Analysis demonstrates a predictive model may improve pneumothorax risk assessment in lung cancer patients, highlighting the value of integrated radiomics and clinical data.
Key Points
The comprehensive predictive model achieves a higher AUC of 0.9262 compared to individual models, enhancing accuracy for pneumothorax prediction.
Model evaluation used ROC curves and DeLong's test, showcasing significant performance improvements over both the clinical and radiomics models.
Retrospective analysis involved data from 111 lung cancer patients undergoing microwave ablation for thorough model development.
Findings indicate the potential for improved diagnosis of pneumothorax, supporting better patient outcomes with accurate prediction methods.