I recently read a very interesting and personally relevant article in IEEE Spectrum Magazine. It was about how IBM and Memorial Sloan-Kettering Cancer Center are working together to teach IBM’s Watson to read patient medical records and search medical literature for suggestions on treatment. While the concept of knowledge/ decisioning systems is not new, one of the major hurdles is the heuristics are only as good as the information and weightings specified. The concept with Watson is that the system learns from the information and the weighting changes or is tuned over time and experiences. This sounds quite promising, as I expect there is lots of data/ information that is both new and often obscure that most doctors may not even be privy to. Thusly the Watson system can act as a system that responds to doctors to consider X, Y or Z treatments explaining why they were chosen (i.e. patient attributes/ conditions supplied) and how they compare to each other.
When I worked for a start-up called WhisperWire, we created a heuristics based Sales Effectiveness System that would guide telecom sellers on what products (or combination of products) should be suggested for customers who were interested in a variety of different needs (i.e. data, voice, web…). The engine heuristics was further tuned by the purchasing companies based on their system engineers inputs. This Watson initiative sounds similar, but for brain cancer and it has the possibility of self tuning.
I do hope to see this technological innovation expand quickly so neuro-oncology practitioners across the globe can benefit and brain cancer treatments are further optimized/ improved that result in better outcomes. For more in-depth reading on this subject, IBM Research has a great posting, IBM Takes on Brain Cancer, on the collaboration.