A Data-Driven Approach to Refine Predictions of Differentiated Thyroid Cancer Outcomes: A Prospective Multicenter...

Area of Science:

• Endocrinology
• Oncology
• Medical Informatics

Background:

• Differentiated thyroid cancer (DTC) risk stratification is vital for clinical decisions.
• The 2015 American Thyroid Association (ATA) guidelines are the standard for assessing recurrence risk.
• Novel features and existing feature relevance are areas of ongoing research.

Purpose of the Study:

• To develop a comprehensive, data-driven model for predicting persistent/recurrent DTC.
• To identify and weigh the impact of all available predictive features.
• To improve upon existing risk stratification methods.

Main Methods:

• A prospective cohort study utilizing the Italian Thyroid Cancer Observatory (ITCO) database (NCT04031339).
• Inclusion of 4773 consecutive DTC patients with follow-up data from 40 Italian centers.
• Development of a decision tree model to assign patient risk indices and analyze variable impact.

Main Results:

• The decision tree model demonstrated superior performance compared to the ATA risk stratification.
• Sensitivity for high-risk classification of structural disease increased from 37% to 49%.
• Negative predictive value for low-risk patients improved by 3%, with significant contributions from age, BMI, tumor size, sex, family history, surgical approach, cytology, and diagnosis circumstances.

Conclusions:

• Existing risk stratification systems for DTC can be enhanced by incorporating additional variables.
• Improved prediction of treatment response is achievable through more precise patient clustering.
• A comprehensive dataset is key to refining patient stratification in DTC.