A blood glucose monitoring and prediction system based on light source compensation

View at publisher Summarise with VIRAT
Authors: Xinrui Shao, Linling Gong
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Non-invasive Blood Glucose Monitoring (NBGM) offers a painless alternative to traditional invasive monitoring methods, yet its accuracy is limited by light source instability. This paper presents a novel monitoring and prediction system with real-time light source compensation. The sensor of the system adopts a dual-calibration strategy: on one hand, it achieves intensity stabilization through Adaptive Gain Control (AGC); on the other hand, it performs spectral correction using a standard whiteboard reference. In terms of data processing, a LSTM model based on PCA method is developed in the study: Principal Component Analysis (PCA) is used to reduce the dimensionality of optical features and remove noise, while the Long Short-Term Memory (LSTM) network is responsible for blood glucose prediction. Experiments validation on 50 volunteers show that the system achieves MAE=0.32 mmol/L, RMSE=0.45 mmol/L, and R²=0.93, outperforming traditional NBGM systems. In addition, the LSTM model based on PCA method outperforms traditional models such as Support Vector Regression (SVR) in terms of prediction stability. The system can effectively suppress the impact of light source variations, improving the accuracy and reliability of non-invasive blood glucose monitoring. It has potential applications in home healthcare and clinical diabetes management.

Fields of Study & Subjects

Spectroscopy Techniques in Biomedical and Chemical Research Optical Imaging and Spectroscopy Techniques Non-Invasive Vital Sign Monitoring

Publication details

Journal / source
International Conference on Imaging Processing for Medical-Oriented Recognition (IPOR)
Pages
10–10
Publication date
19 March 2026
Language
EN