IBM recently announced a partnership with upstart genetics company Pathway Genomics to create a Watson-powered app to provide personalized health information based on a patient’s genetic profile, data collected from personal health trackers, medical literature, and whatever other information it can get its hands on.
As described in a recent Popular Science article, it would go something like this. A diabetic patient would open the Panorama app and ask it a question. Using its natural language processing capabilities to process the question, Watson would then come up with an answer based on information it retrieves from published medical literature & guidelines as well as patient specific genomic data, health history, and even personal health data like recent activity information or blood glucose measurements. Pathway’s CEO Michael Nova describes it like this,
Called Pathway Panorama, the smartphone app will make it possible for individuals to ask questions in everyday language and get answers in less than three seconds that take into consideration their personal health, diet and fitness scenarios combined with more general information. The result is recommendations that fit each of us like a surfer’s wet suit…[Say] you’re diabetic and you’re in a city you have never visited before. You had a pastry for breakfast and you want to know when you should take your next shot of insulin. In an emergency, you’ll be able to find specialized healthcare providers near where you are who can take care of you.
That’s a pretty impressive claim, highlighting both the remarkable potential and apparent risks with an app like Pathway. Nova suggests that the app will be available some time in the middle of 2015. While it’s possible that there is extensive testing being undertaken now to ensure the safety of an app that could provide such specific and potentially high risk recommendations, it’s more likely that this app will be crafted to avoid FDA oversight.
And while there are implementations of Watson in supporting care decisions elsewhere in medicine like oncology, there’s one big difference here. Those applications give recommendations to a “learned intermediary” like a physician or nurse – someone who takes this advice in conjunction with clinical specifics and their own knowledge to make the actual decision. Here, the recommendation would be going directly to the patient and is slated to be highly personalized, unlike say a simple Google search.
Regardless of what the actual scope and capabilities of the released app end up being, the integration of the veritable mascot of “big data” with the smartphone is something of which healthcare professionals should take note.