Machine learning analysis using 77,044 genomic and transcriptomic profiles to accurately predict tumor type

Authors:

Jim Abraham, Amy B. Heimberger, John Marshall, Elisabeth Heath, Joseph Drabick, Anthony Helmstetter, Joanne Xiu, Daniel Magee, Phillip Stafford, Chadi Nabhan, Sourabh Antani, Curtis Johnston, Matthew Oberley, Wolfgang Michael Korn, David Spetzler

Highlights

  • CUP occurs in as many as 3–5% of patients when standard diagnostic tests are not able to determine the origin of cancer.
  • MI GPSai (Genomic Prevalence Score) is an AI that uses genomic and transcriptomic data to elucidate tumor origin.
  • The algorithm was trained on molecular data from 57,489 cases and validated on 19,555 cases.
  • MI GPSai predicted the tumor type out of 21 options in the labeled data set with an accuracy of over 94% on 93% of cases.
  • When also considering the second highest prediction, the accuracy increases to 97%.

External Link