Watson Goes to NYGC for On-the-Job Training in Oncology

March 19, 2014 | Today, the New York Genome Center announced a collaboration with IBM and a large collection of New York health centers to apply Watson's machine learning capabilities to cancer therapy. IBM's Watson supercomputer, which has been given test runs at Memorial Sloan Kettering and the MD Anderson Cancer Center, will now have a chance to design treatment regimens for real patients with glioblastoma. These patients will have their whole tumor genomes sequenced at the NYGC, and that sequencing data will be fed into Watson, which will use the mutations found in the tumor as a starting point for a drug search. Scanning the scientific literature, public drug databases, and clinical trial registries for compounds that hit the appropriate molecular targets, Watson will then suggest a drug cocktail targeted to each patient's individual cancer. After vetting by pharmacologists, physicians, and the patients themselves, these drugs will then be used as a novel personalized therapy.

The most significant new feature of the program is an intelligent feedback loop. After treatment, physicians will record each patient's response, learning how well each drug cocktail treated each tumor. That data can then be given back to Watson, which will gradually learn which of its therapies are most successful, and build on that knowledge for its future recommendations. Xconomy