The field of cancer research is moving towards the development of multimodal machine learning frameworks that integrate various data sources, including clinical biomarkers, MRI features, and genomic data, to improve the prediction of treatment outcomes. This approach has shown promising results in predicting early recurrence of brain tumors and estimating individualized time-to-biochemical recurrence in prostate and bladder cancer. The use of deep learning and radiomics has also been explored, with hybrid frameworks demonstrating robust performance in classifying tumor response to therapy. Notable papers in this area include:

• A study that proposed a multi-modal machine learning framework for predicting early recurrence of brain tumors, demonstrating promising performance and offering a potential tool for risk stratification and personalized follow-up planning.
• A paper that presented a multimodal deep survival framework for prostate and bladder cancer, achieving a concordance index of 0.843 on 5-folds cross-validation and 0.818 on the development set, highlighting its adaptability and potential for clinical translation.
• Research that implemented an automated pipeline to curate a large longitudinal dataset of stereotactic radiosurgery treatment data, resulting in a cohort of 896 brain metastases in 177 patients, and predicting 12 months lesion-level response using classical and graph machine learning, reaching up to 0.90 AUC.
• A study that presented a hybrid deep learning and radiomics approach for predicting brain tumor response to therapy, achieving a mean ROC AUC of 0.81 and a Macro F1 score of 0.50 in the 4-class response prediction task.