BRIDGING FRONTLINE SCREENING GAPS: EVALUATING NONINVASIVE MACHINE LEARNING MODELS FOR EARLY CARDIOVASCULAR STRATIFICATION IN LOW-RESOURCE PRIMARY CARE

Cardiovascular diseases (CVDs) remain the leading cause of global mortality, yet early detection remains severely bottlenecked in low-resource primary care environments due to the cost and logistical constraints of routine laboratory diagnostics. Standard risk-scoring tools frequently rely on comprehensive lipid panels and specialized biomarkers, resources rarely accessible at frontline health posts or community pharmacies. This paper evaluates the feasibility of machine learning algorithms trained exclusively on non-invasive, low-cost clinical markers (age, resting blood pressure, body mass index, smoking status, and simple physiological indicators) to stratify early cardiovascular risk. By analyzing predictive accuracy against clinical utility, dataset shift, and algorithmic fairness, this review outlines how interpretable decision-support models can improve frontline triage without introducing diagnostic disparities in underserved populations.