5 questions clinicians should ask themselves when using AI in healthcare (JAMIA.org)

I like that smart colleagues are starting to write about automation bias, interruptions, skill decline. This academic paper poses 5 questions we should all be asking ourselves. So begins our hard work to welcome a new entity into the exam room, with careful forethought.

https://academic.oup.com/jamia/advance-article/doi/10.1093/jamia/ocaf123/8287602?login=true

From the discussion:

Effective AI integration requires human-centered and adaptive design. Five central research questions address: (1) what type and format of information AI should provide; (2) when information should be presented; (3) how explainable AI affects diagnostic decisions; (4) how AI influences automation bias and complacency; and (5) the risks of skill decay due to reliance on AI.

Read the article. Love the thoughtfulness and humanity.

iPhone notes app is the purest reflection of our humanity (Wired.com) and a medical informatics observation

What’s on your notes app in your phone? WIRED argues that this simple, unfiltered blank page is the easiest place for us to store our unfiltered thoughts. How true. For me: fragments of blog post ideas, books I hear about, movies to watch, hilarious quote from family members, messy to-do lists. Hotel room numbers. Parking garage locations. Who knows? What’s on yours?

https://www.wired.com/story/iphone-notes-app-purest-reflection-of-our-humanity/

Sometimes the simplest note-taking apps are the most profound.

As medical records technologists going back to the 1800’s discovered, if we over-engineer our tools, doctors and nurses will break the bounds of what is allowable documentation to let the story come out.

From Annals of Internal Medicine (requires login) a brilliant history of medicine article by Eleanor Siegel

https://www.acpjournals.org/doi/10.7326/0003-4819-153-10-201011160-00012

The image:

What is fascinating is that: in the 1800’s, hospitals began keeping paper medical records, one book for each HOSPITAL WARD of about a dozen patients. No patient-specific medical records. If you wanted to look back, you would find the ward book for the year, find the day the patient was in the hospital, then look for the patient’s name.

Each patient would be an entry on the page for the day. There was only room for ‘intervention’ and ‘outcome’. No place to write thoughts, observations, theories, learnings.

So, doctors would at times turn the page over and use the blank back of the paper to write (in this case):

This patient came in with what appeared to be an apoplectic stroke. He was interesting in that he had a dextrocardia. He later developed a clinical picture which we could not explain.
Diagnosis: Hemorrhage into cerebrum
Complication: ?Syphilis

UCHealth Parkview reduces sepsis deaths (Beckers)

The story continues. Our EHR, partnered with Epic predictive AI model among other predictive tools have reduced sepsis mortality by 1000 fewer deaths per year compared to our baseline, as we find and treat sepsis earlier with reconfigured teamwork in addition to improved detection tools. Another tale of the Psycho-80: 80% of a project’s success is about the psycho-socio-political skills of the people and 20% of the success is due to technology. Grateful for smart colleagues and partners. (image, our fearless informatics leaders, analysts and trainers having a well-deserved meal after another EHR implementation day)

https://www.beckershospitalreview.com/healthcare-information-technology/ehrs/how-uchealth-uses-its-ehr-to-reduce-sepsis-deaths/

Grok-board: the 2025 update (instantly empathize and understand the EHR)

As tech accelerates, I wonder what has AI done for me lately? Sure it writes my notes, but does it really help me “grok” this patient? Here is my updated Grok-board design take for EHR’s.

I am worried. Just because our whiz-bang technologies can do lots of stuff, I ask: should it? I previously wrote about my Grok-board idea in 2024. In one short year, we have seen ambient notes, chart summarizers and agents take off.

Here, I still aspire to use an EHR that helps me “grok” the patient. Grok comes originally from Robert Heinlein’s “Stranger in a Strange Land” sci-fi novel from 1961, wow more than 60 years ago. He coined the term “Grok” meaning to “instantly empathize and understand.”

In my 2025 version, the left column of the EHR remains intact, with links to many useful parts of an existing EHR patient chart.

What does it mean?

  • Blue Column 1: The Patient story board. Links to patient details: demographics, allergies, primary care doc, care gaps, selected chart items.
  • Green Column 2: This Human: the patient telling us what is important to them, what are their joys and pressures.
  • Orange Column 3: Homunculus visual problem list and an AI summary of the current status of diseases with any available metrics.
  • Pink Column 4: Insights. Active and suggested AI agents, artificial intelligence entities that can obtain data, and then act on predetermined criteria to achieve a general goal, assess the patient’s top risks and suggested next steps.
  • Blue Column 5: Today: pre-visit questionnaires, solicited patient questions, the last progress note you wrote, and an AI Greek Chorus of somewhat adversarial advisors with suggestions for me.

An ideal Grok-board

should be humane and emphasize the patient’s identity and goals, so a physician can connect human-to-human and communicate more effectively. It should also increase the signal-to-noise of the information presented. It should make it quick to grok and then act. It should prioritize the most important next actions, then make doing the right thing easy

Thanks to Gregory Makoul for his groundbreaking work on connecting patient values to physician thinking. Let’s continue to debate the psychology of which displays gives the quickest, most useful view of the patient. Let’s arm-wrestle, not over sorting of the problem list, but over which display leads to better care of the patient with less cognitive burden and more joy for physicians, the clinical team, and the patient.

