IBM’s artificial intelligence system known as Watson (the one that beat Ken Jennings on Jeopardy) is now being used to help treat cancer. This is nothing short of revolutionary in evidence-based cancer treatment.
The goal is to improve the quality and efficiency of patient care.
For a year, they have been teaching Watson oncology. The system has been fed 2 million pages of text from 42 medical journals and clinical trials relevant to oncology.
In addition, Watson has consumed 18,000 historical cases including medical records, patient outcomes, and physicians’ notes. Using natural language processing, Watson is capable of reviewing 1.5 million patient records in seconds. It doesn’t stop there. Watson continues to learn as it is fed new information. With every case and every new study that Watson learns, it modifies its recommendations.
As I mentioned, Watson does give a confidence score for each treatment. I would be willing to bet that Watson does a good job in analyzing the quality of the data. I wouldn’t go so far as to bet my career, though. Keeping that in mind, it’s important to use Watson as a supplement to our knowledge and training as opposed to a replacement. This means we must diligently analyze the evidence that Watson uses to back its recommendations before using them.
In the end, this is an exciting opportunity in oncology. Never before has a physician been able to use a supercomputer to help analyze the evidence and treat their patient.
A longer, more detailed demo is also available.