Reliable Noninvasive Glucose Sensing via CNN-Based Spectroscopy

2025-06-15

AI Summary

To address the longstanding challenges of low accuracy, high cost, and poor wearability in noninvasive blood glucose monitoring (NIBGM), this work proposes a dual-modal fusion sensing system. The first modality employs short-wave infrared (SWIR) multi-wavelength imaging coupled with a convolutional neural network (CNN) to extract spatial absorption features—introducing the novel SWIR-CNN spatial modeling approach. The second modality utilizes miniaturized photodiodes to acquire normalized optical signals, integrated with lightweight random forest regression for real-time prediction. Both modalities are rigorously validated on a unified synthetic blood/skin phantom platform. Experimental results show the SWIR-CNN modality achieves a mean absolute percentage error (MAPE) of 4.82% and 100% Clarke Zone A coverage, while the photodiode modality attains 86.4% Zone A accuracy—surpassing state-of-the-art noninvasive solutions. This study is the first to simultaneously achieve clinical-grade accuracy, practical wearability, and low system cost (<$200), establishing a deployable technical framework for NIBGM.

Technology Category

• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Multimodal Learning

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

In this study, we present a dual-modal AI framework based on short-wave infrared (SWIR) spectroscopy. The first modality employs a multi-wavelength SWIR imaging system coupled with convolutional neural networks (CNNs) to capture spatial features linked to glucose absorption. The second modality uses a compact photodiode voltage sensor and machine learning regressors (e.g., random forest) on normalized optical signals. Both approaches were evaluated on synthetic blood phantoms and skin-mimicking materials across physiological glucose levels (70 to 200 mg/dL). The CNN achieved a mean absolute percentage error (MAPE) of 4.82% at 650 nm with 100% Zone A coverage in the Clarke Error Grid, while the photodiode system reached 86.4% Zone A accuracy. This framework constitutes a state-of-the-art solution that balances clinical accuracy, cost efficiency, and wearable integration, paving the way for reliable continuous non-invasive glucose monitoring.

Problem

• Develop noninvasive glucose sensing using SWIR spectroscopy and AI
• Achieve accurate glucose monitoring with CNN and photodiode systems
• Enable wearable integration for continuous glucose level tracking

Innovation

• CNN-based SWIR spectroscopy for glucose sensing
• Compact photodiode with machine learning regressors
• Dual-modal AI framework for noninvasive monitoring