PREDICTING BLOOD GLUCOSE LEVELS WITH MACHINE LEARNING AND IOT: A META-ANALYSIS AND FUTURE DIRECTIONS IN IOMT DATA FUSION FOR HEALTHCARE TRANSFORMATION

Abstract

The emergence of machine learning and medical IoT is changing healthcare, especially when it comes to managing diseases such as diabetes. Integration of machine learning algorithms and IoT device data offers a promising opportunity to improve glucose management models in context. At the same time, the Internet of Things is changing culture by enabling continuous monitoring, remote communication and increasing efficiency, thus changing delivery and service management. This comprehensive study aims to evaluate the effectiveness of machine learning models using IoT device data to predict blood sugar levels through a meta-analysis. It also examines recent developments, challenges, and future directions for data integration and management in the context of IoMT, highlighting its potential for healthcare reform. We searched electronic databases (such as Scopus, Springer, IEEE Xplore, PubMed, CINAHL, Embase, Web of Science, and Nature) for studies published between 2019 and 2023. Performance of machine learning models for predicting blood glucose. Studies that did not include machine learning models or performance measurements were excluded. The assessment was employed to assess study quality. Our primary outcomes included a comparison of ML models for BG-level prediction across different prediction horizons (PHs). Ten eligible studies were analyzed, focusing on BG prediction across PHs of 15, 30, 45, and 60 minutes. The ML models demonstrated mean absolute root mean square error (RMSE) values of 15.02 (SD 1.45), 21.488 (SD 2.92), 30.094 (SD 3.245), and 35.89 (SD 6.4) mg/dL, respectively. Among these, the Random Forest (RF) model exhibited superior performance across all prediction horizons. Alongside these findings, advancements in IoMT have shown significant benefits, such as enhanced disease monitoring, prevention, care, and diagnosis. However, challenges in managing and securely storing vast amounts of patient data and ensuring data privacy and security persist. The integration of blockchain technology and cloud computing is emerging as a promising solution to these challenges.