Executive Summary : | Type-2 diabetes and Atrial Fibrillation (AF) are significant risk factors for brain stroke, with developing countries bearing the burden of stroke more than developed countries. With 462 million people worldwide expected to have type 2 diabetes, it is the 9th leading cause of morbidity and 7th leading cause of DALYs. Both diabetes and AF place a significant economic strain on global healthcare systems, particularly in developing nations. Current methods for estimating blood glucose concentration (BGC) rely on invasive blood tests and glucometers, which are inconvenient for diabetic and stroke patients. There is no accurate non-invasive and affordable alternative for BGC estimation. Currently, monitoring of AF is done via Holter recorders, which requires several probes over the patient's body and costs 1-3.5 Lakhs. A non-invasive method for continuous BGC estimate can save diabetic and stroke patients from multiple daily pricks and alert them or their carers to unexpected BGC level changes, allowing them to intervene to prevent medical casualties. A wearable device with continuous BGC monitoring and AF detection capabilities can help understand stroke aetiology and prevent future incidences. Smart wearable devices equipped with multiple sensors can be integrated with wearable gadgets to monitor body wellbeing. This study aims to develop an AI/ML-based non-invasive method of BGC estimation and identification of AF by leveraging the capabilities of AI/ML models trained over data from PPG, ECG, gyroscope, and accelerometer sensors. Continuous glucose monitor (CGM) and Holter will be considered the gold standard for all models. |