Non-Invasive Glucose Monitoring Using Optical Sensors and Machine Learning: A Predictive Model for Nutritional and Health Assessment
Glucose monitoring plays a vital role in maintaining metabolic and nutritional balance. However, invasive methods, while accurate, are often painful and impractical for routine use. This study presents a non-invasive glucose monitoring framework that integrates a high-intensity Superbright optical s...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11088090/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Glucose monitoring plays a vital role in maintaining metabolic and nutritional balance. However, invasive methods, while accurate, are often painful and impractical for routine use. This study presents a non-invasive glucose monitoring framework that integrates a high-intensity Superbright optical sensor with an IoT-enabled data acquisition system using ESP32 microcontrollers and Raspberry Pi. The IoT-based architecture enables seamless integration with cloud computing platforms, allowing remote access and scalability for large-scale population-level screening and monitoring. The system captures glucose-related optical signals, which are analyzed using various machine learning algorithms, including a novel Convolutional Neural Network–Attention Hybrid Model (CNN-AHM). CNN-AHM combines spatial feature extraction with attention-based prioritization of relevant signal patterns, enhancing both accuracy and interpretability. Among all models tested, CNN-AHM demonstrated the best standalone performance (RMSE =14.06 mg/dL), representing a 13.18% improvement over its baseline configuration. When combined with complementary strength of the second-best performing Random Forest Regressor (RF-R), the system achieved a substantially improved RMSE of 7.62 mg/dL and an R2 of 0.83. This work underscores the importance of integrating sensor technology, deep learning, and explainable modeling to advance non-invasive glucose monitoring solutions. Future directions include extending the architecture with explainable AI using SHAP (SHapley Additive Explanations) to provide feature-level interpretability, and exploring real-world deployment scenarios. |
|---|---|
| ISSN: | 2169-3536 |