Developing a Non-Invasive Intelligent System for Blood Glucose Level Estimation Using Reflective Optical Sensors

Résumé

This paper introduces a non-invasive system for blood glucose monitoring using a reflective optical sensor (TCRT5000) integrated with an Arduino platform. The system integrates a detection unit equipped with an infrared LED, phototransistor, microcontroller, OLED display, and Bluetooth module-with realtime data visualization and wireless transmission to mobile devices. The TCRT5000 sensor detects light reflectivity changes caused by blood flow, indirectly estimating glucose levels without the discomfort of invasive techniques. Validation was conducted by comparing system readings with a standard glucometer under varied conditions. The system achieved high performance metrics, including 97.14% accuracy, 97.9% precision, 98.59% sensitivity, and 90.91% specificity. Minor variations were observed, attributable to environmental and individual differences. Despite this, the results confirm the device’s reliability and usability, making it a promising alternative for frequent and accessible glucose monitoring. Advantages include affordability, ease of use, and adaptability, which are integral to encouraging patient compliance and improving health outcomes. Future developments may incorporate advanced optical sensors such as NIR spectroscopy or Raman technologies, alongside machine learning algorithms, to improve accuracy and extend functionality. The proposed system highlights the potential for non-invasive methods to revolutionize diabetes care in both personal and clinical settings.