Development and Validation of a Multiomics Model Using Clinical and Radiomics Features for Predicting Hemorrhagic Transformation in Stroke

Motivation: Early identification of hemorrhagic transformation (HT) is crucial for optimizing treatment in patients with acute ischemic stroke. Goal(s): We aimed to develop and validate a multiomics model integrating radiomics features from multiparametric MRI with clinical data to improve HT prediction. Approach: The multiomics model was developed and validated by combining radiomics features from diffusion and perfusion MRI with clinical variables derived from established clinical scoring systems. Results: The radiomics model performed comparably to clinical models. The multiomics model demonstrated improved HT prediction compared to individual radiomics or clinical models in both internal and external validation sets. Impact: The multiomics model, integrating clinical and radiomics features, improves the prediction of HT. These models could assist clinicians in making more informed decisions for managing patients with acute ischemic stroke.