International Journal of Scientific Research and Engineering Development

Diabetes mellitus is a rapidly increasing metabolic disorder that requires early and reliable detection to avoid long-term health complications. Conventional diagnostic techniques are mainly based on blood analysis, which is invasive, uncomfortable, and unsuitable for frequent screening. To overcome these limitations, this work presents a non-invasive diabetes detection system using an AI-based breath analyzer. The proposed system focuses on analyzing acetone concentration present in exhaled human breath, which is a known biomarker associated with abnormal glucose metabolism. Breath samples are collected through a compact breath chamber and analyzed using a gas sensor capable of detecting volatile organic compounds. The acquired sensor data is processed using a Raspberry Pi and classified using a machine learning model trained on pre-recorded breath acetone values. The model predicts whether the individual belongs to normal, pre-diabetic, or diabetic categories. The system is designed to be portable, cost-effective, and easy to use, enabling rapid and painless screening. This approach is especially beneficial for rural and remote areas where access to clinical diagnostic facilities is limited. The proposed solution demonstrates the potential of combining sensor technology and artificial intelligence for reliable, real-time, non-invasive diabetes detection.

Srijyothi P, Praveen Chila, Rohith S, Sahana Ramesh P, Tejas Gowda K C, "Diabetes Detection Using AI Breath Analyzer" International Journal of Scientific Research and Engineering Development, V8(6): Page(2527-2536) Nov-Dec 2025. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.