Using multimodal datasets of imaging and cancer-specific data, we are developing radiomic biomarkers to predict clinical and survivorship outcomes in cancer. In one study, we developed a deep learning biomarker to predict response to immune checkpoint inhibitor therapy; this was presented as a Clinical Science Symposium at the ASCO Annual Meeting. In another study, we validate Biomechanical Computed Tomography as an independent predictor of fracture among men with advanced prostate cancer. BCT-assessed bone density was highly concordant with DXA measurements. Given the low use of guideline-directed DXA screening in the real world, BCT of routinely collected CT scans may offer an accurate and convenient means to screen for fracture risk among patients with mHSPC.
