection and prevention strategies for cardiovascular disease in resource-limited regions.