Predictive Analytics for Detection of Metastatic Cancer
We used data-analytics approaches to develop, calibrate, and validate predictive models, to help urologists in a large statewide collaborative make prostate cancer staging decisions on the basis of individual patient risk factors. These models were used to design guidelines that optimally weigh the benefits and harms of radiological imaging for detection of metastatic prostate cancer. The Michigan Urological Surgery Improvement Collaborative (MUSIC), a statewide medical collaborative comprising more than 90% of urologists in the state of Michigan, implemented these guidelines, which were predicted to reduce unnecessary imaging by more than 40% and limit the percentage of patients with missed metastatic disease to be less than 1%. A subsequent follow-up study a year after implementation confirmed the substantial reduction in imaging without any statistically significant change in prostate cancer detection. The major reduction in imaging was estimated to save more than $260, 000 per year. More importantly, the reduced imaging is associated with reduced anxiety and harm to patients resulting from false positive outcomes that can lead to other more invasive procedures, and the increased access to imaging for patients with other diseases and conditions warranting imaging.
Team:
Selin Merdan, University of Michigan
Christine Barnett, University of Michigan
David Miller, University of Michigan
James Montie, University of Michigan
Brian Denton, University of Michigan