Diabetes management is crucial for millions worldwide and traditional blood glucose monitoring methods often require invasive blood sampling, leading to patient discomfort and poor adherence. This project proposes the development of a non-invasive breath-based glucose monitoring system that leverages gas sensors to detect specific volatile organic compounds (VOCs) in exhaled breath, particularly acetone, which correlates with blood glucose levels. The system will utilize metal oxide semiconductor (MOS) sensors and machine learning algorithms for accurate real-time glucose readings. By eliminating the need for finger pricks, this innovative device aims to enhance the convenience and compliance of glucose monitoring for diabetic patients, ultimately contributing to better disease management and quality of life. The feasibility, accuracy, and usability of the system will be validated through clinical trials, paving the way for future advancements in diabetes care technologies.