Volatile organic compounds (VOCs) are helpful biomarkers because they can reveal various diseases’ underlying pathophysiology and physical abnormalities. Analyzing exhaled breath is a simple and non-invasive way to monitor metabolic rate. Diabetes is a complex metabolic disease. Oxidative stress, inflammatory syndrome, diabetes, and hypertension are all intricately linked. This research utilizes an IoT-based breath analyzer to collect real-time data on diabetes and healthy persons. The study aims to design and develop its IoT-based breath analyzer device to identify the exhaled breath gases of diabetes persons. This method was used to test a group of 26 people with diabetes and 17 people without the disease. The data is gathered through the Android module and serially acquired through the USB cable. The Cool-Term open-source software is used for data acquisition. After the real-time data collection, the disease characteristics are assessed by a deep learning system. The model successfully distinguishes between diabetes and non-diabetes exhaled breath samples with a 97% success rate.