Endocrine Abstracts

JOINT3278

Introduction: Functional thyroid dysfunction (FTD) refers to medical conditions that alter thyroid hormone balance, ranging from mild hypothyroidism to severe hyperthyroidism. The prevalence varies globally, driven by demographic and diagnostic criteria. Hypothyroidism affects 3–7% and hyperthyroidism 0.5–2% of individuals. Subclinical hypothyroidism and hyperthyroidism are more common, accounting for 10% and 5%, respectively. This highlights the need for a patient-centred approach to diagnosis and management. Artificial intelligence (AI), including machine learning (ML), deep learning (DL), natural language processing (NLP), and computer vision (CV), is advancing healthcare by offering tools to analyze intricate datasets. Electronic medical records (EMRs), the cornerstone of modern clinical practice, store both structured data, such as laboratory values and prescription information, and unstructured data, including clinician notes. The integration of AI with EMRs holds the potential to improve the work performance of healthcare professionals, thereby improving patient outcomes.

Objective: We summarised the highlights of AI’s advancements in EMR systems for the management of FTD.

Methods: A systematic review of medical literature from January 2015 to December 2024 was conducted targeting studies on AI- EMR integration for FTD care. Priority was given to research detailing AI methodologies applicable to FTD, emphasizing data quality, developmental stage, and EMR applications.

Results: AI and EMR synergy has led to innovative algorithms that enhance diagnostic precision by integrating imaging with EMR data, customizing levothyroxine regimens, and predicting complications and comorbidities through longitudinal data analysis. Despite the potential to improve FTD diagnosis and management, personalise treatment, and enhance patient care pathways, its adoption in clinical practice remains limited, highlighting a gap between innovation and real-world application.

Discussion: Translating AI-EMR synergy to practical thyroid care faces challenges, including validation issues that limit generalizability, no standardized and fragmented EMR data infrastructure, and the opacity of AI decision-making processes. Emerging solutions like explainable AI (XAI), federated learning, and hybrid human-AI workflows are promising steps forward.

Conclusion: AI-EMR systems have the transformative potential to shift thyroid care from reactive to proactive, precision-driven, patient-centred, safe, and equitable strategies. Addressing the current limitations through data standardization and harmonization, equity-focused safe AI frameworks, and accelerated real-world validation through prospective clinical trials is crucial. Achieving this vision demands urgent collaborative multidisciplinary efforts among professionals and researchers in endocrinology, clinical practice, data science, epidemiology, and policy-making in FTD care. By prioritizing patient-centred design and rigorous validation, AI-EMR integration can evolve from experimental to essential in optimizing FTD care.