Multi-cancer Brain Metastasis Risk Score Development and Validation using 220,246 Whole Transcriptomes and Machine Learning


Jim Abraham, Carey Anders, Adam Brufsky, Michael Glantz, Priscilla Brastianos, Luke Pike, Amy Heimberger, George Sledge, Matthew Oberley, David Spetzler


Application of AI to large molecular data sets covering transcribed genes is in its infancy. To date, there have not been large enough data sets to take advantage of the power of AI technology to predict disease progression for patients with cancer. Here we show that application of ML/AI methodologies to a large collection of molecular and clinical data, generates insight into disease progression. While meaningful predictions can be made on cohorts of 100,000 patients, it is clear that more data will enable even more accurate predictions. Generation of WTS data on larger patient cohorts is essential to maximize precision medicine and advance the science and medicine of cancer care. These results can have a direct impact on current patients as well as provide insight into future drug develop efforts.


Whole transcriptome data can be leveraged by a machine learning platform that employs dimensionality reduction techniques along with transfer learning to predict the risk of brain metastasis. This tool can be used to augment the clinical picture of cancer patients an unmet clinical opportunity to inform prognosis, monitoring, and therapeutic selection.

Download Publication