From the course: DJ Patil on Data Science: The Ask Me Anything Conversations

What is AI's place in healthcare?

From the course: DJ Patil on Data Science: The Ask Me Anything Conversations

Start my 1-month free trial

What is AI's place in healthcare?

“

- [Interviewer] Where do you see AI's place in healthcare and drug discovery? - Well, one of the most exciting things that is taking place is the application of these technologies like AI into healthcare. But the first part that I think we need to start with is where is AI going to have the biggest impact? Everyone's talking about AI replacing the doctor and here's the interesting thing is that the biggest applications oftentimes of AI are in the unsexy most boring problems. Where is the bulk of energy go right now in a hospital? It's in the back office and it's filing paperwork and it's doing this, it's faxing papers back and forth. What would it look like if a lot of these paper pushing activities were just automated by machine learning? You'd get incredible amount of efficiencies, but you'd also reduce the amount of errors. The other part that starts to happen in there is what about when the places where most people are harmed, which is in a hospital is you get the wrong type of drug, you get the wrong type of pharmaceutical. So medical errors. Where could you find medical errors? When you could you see signals that are bad? There's a case that was a drug called Vioxx, which is a painkiller, and if you look at the data, there's an incredible amount of people who had heart complications as a result. Why couldn't you see that signal? Because no one was looking. But with machine learning and AI, you're always looking through the data, you're crawling to the data constantly to find patterns that are anomalies. And so you start to see that. The other area that I think is going to be transformative is in this idea of population health. And this is the idea of how do you take the broad set of data around a population to try to understand connections and relationships to see what might be correlated, how diseases might interact. And so questions around autism, diabetes, what are the sources? Are there environmental relationships? Are there other things that we might not think of? It wasn't that long ago that people used to think ulcers were of one cause and then it was finally shown that actually you can treat a lot of ulcers through antibiotics. That was a super mind blowing result that got a team a noble prize. So what other relationships might we discover? We bring that data and we search through it to find new kind of relationships. That's where the biggest power is. And a bunch of that is going to also come with pharmaceutical drug discovery because there's treatments or certain drug types that might be slightly chemically similar, but we haven't thought about like, "Hey, that's actually similar." But a machine can process those just so effectively. And when that comes together in these broad platforms that are being built, I think we're going to move in a transformative way. And then it comes down to the question of how fast can we test these drugs? How can we actually put this into the broad clinical care so it benefits everyone versus a small set of the population that live next to just the top tier hospitals?

Contents