Top Line: The treatment of prostate cancer is all about risk stratification.
The Study: Today we bring you a very cool concept and an applaudable study. Baseline clinical data, digitized histopathology slides of pretreatment prostate biopsies, and longitudinal outcomes from five landmark NRG/RTOG trials (9202, 9408, 9413, 9910, and 0126) with over 10 years median follow-up were used to train and validate models predicting long-term clinically relevant outcomes based on pre-treatment digitized prostate biopsies. The models predicted all endpoints better than NCCN guidelines, with roughly 9-15% relative improvements across the board. What’s more, the model continually learns from ongoing input, which should result in not only ever-improving accuracy but also superb generalization across practice settings. This all practically means the global scalability of access to this state-of-the-art predictive technology is limited only by availability of digital scanners.
TBL: “This artificial intelligence-based tool improves prognostication over standard tools and allows oncologists to computationally predict the likeliest outcomes of specific patients to determine optimal treatment.”
- Esteva, NPJ Digit Med 2022