PREDICTIVE ANALYTICS USING LOGISTIC REGRESSION FOR EFFECTIVE HEALTH RISK MANAGEMENT: A CASE STUDY

Abstract

Predictive analytics plays an increasingly important role in healthcare by enabling early identification of individuals at elevated risk of adverse health outcomes. This study aimed to assess the effectiveness of logistic regression in predicting high health risk using demographic, behavioural, and clinical indicators. A quantitative, cross-sectional design was employed using a dataset of 800 observations representing adult individuals with varying health profiles. Health risk status was modeled as a binary outcome, and predictors included age, body mass index, blood pressure, cholesterol levels, glucose levels, smoking status, physical activity, family history, and sex. Logistic regression analysis was conducted to estimate the probability of high health risk, with model performance evaluated using odds ratios, goodness-of-fit measures, and classification metrics. The results indicated that age, diastolic blood pressure, glucose level, smoking, physical activity, and family history were significant predictors of health risk. Smoking substantially increased the odds of high health risk, whereas physical activity demonstrated a protective effect. The model exhibited satisfactory fit and strong discriminative ability, with a McFadden's R² of .241 and a receiver operating characteristic area under the curve of .818. Although specificity was high, sensitivity was moderate, highlighting a trade-off in detecting all high-risk individuals. Overall, the findings demonstrate that logistic regression provides an interpretable and effective approach for health risk assessment and may support data-driven decision-making in preventive healthcare contexts.