Harmonizing Radiomics and Interpretable AI: Precision and Transparency in Oncological Prognostication

Abstract

This research aims to improve the understandability and trustworthiness of prediction models in oncological prognostication by incorporating advanced methods like SHAP (SHapley Additive exPlanations). By integrating multi-modal radiomic data, which includes a wide range of variables collected from medical images, we have developed predictive models that capture the intricate nature of illness progression. Clinicians can gain a visual understanding of feature effects, facilitating validation and informed decision-making, by utilizing SHAP summary plots and individual instance explanations. This research provides doctors with a clear and logical basis for making precise predictions. It strengthens trust and collaboration between computational technologies and medical competence.

Ähnliche Arbeiten