Machine learning-based prediction of central lymph node metastasis in unifocal papillary thyroid microcarcinoma.

OBJECTIVE

This study aims to develop a machine learning (ML) model to predict the risk of central lymph node metastasis (CLNM) in patients with papillary thyroid microcarcinoma (PTMC) using a combination of clinical and ultrasound features.

METHODS

Multiple ML models were integrated, with least absolute shrinkage and selection operator regression applied for feature selection and a LightGBM model optimized for prediction. Clinical and ultrasound features were used to construct the predictive model.

RESULTS

The model demonstrated high predictive accuracy in the validation cohort, with an area under the curve of 0.87. Key features associated with CLNM risk included tumor size, extrathyroidal extension and vascularization.

CONCLUSIONS

The ML model showed strong potential for predicting CLNM in PTMC, and interpretability analysis enhanced model transparency. These findings provide valuable support for personalized treatment strategies in clinical practice.