Advanced Artificial Intelligence and Machine Learning Driven Data Analyses in Diabetes Mellitus Research

Diabetes mellitus is an endocrine disorder of global significance. This PhD research project harnesses the capabilities of machine learning techniques for the meticulous investigation of data associated with diabetes mellitus. The formidable prevalence of the disease across the world, alongside the consequential burdens it imposes on healthcare systems, underscores the paramount importance of this research. The research is characteristically subdivided into three focal domains: predictive analysis, glucose quantification, and risk assessment relating to diabetes. Specifically, the research delves into advanced deep learning architectures for forecasting blood glucose levels, proposes methodologies for improving glucose quantification, and provides thorough risk assessments for COVID-19 patients with pre-existing diabetes mellitus. For each of these three domains, the research deploys state-of-the-art machine learning algorithms as a powerful apparatus to navigate the complexities of diabetes data, culminating in two research publications in reputable, peer-reviewed academic journals. Each publication illuminates the transformative potential of machine learning as a conduit for novel advancements within the respective domain. This in turn contributes to a more nuanced understanding of the disease, enhancement of patient care, and optimisation of healthcare resource allocation. Composed in a publication format, this dissertation is structured as a compilation of the six resultant articles, which are interconnected within the overarching framework of machine learning applications in diabetes research. As a whole, this extensive exploration of diabetes data through machine learning pipelines proffers novel insights and aims to make a substantial contribution to the academic field.