Can Smart Computers Help Us With Patient Follow-Up?

\r\nHe has published in several medical journals, including recent research on use disparities in telehealth visits & effective community outreach interventions to encourage COVID vaccination. His weekly MedPage Today column, \u201cBuilding the Patient-Centered Medical Home,\u201d focuses on improving patient care.<\/p>“,”affiliation”:””,”credential”:”MD”,”url_identifier”:”fp4223″,”avatar_url”:”https:\/\/assets.medpagetoday.net\/media\/images\/author\/Pelzman_330px.png”,”avatar_alt_text”:”Fred Pelzman”,”twitter”:””,”links”:null,”has_author_page”:1,”byline”:”Contributing Writer, MedPage Today”,”full_name”:”Fred Pelzman”,”title”:”Contributing Writer, MedPage Today, “,”url”:”https:\/\/www.medpagetoday.com\/people\/fp4223\/fred-pelzman”,”bluesky”:””}]”/>

Throughout our days, through the course of our interactions with patients and the follow-up period in between visits, we are constantly faced with data that we need to address: findings on history, physical exam, blood test results, even information that’s hidden in old medical records that come to us by fax or by mail.

And it’s up to us to decide how this helps or hurts our patients — how it guides our therapies or changes our management decisions, how this data can be useful to help engage our patients and help them step onto a healthier pathway, and even how it ultimately might prove to be the key to something dramatic that makes a huge difference in their health in the short term or the long term.

Some of this is easy. If a patient comes in and says they are smoking, not exercising, and feel that they’re overweight and want to lose a few pounds, this conglomeration of data leads us to a fairly easy decision, a set of recommendations we can make and tools we can give them to help address these problems, and perhaps stop them in their tracks. Not that the doing of this is easy, but knowing just what to do isn’t that hard.

On the other hand, when a patient gets a CT scan or blood test results or undergoes a stress test, the results may present a logical next step, which is easy, or open up a whole new can of worms, which can be hard. Uncertainty seeds confusion and fear, and incidental findings may lead to increased patients’ anxiety, more testing, and even unforeseen complications down the road.

Figuring Out Next Steps

Engaging our patients at each of these branch points through shared medical decision-making, and ensuring that they understand what all of this means and what all of their options are, can help them make the right choices moving forward. But sometimes that’s hard to do. We might see a subtle deflection in the trend of someone’s laboratory results, a complaint that doesn’t quite fit into the bread-and-butter way we think about a certain presentation, one that might even give us the missing clue to what is really going on, or even a really scary result that portends something awful lying hidden just up ahead.

These are the ranges of cognitive tasks we face every day, this intellectual activity that combines our knowledge and experience, and helps us fulfill our mandate to take care of our patients. But how often can we completely and safely fulfill all of these demands?

Sure, when someone’s blood pressure has been persistently elevated, and we’ve gone through nonpharmacological interventions, and the high BP persists on home and in-office monitoring, then it’s fairly easy to come up with a game plan to manage their hypertension. And when a lab result, imaging finding, or results of a biopsy point to an obvious next step, as devastating as this sometimes can be, the path forward from there is usually fairly straightforward.

How we address it with the patient, how we gently tell them what’s going on and guide them through the rocky roads ahead, makes all the difference in the world. I think we’d all love some more help with all of this stuff, all along the spectrum from the easy to the incredibly difficult.

Inventing a Smarter System

I’ve heard that people are working on systems out there to help present patients with detailed information about their lab test results through the patient portal, and something smart enough to answer their questions and fit this into the clinical context of their own health history, something which ultimately may be bordering on “practicing medicine.”

Right now, when a patient sends us a portal message, our electronic medical record generates an artificial intelligence-based first pass response that we can use or edit. When a patient portal message says, “Dr. Pelzman, why are my basophils so low, and why is my CO₂ one point below the upper limit of normal? Is there anything I can do about it? Do I need to take medications? Is this something serious? Is this something that could harm me?,” then maybe the system can start to do that second-level depth of thinking, providing better guidance. “No, this is not clinically meaningful; no harm awaits. Keep doing what you are doing. These ranges are just population-based averages.”

This smarter thinking could include data from prior lab results in our system and the clinical context they presented with on the day when the labs were done, and even draw on a line of text buried in decades-old records that were faxed over from their long-ago PCP, using pattern recognition and predictive modeling on all of this to help decide whether this is a benign finding or something more insidious.

When a radiology report contains in the impression the suggestion for follow-up repeat imaging in 3 to 6 months, this message to the patient could explain the thought process behind this recommendation, why it is best not to proceed with a biopsy or surgery or chemotherapy at the present time, that these recommendations are not made lightly, and that if they’re concerned we can certainly reconsider and discuss further options.

Working Alongside Us, Not Instead of Us

Ultimately, I see a smart system like this running along beside us, listening in on our conversations with patients, listening in as we call out the results of the physical exam, understanding as we speak our plan and instructions for the patient, and then synthesizing these results and helping present the patient with the information they need to make the best health decisions for themselves, always under our guidance.

I don’t think any of us want to completely give up the reins of dealing with all of this stuff to a computer, no matter how smart and foolproof anyone tells us it is. But if it’s there with us through the day, getting input through some sort of device like an ambient listening system, smart glasses, or a body camera, maybe it can augment our care, enhance our decision-making, and improve patient’s feelings that they’ve had everything listened to and explained to them. Maybe its presence can reassure the patient that all the loops have been closed, all the i’s dotted, and all the t’s are crossed, and that they’re less alone than they probably feel right now in this overwhelmed, burned-out, and frustrating healthcare system.

Sounds easy enough.