FROM SENSORS TO INSIGHTS: A SCOPING REVIEW OF AI/ML APPLICATIONS IN WEARABLE HEALTH MONITORING FOR DIABETES

Abstract

Advancements in wearable health technologies, combined with artificial intelligence (AI) and machine learning (ML), are revolutionizing how diabetes is monitored and managed. This paper presents a detailed review of cutting-edge AI/ML techniques utilized in wearable systems, with a particular emphasis on noninvasive, real-time blood glucose monitoring methods. The review explores the application of wrist-based photoplethysmography (PPG) signals for estimating glucose levels, along with the integration of IoT-enabled continuous glucose monitoring (CGM) systems. A systematic analysis was performed using databases such as IEEE Xplore, PubMed, Scopus, and Web of Science, selecting studies that demonstrate strong innovation, technical robustness, and clinical importance. The review identifies significant improvements in the precision and efficiency of AI/ML algorithms, as well as progress toward their practical implementation. However, challenges such as data variability, signal interference, and effective sensor fusion remain. The potential for these technologies to support early diagnosis, enable individualized treatment plans, and enhance patient care is substantial. The findings highlight the importance of collaborative, interdisciplinary research to overcome current limitations and bring these innovations into mainstream clinical use.