Machine Learning for Thyroid Cancer Detection, Presence of Metastasis, and Recurrence Predictions-A Scoping Review

Area of Science:

• Endocrinology
• Oncology
• Medical Informatics

Background:

• Thyroid cancer (TC) is a common endocrine malignancy where early detection is crucial.
• Conventional diagnostic methods for TC have limitations.
• Machine learning (ML) offers potential to improve TC patient care through enhanced prediction and risk stratification.

Purpose of the Study:

• To conduct a scoping review of ML applications in thyroid cancer research.
• To map the existing literature on ML use with clinical data and Electronic Medical Records (EMRs) for TC.
• To identify trends, challenges, and future directions for ML in TC.

Main Methods:

• Systematic search and analysis of scientific literature.
• Screening of 1231 papers, evaluation of 203 full-text articles, and selection of 21 relevant studies.
• Categorization of ML applications into malignancy prediction, metastasis prediction, and recurrence/survival prediction.

Main Results:

• Identified key ML applications in TC, including nodule classification and malignancy prediction.
• Reviewed studies on predicting other metastases derived from TC.
• Examined ML's role in predicting recurrence and survival for TC patients.
• Synthesized findings on ML-driven TC research trends and challenges.

Conclusions:

• ML demonstrates significant potential to improve diagnostic accuracy, risk stratification, and prognosis prediction in thyroid cancer.
• Further research is needed to enhance the clinical integration of ML for precision medicine in TC.
• ML can transform thyroid cancer patient care by reducing errors and improving outcomes.