<![CDATA[Diabetes is a chronic disease with a high prevalence in Indonesia, requiring routine blood glucose monitorin1. However, the standard method for measuring Glycated Hemoglobin (HbA1c) is invasive, painful, and costly. This study aims to summarize and discuss the non-invasive estimation of HbA1c levels using Photoplethysmography (PPG) signals. PPG, a non-invasive optical technique, detects microvascular blood volume changes. Its pulse wave morphology is affected by biomechanical and hemodynamic alterations due to HbA1c accumulation, such as increased arterial stiffness. Various studies have explored the extraction of PPG signal features (statistical, physiological, and AC/DC ratio), which are then processed using machine learning and deep learning algorithms like 1D-CNN, XGBoost, Random Forest, and QSVM. The results demonstrate promising performance, with some models achieving Pearson correlation coefficients up to R = 0.96 and a clinical accuracy of 100% estimation points falling within Zone A of the Clarke Grid Analysis (CGA). The non-invasive approach based on PPG and artificial intelligence offers an accurate, fast, and comfortable solution for HbA1c monitoring, marking a crucial advancement in diabetes management.]]>
Copyrights © 2025
