Machine learning in endocrinology: current applications and future perspectives.

Area of Science:

• Endocrinology
• Computational Biology
• Medical Informatics

Background:

• Endocrinology research is increasingly leveraging machine learning (ML) for data analysis.
• ML can uncover complex patterns in large datasets, aiding in understanding endocrine disorders.
• Applications of ML in healthcare promise optimized outcomes and novel insights.

Purpose of the Study:

• To review basic machine learning concepts relevant to endocrinology.
• To highlight specific endocrine areas with significant potential for ML application.
• To provide an overview of ML methodologies in endocrine research.

Main Methods:

• A narrative review approach was employed.
• Systematic literature search conducted for studies published between January 2000 and December 2024.
• Focus on endocrine conditions analyzed using ML statistical methods.

Main Results:

• Analysis of 1130 studies revealed thyroid research as most common, followed by pituitary, adrenal, and parathyroid studies.
• ML applications span medical imaging, tumor classification, treatment response prediction, risk estimation, and biomarker identification.
• Significant ML application in diagnosis and understanding of endocrine diseases.

Conclusions:

• Machine learning holds substantial potential to improve diagnosis, treatment, and understanding of endocrine diseases.
• Current limitations include lack of model transparency, data imbalance, and clinical implementation challenges.
• Further validation, interdisciplinary collaboration, and standardization are crucial for safe and effective ML integration in endocrinology.