A Stratified Modeling-Machine Learning Approach to Improve the Accuracy of Non-Invasive Blood Glucose Estimation Using Photoplethysmography Signals

Aug 5, 2026

Upload by Sage

Diabetes requires continuous blood glucose monitoring, yet invasive approaches introduce pain, discomfort, and high costs while offering limited practicality for routine use. This study develops a non-invasive estimation method based on photoplethysmography (PPG) signals using stratified modeling-machine learning. Stratified regression applies linear modeling in the non-diabetes stratum and logarithmic modeling in the diabetes stratum, combined with classifiers such as SVM, KNN, Naïve Bayes, decision trees, and neural networks.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages21

Reading Time53 min

AccessFree to download

Introduction

Diabetes burden and need for continuous monitoring

Invasive versus non-invasive blood glucose measurement

Method

Stratified modeling-machine learning approach

Tested classifiers and regression models

Results

Error metrics and correlation performance

Clarke error grid analysis (CEGA) and clinical acceptability

JOE International Journal of Online and Biomedical Engineering [Onli](Online-Journals.org)[ne-Jo](Online-Journals.org)[urnals](Online-Journals.org)[.org](Online-Journals.org) iJOE | eISSN: 2626-8493 | Vol. 21 No. 6 (2025) | [https://doi.org/10.3991/ijoe.v21i06.53815](https://doi.org/10.3991/ijoe.v21i06.53815) PAPER A Stratified Modeling-Machine Learning Approach to Improve the Accuracy of Non-Invasive Blood Glucose Estimation Using Photoplethysmography Signals Fitrilina1,2, Muhammad Ilhamdi Rusydi1(􀀍), Rahmadi Kurnia1, Noverika Windasari3, Raffi Zahrandhika Putra2 1Department of Electrical Engineering, Faculty of Engineering, Universitas Andalas, Padang City, Indonesia 2Department of Electrical Engineering, Faculty of Engineering, Universitas Bengkulu, Bengkulu City, Indonesia 3Department of Forensic and Legal Medicine, Faculty of Medicine, Universitas Andalas, Padang City, Indonesia [rusydi@eng.unand.ac.id](rusydi@eng.unand.ac.id) ABSTRACT Diabetes is a silent killer that can only be controlled with continuous monitoring of blood glucose levels. The method commonly used is invasive and has various weaknesses, but it is more accurate than non-invasive methods. This research aims to develop a method to increase the accuracy of non-invasive estimation of blood glucose levels using photoplethysmography (PPG) signals. The proposed method is to carry out stratified modeling-machine learning. The tested classifiers were support vector machines (SVM), KNN, Naïve Bayes, decision tree, and neural network. The prediction model used simple linear, logarithmic, second-order polynomial, exponential, and power regression. Applying stratified modeling using linear regression in the non-diabetes stratum and logarithmic regression in the diabetes stratum obtained a mean absolute relative difference (MARD) value of 4.5%, root mean square error (RMSE) of 18.9 mg/dl, Pearson correlation 0.985 and Clarke error grid analysis (CEGA) 96% in region A and 4% in region B. The implementation of stratification reveals a marked improvement in efficacy, manifested as a reduction in the MARD by 77.83%, a decrease in the RMSE by 51.91%, an enhancement in the Pearson correlation by 0.065, and a CEGA by 100% in regions A and B, thereby being clinically acceptable. Implementing a stratified modeling-machine learning approach can improve the accuracy of non-invasive blood glucose level estimates. KEYWORDS blood glucose non-invasively, machine learning, photoplethysmography (PPG), classification, regression 1 INTRODUCTION The International Diabetes Federation (IDF) identifies diabetes as one of the top ten causes of adult mortality. In 2017, approximately 424.9 million adults (aged 20–79 years) were diagnosed with diabetes, resulting in 4 million deaths and global Fitrilina, Rusydi, M.I., Kurnia, R., Windasari, N., Putra, R.Z. (2025) . A Stratified Modeling-Machine Learning Approach to Improve the Accuracy of Non-Invasive Blood Glucose Estimation Using Photoplethysmography Signals. International Journal of Online and Biomedical Engineering (iJOE), 21(6), pp. 76–96. [https://doi.org/10.3991/ijoe.v21i06.53815](https://doi.org/10.3991/ijoe.v21i06.53815) Article submitted 2024-12-12. Revision uploaded 2025-01-30. Final acceptance 2025-01-30. © 2025 by the authors of this article. Published under CC-BY. 76 International Journal of Online and Biomedical Engineering (iJOE) iJOE | Vol. 21 No. 6 (2025) A Stratified Modeling-Machine Learning Approach to Improve the Accuracy of Non-Invasive Blood Glucose Estimation Using Photoplethysmography Signals healthcare expenditures of USD 727 billion [1] . By 2019, these numbers increased to 463 million cases, 4.2 million deaths, and USD 760.3 billion in expenditures. The global prevalence rose from 8.8% in 2017 to 9.3% in 2019, with projections indicating a continued upward trend. In 2019, Indonesia ranked seventh globally and was the only Southeast Asian country among the top ten nat

FAQs

What problem does the paper address about diabetes monitoring?

It targets the need for continuous blood glucose monitoring while avoiding the drawbacks of invasive methods such as pain and high cost.

How does the proposed method improve non-invasive glucose estimation?

It uses stratified modeling-machine learning, separating non-diabetes and diabetes strata and applying different regression forms to each.

Which models and evaluation metrics are used to assess performance?

The study tests classifiers including SVM, KNN, Naïve Bayes, decision tree, and neural network, and evaluates with metrics such as MARD, RMSE, Pearson correlation, and Clarke error grid analysis (CEGA).

Aug 5, 2026

Upload by Sage