Abstract

Background: As a data-intensive discipline, endocrinology relies heavily on the interpretation of vast, multi-source, and heterogeneous data, such as laboratory metrics, medical images, and unstructured text within electronic health records. Traditional methods face significant challenges in managing this complexity. Artificial intelligence (AI), particularly deep learning and natural language processing (NLP) technologies, is emerging as a key force driving a paradigm shift in the field due to its exceptional capabilities in pattern recognition and automated information extraction.

Objective:This review aims to systematically delineate the pathway through which AI technologies are advancing endocrinology from automated data processing towards intelligent clinical decision support. It provides a comprehensive overview of the current application landscape and offers a critical analysis of the core challenges and future directions.

Methods: By systematically examining and synthesizing existing literature, this review focuses on key application scenarios of AI in endocrinology. These scenarios encompass NLP-based mining and structuring of clinical text, the development of disease risk prediction models, intelligent interpretation of medical images, drug discovery and safety monitoring, and innovative practices in public health management.

Results:The study finds that AI is deeply integrated across the entire lifecycle of endocrine disease management. It can automatically extract key clinical information from free text, enable accurate prediction of risks for conditions like diabetes and its complications, demonstrate high precision in analyzing images such as thyroid ultrasounds, and accelerate drug target discovery and adverse reaction monitoring. However, the clinical translation in this field continues to face substantial challenges, including issues of data quality and privacy, algorithmic "black box" issues and interpretability, insufficient model generalizability, and difficulties in clinical workflow integration alongside ethical and regulatory hurdles.

Conclusion: AI is undoubtedly reshaping the practice of modern endocrinology, propelling it towards new heights of precision and personalized medicine. However, to successfully transition from "promising models" to "routine clinical application," future efforts must focus on generating high-level evidence through prospective clinical trials, developing human-centric, explainable, and unbiased AI tools, and fostering interdisciplinary collaboration ecosystems alongside establishing appropriate ethical and regulatory frameworks.