Abstract:

Abstract The widespread chronic ailment known as diabetes affects millions of people worldwide. Early detection and an understanding of the underlying reasons can significantly enhance the outcomes for patients and public health initiatives. We propose a non-invasive self-care system that uses IoT and machine learning (ML) to check blood sugar and other critical markers for early diabetes prediction in response to the growing need for IoT-based mobile healthcare applications to anticipate diseases, including diabetes. Our main objective is to offer cutting-edge diabetes management tools that facilitate patient monitoring and technology-aided decision-making. Our objective was to create a hybrid ensemble ML system that used boosting and bagging methods to anticipate the onset of diabetes. In order to collect data from 13,421 participants and validate the model, an offline survey and an online application based on the Internet of Things were utilized. The fifteen items on the form were all about lifestyle, family history, and health. Our ML model performs better than existing methods, according to the experimental findings from both bases, making it a promising method for better diabetes prediction and management. Our technology has the potential to greatly improve early identification and care for those who are at risk of acquiring diabetes around the world by combining the Internet of Things and machine learning.

Keywords: