Illuminating the Path to Thyroid Disorder Management Using Artificial Intelligence: A Narrative Review

Dysregulation of thyroid function, manifesting as hyperthyroidism or hypothyroidism, can profoundly impact an individual's overall health. In this context, artificial intelligence (AI) applications have the potential to revolutionize diagnostic approaches, treatment strategies, and patient monitoring.

This study comprehensively reviews the latest literature on AI applications in thyroid functional and autoimmune disorders.

An online search was conducted on databases using search queries crafted with MeSH terms related to AI and thyroid disorders. After screening, studies aligned with our research focus were selected for this narrative review.

Multiple studies have explored the use of AI technologies, including machine learning (ML) and deep learning (DL), to enhance laboratory workflows for thyroid function tests (TFT) and improve the accuracy of TFT interpretation by incorporating clinical data. In imaging, DL-based models have demonstrated the potential to assist less experienced radiologists in interpreting scintigraphy and ultrasound images. Artificial intelligence has also provided valuable insights into identifying diagnostic genes for thyroid-related autoimmune disorders and understanding the effects of environmental factors, such as chemicals, on thyroid gland function. Some ML models have been developed to predict the risk of hypothyroidism following radioiodine therapy (RAI). Furthermore, AI has shown promise in personalized levothyroxine dose adjustments, predicting treatment responses, and accurately diagnosing complications such as thyroid-associated ophthalmopathy (TAO). Finally, ML-based models forecasting the risk of suicide attempts in patients with major depressive disorder (MDD) and predicting pregnancy outcomes, such as gestational diabetes mellitus (GDM) and preterm delivery, based on TFT results, appear beneficial in addressing these significant health issues.

The current state of AI in diagnosing and treating thyroid function disorders is promising, with applications primarily focused on improving diagnostic accuracy, consistency, and personalized treatment approaches. However, challenges remain that prevent these models from fully substituting professionals. Addressing these challenges is crucial to ensure AI effectively contributes to the management of patients with thyroid diseases.