Related Experiment Video
Updated: Jun 28, 2026
08:27
Highly Sensitive and Rapid Fluorescence Detection with a Portable FRET Analyzer
Published on: October 1, 2016
9.2K
Detection of fasting blood sugar using a microwave sensor and convolutional neural network
Mohammad Amir Sattari1, Mohsen Hayati2
1Electrical Engineering Department, Faculty of Engineering, Razi University, Kermanshah, Iran.
Scientific Reports
|July 2, 2025
Summary
This study introduces a novel microwave sensor for non-contact blood glucose monitoring. Combined with a convolutional neural network, it accurately estimates fasting blood sugar levels, paving the way for wearable diabetes management.
Area of Science:
- Biomedical Engineering
- Sensor Technology
- Artificial Intelligence
Background:
- Fasting blood sugar (FBS) monitoring is crucial for diabetes management.
- Conventional methods require invasive blood sampling.
- Microwave sensing offers a promising non-contact alternative for glucose detection.
Purpose of the Study:
- To develop and validate a miniaturized microstrip microwave sensor for non-contact FBS detection.
- To utilize deep learning for accurate interpretation of sensor data.
- To assess the feasibility of this technology for wearable health systems.
Main Methods:
- A miniaturized microstrip microwave sensor was designed and fabricated.
- FBS levels were measured in 78 individuals using a clinical auto-analyzer.
- Sensor responses (S21) across 30 kHz-18 GHz were recorded for each sample.
- A convolutional neural network (CNN) was trained on sensor data for FBS estimation.
Main Results:
- The developed microwave sensor demonstrated high sensitivity to FBS levels.
- The CNN model achieved a mean relative error (MRE) of 1.31% in FBS estimation.
- The study confirmed the feasibility of non-contact glucose measurement using this integrated approach.
Conclusions:
- Combining microwave sensing with CNN provides a reliable method for non-contact FBS measurement.
- This technology shows significant potential for integration into user-friendly wearable diabetes monitoring devices.
- The approach reduces the need for frequent conventional blood sampling, improving patient convenience.
