BigMedilytics gets silver medal at the BMMR2 Challenge
IBM Research team from Haifa came second-place winner (AUC 0.8380) with marginal difference from Penn Medicine (AUC 0.8397) in the international challenge, Breast Multiparametric MRI for prediction of NAC Response (BMMR2), organized by the Breast Imaging Research Program of UCSF. The competition was aiming at discovery of multimodal prognosis biomarkers based on retrospective analysis of a multicenter clinical trial of cancer patients who completed neoadjuvant chemotherapy prior to surgery.
The team used advanced AI techniques ranging from classical computer vision to advanced machine learning, to identify multimodal prognostic biomarkers derived from diffusion-weighted magnetic resonance imaging in combination with other medical images and clinical data. The challenge utilized prediction of pathological complete response following neoadjuvant chemotherapy for invasive breast cancer as a performance measure. The open-source technology that the team shared half a year back called FuseMedML played a significant role in the team’s ability to quickly experiment with multiple different models and variations and select the best performing one.