Related Experiment Video
Updated: May 10, 2026
12:02
Automated Modular High Throughput Exopolysaccharide Screening Platform Coupled with Highly Sensitive Carbohydrate Fingerprint Analysis
Published on: April 11, 2016
12.1K
An accurate glucose detection platform using colorimetry and supervised learning algorithms
Mithun Kanchan1, Pragna Harish1, Omkar S Powar1
1Manipal Institute of Technology, Manipal Academy of Higher Education, Manipal, Karnataka 576104, India.
Biomedical Physics & Engineering Express
|March 9, 2026
Summary
This study presents an affordable, precise Point-of-Care diagnostic platform for rapid blood glucose monitoring using microfluidics and image analysis. The system achieves high accuracy, offering a scalable solution for diverse healthcare settings.
Area of Science:
- Biomedical Engineering
- Analytical Chemistry
- Medical Diagnostics
Background:
- Accurate blood glucose monitoring is crucial for managing diabetes and preventing complications.
- Existing methods can be costly, time-consuming, or require specialized equipment.
- Point-of-Care (POC) diagnostics offer potential for rapid, decentralized glucose assessment.
Purpose of the Study:
- To develop an affordable, reliable, and precise POC diagnostic platform for glucose detection.
- To integrate microfluidic and colorimetric principles with image analysis for glucose estimation.
- To create a user-friendly system suitable for real-time monitoring in various healthcare settings.
Main Methods:
- Fabrication of a custom microfluidic chip for enzymatic color reactions with minimal sample volume (~20 µL).
- Development of a compact, 3D-printed imaging module with a high-resolution camera for stable image acquisition.
- Application of supervised machine learning models (Random Forest, SVM, KNN, FNN) on engineered image features for glucose level prediction.
Main Results:
- The Random Forest model achieved 98% cross-validation precision and near 100% specificity for glucose level distinction.
- The system demonstrated rapid color development within 3-4 minutes and minimal misclassification (mean AUC near 1).
- The platform is USB-powered, compatible with embedded systems/laptops, and eliminates the need for smartphones or external calibration.
Conclusions:
- The proposed image-based glucose estimation approach is a cost-effective, scalable, and accurate POC solution.
- The system's low reagent consumption, rapid analysis, and ease of operation are advantageous for decentralized screening and resource-limited settings.
- Future work includes expanding concentration ranges, clinical validation, and automated calibration for enhanced usability.
