Development of a blood HbA1c level detection model based on Support Vector Regression (SVR) using microtest data

¹ Department of Physics, Faculty of Mathematics and Natural Sciences, IPB University, Bogor, 16680, Indonesia
² Equipment Manufacturing Technology Research Centre, National Research and Innovation Agency (BRIN), Serpong, 15346, Indonesia
³ Department of Community Nutrition, Faculty of Human Ecology, IPB University, Bogor, 16680, Indonesia
⁴ Faculty of Medicine, IPB University, Bogor, 16680, Indonesia
⁵ Master of Public Health Program, Faculty of Health Sciences and Technology, Binawan University, Jakarta, 13630, Indonesia
⁶ Directorate of Research, Development, and Innovation, Indonesian Artificial Intelligence Society, Jakarta, 12930, Indonesia
⁷ Computer Science Study Program, School of Data Science, Mathematics and Informatics, IPB University, Bogor, 16680, Indonesia
⁸ Institute of Nano Electronic Engineering, Universiti Malaysia Perlis, Kangar, Perlis 01000, Malaysia

^* Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it. , This email address is being protected from spambots. You need JavaScript enabled to view it.

Abstract

Glycated hemoglobin (HbAlc) is a key indicator of long-term glycemic control and a marker of diabetes diagnosis. Rapid and cost-effective prediction from microtest data may support screening in resource-limited settings. This study developed and evaluated an HbAlc prediction model using Support Vector Regression (SVR) on small-scale primary microtest data (10 subjects, three repeated sessions) with strict procedures to prevent data leakage. Clinical and biometric numerical variables were standardized and modeled using an SVR with a Radial Basis Function (RBF) kernel. In 5-fold cross-validation, Spearman correlation was applied exclusively to the training data to select the top 10 features per fold, followed by hyperparameter optimization (C, epsilon, gamma) using grid search with cross-validation. Model performance was evaluated using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R². The SVR achieved MAE ≈ 0.705, RMSE ≈ 1.285, and R² ≈ −0.162, indicating performance close to the mean baseline under leakage-free validation. Frequently selected predictors included HbA1c measurements at multiple time points and clinical indicators such as Impaired Glucose Tolerance (IGT). While predictive performance remains limited by sample size, the study establishes a methodologically robust framework for small-scale HbA1c modeling.