Thursday, December 5 | 8:20 a.m.-8:30 a.m. | R1-SSR04-4 | Room S401
A computer-vision-based AI model that also incorporates assessment of clinical variables can predict treatment outcomes in prostate cancer patients, according to this presentation.
Researchers from the U.S. National Institutes of Health (NIH) led by presenter Benjamin Simon have developed a deep-learning pipeline that can predict the probability of biochemical recurrence (BCR) for patients with prostate cancer undergoing definitive radiotherapy with or without androgen deprivation therapy.
After training and testing three different types of algorithms, they found that the best results were achieved with a multimodal model. In testing, it yielded 86% sensitivity, 90% specificity, 90% accuracy, and an area under the curve of 0.92.
“Leveraging AI to predict [biochemical recurrence] from MRI has the potential to improve [prostate cancer] risk stratification over clinical variables alone,” the authors wrote. “After further external validation, future models based on this pilot work may inform treatment selection.”
Learn all about their model by attending this Trainee Research Prize-winning talk on Thursday morning.