Ezekiel Emanuel on Why Analytics and AI are "Substantially Overstated" in Medicine

The outspoken physician, healthcare administrator, and author does belive that machines will replace radiologists sooner rather than later, though.

Outspoken physician, healthcare administrator, and author Ezekiel Emanuel, MD, sat down with Healthcare Analytics News™ backstage at the Mayo Clinic Center for Innovation's Transform meeting in Rochester, Minnesota. Emanuel was founding chair of the Department of Bioethics at the National Institutes of Health and is currently vice provost for global initiatives at the University of Pennsylvania.

At the conference to participate in a televised debate on the state of American healthcare, Emanuel covered a host of topics in our conversation. Here, he gives his feelings on the role of advanced analytics and machine learning in medicine, which he finds to be exaggerated.

Transcript:

So how do analytics play into this?

It's both vital and, I think, substantially overstated. So what's vital?

Look, having data on which patients are sick, on where you're spending money on those patients, on which providers are not performing well, providing information back to providers about who their patients are that have gaps in care that they need to close…? Vitally important.

I think where the analytics are overstated is when people talk about how we're going to have minute-to-minute blood pressure or heart rate monitoring, and we're going to feed that back to patients, and that's going to change what we're going to alert patients when their weight goes up too much or their breathing isn't quite right those all the studies with those kind of analytics and minute-to-minute patient outcomes or patient medication adherence they seem to have failed and I don't think that's an avenue that we really should go down.

Are artificial intelligence and machine learning going to transform medicine? Well, maybe we could argue about what “transform medicine” means, but I'm actually skeptical that they're going to be the key element of transformation.

There are places in which artificial intelligence and machine learning are definitely going to make medicine very different. For example, radiologists and pathologists are definitely out of a job in the next decade because of machine learning, you've got all these digital images, whether X-rays or CT scans or pathology slides, and machines are just going to be better at reading them and detecting pathology than human beings. And machines don't get tired, as people often say, and they learn much better by repetition, so I think that's one area I think another area where sort of high-tech is going to be important.

Also in some elements of telemedicine, especially for rural areas where specialists are in scarce supply, like pediatric specialists or cardiac specialist or GI specialists, or psychiatrists or dermatologists. Being able to get a consultation where you're not going to have a lot of the access to these actors is going to be important.

Are machines going to interact with patients, give them diagnoses, and manage them? No that's not happening, and I think that's a total pipe dream that somehow machines are going to disintermediate healthcare providers from patients. I just don't see that in the future at all