Research into the use of artificial intelligence (AI) in the diagnosis and intervention of metabolic diseases has seen significant growth, with the number of publications increasing by 394% over the past 5 years. An analysis of 1,059 scientific papers from the WOSCC and Scopus databases showed that China (21.87%), the USA (17.10%) and the United Kingdom have the largest share of publications. The most prominent institutions include Harvard Medical School, which has the strongest academic influence, and Chinese institutions with the highest number of publications (298). Key research areas include machine learning, deep learning, data mining, metabolomics and metabolic disease diagnosis. The study identified three main directions for future research: prevention using smart devices, multimodal diagnostic methods, and intervention strategies guided by large language models (LLM). These findings indicate a significant shift in the management of metabolic diseases thanks to AI technologies and create the basis for the further development of intelligent diagnostic and therapeutic systems.