Development and External Validation of Machine Learning-based Models for Predicting Survival Outcomes in Endometrial...

Area of Science:

• Oncology
• Medical Informatics
• Biostatistics

Background:

• Endometrial cancer prognosis is challenging, especially for recurrent cases.
• Accurate prognostication is crucial for effective patient management.
• Machine learning offers potential for improved cancer survival prediction.

Purpose of the Study:

• To develop and validate machine learning models for predicting endometrial cancer survival.
• To assess the accuracy of these models in forecasting overall survival (OS) and cancer-specific survival (CSS).

Main Methods:

• Utilized the Surveillance, Epidemiology, and End Results (SEER) database for model construction.
• Incorporated patient demographics, pathological, and therapeutic factors.
• Validated models using both internal (SEER) and external datasets.

Main Results:

• Models demonstrated strong predictive performance for OS and CSS.
• The best model achieved an AUC of 0.83 for OS in internal validation and 0.85 in external validation.
• Models showed higher accuracy in predicting CSS compared to OS.

Conclusions:

• Machine learning models can effectively predict prognosis in endometrial cancer patients.
• These predictive tools can support clinical decision-making and patient care strategies.