Taking it further

Let’s design a video gamer’s chair that helps you use all your senses. What if your left forearm vibrates to warn you of a medication allergy when you’re writing a prescription? What if your right calf feels warm if there is kidney impairment or left calf feels warm if there is liver impairment affecting your next decision. What if your lower back vibrates if you’re about to close a chart with unaddressed care gaps. What if we had smell-o-vision to detect the fruity breath of a patient in keto-acidosis?

CMIO’s take?  We are underusing our senses and our pattern-matching skills. Let’s build a way for our physicians and teams to grok the patient and then make it easy to do the right thing. We must intentionally build our humanity into this future.

Epic UGM 2025 FOMO generator #16. ‘COSMOS AI does not speak English, it speaks Events’ –Karen Wong MD

This is the quote of the meeting. From a chance meeting with Dr. Wong.

COSMOS AI preprint is available. My mind is blown and I want to know what this means. And I think I won’t know for awhile until the tech bubbles up to something I can get my head around.

https://www.alphaxiv.org/overview/2508.12104v1

I was chatting with Karen Wong Epic physician in the physician lounge in Voyager. I was bemoaning my trying to understand what COSMOS AI could do. Then she let this epigram loose:

COSMOS AI doesn’t speak English. It speaks Events.

This is perhaps the quote of the conference. Thank you Karen. And now, off to read the ARXIV article and pretend to understand it.

Epic UGM 2025 FOMO generator. #14. Adopting Gen AI at scale. Outpatient and inpatient insights and draft hospital course.

UGM229 Scaling Generative AI Chart Summary Features. Mayo Clinic

Getting from ‘blob’ summary to problem based insights. It got much better: ‘Revolutionary’. Now at 290 users. Surveying for usefulness.

Way to determine the value of this summary? Looking at usage. Else what is the point?

Outcomes

Powerful outcomes. Concise, burden reducing. Accurate. CT observation today: Ok we need this.

Draft hospital course. Compared to human written prior to go live. Now about 500 users. LLM did better than human in every category.

So helpful to see Mayo’s analysis plan: value, maturity, scalability. token limit problem early on. Now able to see many more notes to summarize. Including notes prior to cosignature. Important else missing data. How does this overlap with inpatient insights.

Success!

About 500 pilot users. initially providers but now all clinical staff. Nutrition, PT, RN. Could take 5-10 minutes to generate a summary in morning if many requested at once.

Downward trend of use? Unknown why.

Impressive work.

Epic UGM 2025 FOMO generator. UCHealth neurology and neurosurgery. Expecting the unexpected.

UGM177 Expect the Unexpected: Streamlining Documentation in Neurosurgery and Neurology. What does this mean?!

Kleptomania!

Vizient background

Epic automation? Rules?

Super easy. If malnutrition criteria met: statement put into note. If not, blank.

Thrombocytopenia example.

Our collection of rules.

Wow. All the CDI queries big drop off because the notes include the right documentation automatically. this is neurosurgery.

Automated smartlink for quality items allows coders to send queries for other high importance topics.

Looking at stroke patients

By adding smart links for encephalopathy and conditions changed the expected data. See the O/E improved over years in neurosurgery.

So physician builder stole ideas from other orgs like UW Madison. (disappearing help text)

Disappearing help text built this way, stolen from UCSD’s talk from 2019.

Also using Diagnosis Aware Notes.

Getting inpatient teams to use DAN improves charting of diagnosis.

AgileMD pathway with ability to teach residents follow pathways and insert text into the note.

This is the collection of tools to capture quality metrics and influence provider behavior and documentation.

Epic UGM 2025 FOMO generator #13. Ambient at scale. AdventHealth

UGM151 Ambient Voice Technology: 1,000 Down, Only 19,000 Users To Go. Advent tell us the road map

Early training model during the pilot. Support with love chat group on teams. For 14 days. Immediate support and success stories. Transitioned to live chat group with all clinical informatics support teams.

Support web page!

But no on-demand training. and a time zone problem for large organization. Now using QR 20 minute video.

Gives the option AFTER the video to sign up for license.

Tracked signal data. Criteria of DAX used in 25% of notes.

Create a training program in Signal. That way Epic can track time zero. Patrick Sweeney from Epic helped set this up. Survey from Microsoft.

23% drop in time in notes.

Typing drop by 20%. Drop copy/paste by 50%. Documentation length goes up.

Drop in documentation time but growth of time in chart review. Docs are taking that time and it goes back to patient care.

Epic UGM 2025 FOMO generator #12. The dinner spaceship cruise.

At the end of the day. Brain is full. Epic big tent does fun. Spaceship cruise. Join us!

Amanda Brill, Laura Macke, Rachel Rodriguez, Adam Ward, Frances D’Arcy, this blogger.

The author and Kelley Aurand, CMIO extraordinaire at Legacy Health who taught me about ambient.
Is this what AI really thinks I look like?
AI printers.

End of day introvert face. Thanks to Esther Park Christina Jung and David Bar-Shain for my PACmentor lapel badge for 2025-26. Go PACmentors!

Thanks Epic colleagues for another amazing day of learning.

Another day tomorrow. Anyone coming to my book club talk at 230pm? UGM255. Book Club for piglets. Mystery in the Midwest. Ukulele song. Fair warning.

Epic UGM 2025 FOMO generator #11. Mark Mabus ‘You had me at login’

UGM066 What’s New: A Tailored Upgrade Training Experience. Sure sure lots of great ideas but: a sound-track-backed anthem?!

Hear more about Epic’s new system training tool. Go see the slides and talk recording later. Now, enjoy 30 seconds of Dr. Mabus belting out this anthem.