Predicting the prognosis of epithelial ovarian cancer patients based on deep learning models
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.This study developed prognostic models to predict overall survival in epithelial ovarian cancer patients. The DeepSurv machine learning model demonstrated the best predictive performance, aiding clinical decision-making.
Area Of Science
- Oncology
- Biostatistics
- Machine Learning in Healthcare
Background
- Epithelial ovarian cancer (EOC) presents high mortality and morbidity rates.
- Accurate prognostic models are crucial for improving survival outcomes in EOC patients.
Purpose Of The Study
- To develop and validate prognostic models for predicting overall survival (OS) in epithelial ovarian cancer (EOC) patients.
- To compare the performance of machine learning (ML) models against traditional regression methods.
Main Methods
- Utilized SEER database (N=10902) and a hospital-based cohort (N=116) for model development and validation.
- Employed COX regression for prognostic factor identification and developed three ML models (including DeepSurv) and a nomogram.
- Evaluated models using C-index, ROC curves, calibration curves, and decision curve analysis (DCA).
Main Results
- Identified 12 independent prognostic factors for OS in EOC patients.
- The DeepSurv model exhibited superior performance, with a C-index of 0.715 (internal) and 0.672 (external validation).
- Achieved high accuracy in predicting 3- and 5-year survival rates, with ROC curves ranging from 0.731 to 0.766.
Conclusions
- Successfully developed and validated multiple predictive models, including a nomogram and ML-based approaches.
- The DeepSurv model shows significant potential as a clinical decision support tool for EOC patient management.
- These models offer valuable instruments for predicting overall survival and guiding treatment strategies in epithelial ovarian cancer.
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.