Genetic-lifestyle risk prediction model for hypertension and diabetes using logistic regression

Abstract

This study focused on genetic-lifestyle risk prediction modelling of hypertension and diabetes using the logistic regression method. The sample sizes, n =1000 patients selected from a projected population, N= 567,694 in Lafia, Nasarawa State. A cross-sectional analytical design was employed, utilizing secondary data obtained from the District Health Information System (DHIS2) database. Classical logistic regression and LASSO logistic regression models were fitted to identify significant predictors of hypertension and diabetes. The results show (38.2%) had hypertension, and (29.5%) had diabetes. Many factors are indicated as diabetes risk: older age (OR = 1.053, p = .002), higher BMI (OR = 1.099, p = .001), family history (OR = 2.410, p = .001), and smoking (OR = 1.419, p = .041). Staying physically active, show protective indicator (OR = 0.657, p = .018). Hypertension indicate that age (OR = 1.046, p = .001), BMI (OR = 1.085, p = .001), family history (OR = 2.055, p < .001), and smoking (OR = 1.507, p = .021) increase risk, while physical activity gives protection (OR = 0.694, p = .030). The models show good testing, with AUC values of 0.78 for diabetes and 0.82 for hypertension which is suitable for detecting individual at higher risk. The findings show both diseases share a common genetic and lifestyle factor, which means a single predictive model could realistically screen for both at once which helps health workers in the study area a statistical tool for detecting individuals at risk early and prevention intervention.