The ASCO 2017, the annual Meeting of the American Society of Clinical Oncology (ASCO), new data were presented regarding the experimentation of Watson for Oncology, the version of IBM’s cognitive computing system trained to recognize the symptoms of various forms of cancer and suggest a therapy. Developed within the Watson Health business unit, it showed a level of agreement with human doctors that reached 96% in cases of lung cancer.
IBM started developing a version of its Watson cognitive computing system to be used in health-related projects a few years ago. The project was developed to the point that in April 2015 IBM announced the opening of a business unit called Watson Health to better manage resources and collaborations with the sector’s companies, clinics, hospitals and other entities. The Watson Health Cloud Platform, later called the Watson Platform for Health, was launched to securely collect and manage necessary health information on users.
IBM Watson’s applications in the oncology field begun before the opening of IBM Watson Health and their development continued with the creation of Watson for Oncology, a version of the IBM system specializing in this field, but not only. IBM also developed Watson for Clinical Trial Matching (CTM), a version specialized in helping physicians to evaluate candidates for clinical trials. In recent days, the results of various ASCO’s studies related to these IBM Watson Health solutions were showed, which gave very positive results.
Various studies were conducted in recent years and Watson for Oncology showed a level of agreement with human doctors that went from 73% for colon cancer in high risk cases to 83% in a study that included various forms of cancer up to 96% in a lung cancer study.
It would be interesting to understand the reasons for disagreements between Watson for Oncology and human doctors in the other cases. A while ago, there was a case where a therapy decided by human doctors for a case of leukemia gave no improvements and Watson diagnosed a different form of leukemia leading to a more appropriate therapy. In essence, there were already cases where Watson turned out to be better than human doctors.
IBM also announced a collaboration with the non-profit organization Hackensack Meridian Health to combine Watson for Oncology with the Cota platform, used to create medical databases to help all the parties involved in patient care. The aim is to create a synergy between the two platforms to further improve patient care.
The Watson for Oncology system is continuously improving through training with the acquisition of new data on real cases and collaborations that allow it to work with human doctors like at Jupiter Medical Center in Florida. In cases where deciding quickly the best therapy can make the difference between life and death, a system that can assess the data at speeds impossible for humans will become more and more important to help.