Development and validation of a multidimensional machine learning-based nomogram for predicting central lymph node...

Area of Science:

• Oncology
• Medical Informatics
• Surgical Pathology

Background:

• Papillary thyroid microcarcinoma (PTMC) presents a diagnostic challenge, with central lymph node metastasis (CLNM) increasing recurrence risk.
• Current prediction models for CLNM in PTMC lack comprehensiveness, integrating only limited clinical or imaging data.
• This limits effective preoperative risk stratification and personalized surgical planning for PTMC patients.

Purpose of the Study:

• To develop and validate a machine learning-based nomogram for predicting CLNM in PTMC.
• To integrate multidimensional predictors including clinicopathological, ultrasonographic, and serological features.
• To enhance preoperative risk stratification and guide personalized surgical decision-making for PTMC.

Main Methods:

• A retrospective analysis of 503 PTMC patients undergoing thyroidectomy was performed.
• Least Absolute Shrinkage and Selection Operator (LASSO) regression identified key predictors for a logistic regression model.
• Model performance was assessed using ROC curves, calibration plots, and decision curve analysis (DCA).

Main Results:

• CLNM was confirmed in 28.8% of patients; age, gender, tumor size, location, and ETE were significant predictors.
• The nomogram demonstrated strong predictive performance with AUCs of 0.88 (training) and 0.78 (validation).
• Excellent calibration and clinical utility were observed, supporting its applicability in PTMC risk assessment.

Conclusions:

• The developed machine learning nomogram is a reliable tool for assessing CLNM risk in PTMC patients.
• This predictive model supports the implementation of personalized surgical strategies for PTMC management.
• Further external validation is recommended to confirm the generalizability of the nomogram.