Enhancing cancer prognostics with group penalty models: a comparative study on radiomics feature selection in lung adenocarcinomas and meningiomas

Key Points

• Group penalty models provide greater feature selection stability and maintain predictive performance.
• The study compares group penalty models with Lasso in selecting features for cancer prognosis.
• Using biologically meaningful feature groups enhances interpretation and clinical reasoning in radiomics.
• This analysis highlights the potential for improved cancer prognostics based on group penalty modeling.