Opinion | Small Wins

  • author['full_name']

    Fred Pelzman is an associate professor of medicine at Weill Cornell, and has been a practicing internist for nearly 30 years. He is medical director of Weill Cornell Internal Medicine Associates.

Last week, we had a guest speaker at our faculty noon conference, Ashley Beecy, MD, the medical director of artificial intelligence operations for our hospital, who came to give us an overview of what’s happening around artificial intelligence (AI) in healthcare across our enterprise.

It was a terrific in-depth talk, showing us a lot of what is going on behind the scenes to develop new AI applications to help improve clinical care across the spectrum of everything we do.

There has been a lot of talk about artificial intelligence, on a lot of new areas of implementation, places where they’re starting to make inroads in using this new technology to improve the lives of patients and the healthcare providers who work with them. We all hold out hope that this is going to go well, that we aren’t going to look back years from now and say we wish we hadn’t turned this stuff on, that it’s not going to get out of hand, run away from us, or further entrench imbalances and inequities in healthcare access and utilization.

How AI Might Help

A recent article in JAMA Internal Medicine talked about the ways that artificial intelligence might help improve the lives of those in primary care. The four main areas the authors felt the field should focus on were inbox management, clinician documentation, between-visit panel management, and individualized decision support. While it’s clear that they are going to be able to develop some fancy tools to make this stuff functional, efficient, and effective for us and for our patients, it feels like there certainly are places where we can get some bang for the buck before these new widgets are released into the wild.

Right now, our electronic health record (EHR) system possesses an enormous amount of functionality that we’re not taking full advantage of, that I think could lead to amazing improvement in the efficiency of the care we provide and improved satisfaction for patients and caregivers, and could allow healthcare providers to spend more time taking care of patients and less time being data entry clerks in the EHR.

The system has best practice alerts and reminders and pop-ups that can be used to help ensure that patients get everything they need, right when they need it. Wouldn’t it be great if there was a system that churned through every patient’s chart and helped fill in every healthcare gap that existed?

I love it when my patients use the portal, send me a message saying that they got a reminder that they’re overdue for a shingles vaccine, or Hepatitis C screening, or their mammogram, and can this be set up? Wouldn’t it be great if we could automate this, if this could happen without a lot of effort from the provider?

Assistance With Medical Decision-Making

If the system says they are due for a colonoscopy, and the patient says they want one, can’t this all just happen automatically? A conversation between the patient and some sort of electronic assistant could go back and forth through some shared medical decision-making, perhaps some educational materials, and once a decision is made to go ahead with the colonoscopy, a referral can be placed automatically, using the correct ICD-10 code and sending this off to the endoscopist’s team to help the patient schedule the procedure, coupled with information from their office about the prep and timing of the procedure and all of the details they need to know to get it done.

If the system says a patient is due for a flu shot, and they want to come in and get one, can’t that happen without a lot of back-and-forth messaging to the doctor?

Right now, patients are able to call up and self-schedule their own mammograms at our institution, and at which point we get a request from Radiology to “put in an order” for this procedure. The order is exactly the same each time, the specific breast cancer screening mammography order that is in place in our EHR — and there are a few acceptable codes for routine breast cancer screening that get the procedure covered by insurance. Why should we be involved in this at all?

True, it only takes a few seconds. I get the message in my inbox, I open it up, I click to the order section, I start typing mammogram bilateral until the right order pops up, hit enter, then toggle down to where it says the reason for the test, type in breast cancer screening or click the button if it’s already there, then sign the order, and off it goes to Radiology. Not much clinical medical decision-making going on here.

For the most part I’m not having a discussion with the patient about the risks, benefits, and alternatives of mammography, and we’re not doing much shared medical decision-making. If the patient is able to schedule this on their own, and my job is really just putting in a referral, can’t we just take me out of this process altogether?

Using AI to Improve Disease Management

Think about how we could do this for chronic disease management as well. Let’s say I want to improve my patient’s control of diabetes, hypertension, heart failure, high cholesterol, or COPD. Perhaps we could design a set of interactions — a package of office visits and other touches; a way to escalate doses of medications, endless patient education, telehealth and other video visits, and remote patient monitoring; and the support of patient navigators, community resources, pharmacists, and nurses — and then set this in motion and let it run on its own.

Or what about when a patient calls for a refill of a medicine but they haven’t been seen in well over a year, and they are overdue for necessary monitoring and laboratory testing? Suppose that the system could prompt them to schedule an appointment for blood tests and an in-person physical exam before further refills are given?

I’m hopeful that someday we’ll have powerful computer systems that will scan through all of a patient’s EHR and plunge into the depths of the medical literature to find clues as to what might be going on with the patient, to help guide us to a better differential diagnosis, to suggest next steps, or to spot diseases before they manifest. A “Dr. AI,” if you will. But for now, I think we’d all really appreciate some small wins, some ways to make things easier, to smooth out the rough edges, to ease the burden of all that everybody is doing to try to get through these days.

As the singer Dave Edmunds once said, in a song written by Bruce Springsteen, “From small things, Mama, big things one day come.”

Please enable JavaScript to view the

comments powered by Disqus.