A machine learning model utilizing Delphian lymph node characteristics to predict contralateral central lymph node...

Area of Science:

• Oncology
• Radiology
• Artificial Intelligence

Background:

• Papillary thyroid carcinoma (PTC) poses challenges in predicting contralateral central lymph node metastasis (CCLNM).
• Accurate prediction of CCLNM is crucial for determining appropriate surgical intervention in PTC patients.

Purpose of the Study:

• To develop an AI-driven model for predicting CCLNM in PTC.
• To identify effective predictors of CCLNM by integrating radiological and clinical pathological data.
• To establish a clinically applicable model for guiding surgical extent in PTC.

Main Methods:

• A prospective cohort study involving 603 PTC patients from three centers.
• Development of a machine learning (ML) model using logistic regression and other algorithms.
• External validation of the final Random Forest model for predicting CCLNM.

Main Results:

• Multiple pathological factors were associated with CCLNM.
• Delphian lymph node metastasis ratio, ipsilateral metastasis number, and presence of ipsilateral metastasis were independent risk factors.
• The Delphian lymph node-CCLNM (D-CCLNM) model achieved high predictive accuracy with AUCs ranging from 0.8907 to 0.9273.

Conclusions:

• A novel ML-based method effectively predicts CCLNM in PTC patients.
• The D-CCLNM model integrates Delphian lymph node data and clinical characteristics.
• This approach provides a foundation for guiding surgical decisions in clinical settings.