If you are interested in AI’s potential in healthcare, by this time you’ve definitely read about how the emerging technology so far has made a bit of a faceplant when it comes to clinical decision support and telling doctors what to do. IBM Watson should ring a bell.
But if AI (and deep learning and machine learning) aficionados and true believers set their sites a bit lower, then there is still a lot to be gained in improving healthcare. At least that’s what a couple of Silicon Valley VCs — Bob Kocher and Bryan Roberts, both partners at the Palo Alto office of Venrock — appeared to advocate at the StartUp Health Festival on Monday in San Francisco.
When Investment Rhymes with Canada
Canada has a proud history of achievement in the areas of science and technology, and the field of biomanufacturing and life sciences is no exception.
Roberts spoke about it at some length.
I think you will see it dramatically on the administrative side of healthcare,” declared Roberts. “Which is less sexy to talk about than outperforming radiologists and stuff like that. But one could imagine a complete restructuring of the claims process and the admin process, the pre-op process, and risk adjustment by using deep learning and machine learning. That’s the easiest place to see it.
One could imagine applying it to the whole realm of oncology treatment, ranging from molecular diagnostics to treatments and treatment patterns….
The drug discovery process takes so darn long – that has to catch up [in order] to give you enough choice that you can’t do it on your fingers and toes. That’s going to take a little bit longer in my mind.
Roberts noted that to him AI becomes interesting when it can be broadly applied across healthcare instead of in just one area. He pointed to AI’s use in image recognition as an example of what he isn’t interested in. That’s because you have to use loads and loads of the data about the same thing to make the algorithms smart enough to detect patterns.
“It’s that set of applications that are useful horizontally across a lot of healthcare that has gotten our attention,” he explained. “I think the application that we’ve spent a bunch of time on lately sits at the intersection of machine learning, AI, and voice. It’s different, it’s passive — it’s not asking docs to do something else, and improving both the efficiency and quality of data capture for docs could be awesome.”
Kocher also indicated that it’s best for AI to stay away from human-body related applications.
“In biology, we don’t understand as well as we think, so I think AI may be more applicable in less clinical and more everything else in healthcare….That way we don’t have to worry about the FDA, reimbursement, hurting somebody. You can work on all other back office stuff..,” Kocher said.
In other words, there’s plenty of inefficiencies in non-clinical aspects of healthcare where AI can be applied successfully and be valuable, all without having to answer worrisome questions of is AI going to replace doctors and could it hurt a patient.
Photo: chombosan, Getty Images
