Modelling Stroke Risk Based on Age, Objective Factors, and Food Consumption Using Nonparametric Binary Logistic Regression with Least Square Spline Estimator

Author : Marisa Rifada, Nur Chamidah, I Nyoman Budiantara, Muhammad Anshari, Naufal Ramadhan Al Akhwal Siregar, Renatalia Fika

Abstract :Stroke is a leading global cause of mortality and disability, which is categorized into ischemic and hemorrhagic stroke. This study develops a nonparametric binary logistic regression model based on the least square spline estimator (LSS-NBLR) to assess the risk of ischemic and hemorrhagic stroke. The model estimates the odds ratio for each risk factor, allowing for flexible analysis of how variables impact stroke risk. Risk factors analyzed include age, LDL cholesterol, uric acid, triglycerides, blood glucose and sodium levels. Data were sourced from Dr. Drs. M. Hatta Brain Hospital, Bukittinggi, Indonesia, in 2023. The results indicate that age ≥ 50 years, LDL ≥ 125 mg/dL, triglycerides ≥ 125 mg/dL, and blood glucose ≥ 150 mg/dL increase the odds of ischemic stroke. In contrast, uric acid and sodium levels consistently showed odds ratios below 1 across all spline segments, indicating a decreasing tendency in the odds. The LSS-NBLR model achieved 87.14% accuracy for in-sample data and 93.3% for out-of-sample data. These findings highlight the model’s flexibility and accuracy in stroke risk prediction. Furthermore, this study contributes to SDGs by supporting early detection and prevention strategies for stroke through data-driven biostatistical modeling.

Keywords :Stroke Risk Prediction, Ischemic Stroke, Hemorrhagic Stroke, Nonparametric Binary Logistic Regression, Least Square Spline Estimator, Biostatistical Modeling, Odds Ratio Analysis, Cardiovascular Risk Factors, SDG 3

Conference Name :International Conference on Statistical Genetics and Biostatistics (ICSGAB - 25)

Conference Place Seoul, South Korea

Conference Date 23rd Dec 2025