If you're developing predictive models using machine learning, you know how important it is to continually test and update your model. Technology rarely stands still and improvements in models are nearly always warranted, as evidenced by a new report from researchers in the Molecular Imaging Branch in NCI’s Center for Cancer Research.

NCI’s Dr. Baris Turkbey and his colleagues made updates to their clinical diagnostic model to align with the latest diagnostic criteria (i.e., the Prostate Imaging-Reporting and Data System [PI-RADS], version 2.1). The researchers also integrated additional imaging data (i.e., multiparametric magnetic resonance imaging, or mpMRI data).

After testing with data from 1,319 patients, the researchers found the updated PI-RADS model with mpMRI data was a clear winner, outperforming their traditional clinical model and two others.

Dr. Turkbey noted, “By combining mpMRI with our clinical model, we were able to create an integrated approach that revolutionizes the way we interpret biopsies and manage prostate cancer. This type of model offers more precise identification and could reduce the need for additional and often unnecessary biopsies.”

He added, “Although we saw similar results across a highly diverse patient group, we’ll be conducting additional testing to assess our model’s generalizability in a community setting.”