Generative artificial intelligence in diabetes healthcare

Abstract

The rapid advancement of generative artificial intelligence (AI) has been fueled by breakthroughs in large language models and applications across diverse domains, from creative content to scientific discovery. Its strength lies in modeling, simulating, and generating high-fidelity data. In diabetes care, generative AI enables solutions to challenges such as data scarcity, patient variability, and personalization. This article explores key deep generative models, including variational autoencoders, generative adversarial networks, transformers, and diffusion models applied to tabular, time series, image, and text data. These models enable synthetic patient data generation, dataset augmentation, glucose-insulin dynamics simulation, and the development of virtual coaches and digital twins. Despite these advances, challenges persist, including model instability, high data requirements, and output interpretability. This article reviews the current literature and outlines opportunities, limitations, and ethical considerations for the use of generative AI in diabetes healthcare.

Ähnliche Arbeiten