What happens when CT Lin allows ChatGPT to help write EHR parody lyrics about ChatGPT? Does the Universe invert itself?
Recorded at the Vanderbilt Annual Otolaryngology Seminar at Vail, Colorado. An invited talk by CT Lin on AI in Healthcare, but wait, does he end with a NEW song?
Yes. Here it is: “ChatGPT” sung to the melody of “Sweet Caroline.”
THANK YOU to Dr. Eben Rosenthal at Vanderbilt University, Sarah Rosenthal at UCSF, and the entire VUMC Otolaryngology Group.
Here we go: Ben Zhao, a professor at the University of Chicago, led a team to create NightShade, a tool that lets artists alter pixels in their images so that if/when ingested to an AI training set, will cause the resulting model to “break.”
From MIT Tech Review
https://www.technologyreview.com/2023/10/23/1082189/data-poisoning-artists-fight-generative-ai
This is very much like the antibiotic / antibiotic resistance battle that humans have with bacteria.
I will be very interested to see where this journey leads.
I celebrate you, my colleagues in the arena. For all of our failures, our setbacks, our licked wounds, we are making a difference.
Bringing my uke to another informatics conference.
A note of appreciation to my colleagues doing practical informatics work on the frontlines of healthcare
It may seem that everyone else out there has it easy: after all, JAMIA, JMIR and ACI informatics journals are filled with amazing colleagues who write about their amazing research, clinical innovations, successful project and implementations. Be careful not to let the FOMO get you down, as it does to me on occasion.
It is so easy to judge oneself: There must be something wrong with ME, why can’t I get anything done? Why can’t I be successful, why is it so hard for me?
Watching the finale of Ted Lasso recently, I was, as they say in French, ‘triste’. An emotion of sadness, tinged with joy. (I think that is right). On the one hand that the series is over, and on the other to recognize that there is such great art, such great writing in the world, about being a good person.
Ted talked about Theodore Roosevelt’s ‘the man in the arena.’ Or that the person doing the work gets dirty, suffers the slings and arrows, gets criticized. On the other hand, that same person is also truly living life, experiencing joy, making a difference. This contrasts with the person in the relative safety of the sidelines— the critic who points out flaws, denigrates.
CT deploys a pre-populated start date for schedule II controlled substance prescriptions to simplify prescriber workflow. Great, except terrible. The pre-populated dates cause unexpected havoc. CT removes the tool after a week and apologizes to 4000 physicians and APPs.
CT deployed radiology “indications of use” as checkboxes to improve ordering provider workflow. Great, except, terrible. Pages of scrolling checkboxes are unusable and are removed after 2 days. CT apologizes.
CT advocates for and then deploys messaging between patients and their clinics and providers to increase patient connection and engagement. Great, except terrible. CT apologizes for ruining healthcare.
These projects had a difficult birth, and yet, with persistence and great teams and supportive informatics colleagues, achieved success.
CMIO’s take? Keep in mind the long game. Let not the inevitable sarcastic shouting of the critic silence our efforts.
I celebrate you, my colleagues in the arena. For all of our failures, our setbacks, our licked wounds, we are making a difference.
How can a plastic water bottle contain so much wisdom, you ask?
Image: This AI-generated image is Not Quite What I Wanted.
The images in this post were generated by Bing Chat Dall-E image generator to different versions of the prompt below:
Draw an impressionistic picture of a doctor and a patient; we can see the doctor’s face clearly. The upset patient is in the act of throwing with the right arm; a plastic water bottle has just struck the chest of the doctor. The doctor is surprised, the exam room is out of focus. The doctor now has to figure out how to repair this conversation.
You be judge of how
a) I am not yet a great prompt engineer
b) Bing does not yet understand staging and body postures and
c) generally, WTF?
Two decades ago, after just a few years spent seeing patients in a busy internal medicine practice in Sacramento, I had burned out. Sure, I was great at all the internal medicine stuff: diabetes, blood pressure, heart, lung, stomach problems and more. But, I was tired of the endless parade of vague symptoms and self-destructive behaviors (lack of exercise, alcohol- and cigarette-consumption) for which my skills were ineffective, and I often felt powerless.
In the late 1990’s, all the junior faculty at the University of Colorado were required to attend a four-hour communication workshop held by the Bayer Institute (now, the Institute for Healthcare Communication).
At this workshop, I found communication tools and an entirely new way of thinking about patient care, one that emphasized demonstrable caring, listening, and relationship-building. And I came away with not just theory, but practical words I could use myself.
I was excited enough that I spoke to my workshop instructor, Dr. Fred Platt after the course and asked if I could learn to teach the material as a way of understanding it better myself. To my surprise, he took me under his wing.
I traveled to West Haven, Connecticut to learn to be a facilitator of the physician-patient communications workshop. I was proud to receive this honor. I met about 30 other physicians sent by their respective organizations to learn how to teach the communications workshop. We were the chosen! We were the elite!
The next day at the training center, my puffy big head and I were not at all prepared for the intensity of the interview with my first “standardized patient” (an actor mimicking a patient scenario in a standardized way).
I was told “You are a surgeon going to see your first patient of the clinic morning.” I walked in, sat down, and said: “Good morning, Ms. Smith. How can I help you today?”
The patient screwed up her face and … and threw her plastic water bottle at my chest. She yelled: “You have f***ing ruined my life! And you DON’T EVEN CARE!”
My Best AI-generated Image: This is as close to my intent as I got (patient throwing water bottle at surprised doctor)
Eyes wide, I froze. After a few seconds of terrified silence, the coach sitting at my elbow signaled “time-out.”
“Maybe you’d like to pause here? What is going through your mind right now?”
What was going through my mind was: “OMG OMG OMG. I have to get out of here.”
Less correct image: jet-powered water bottle? No wonder the doc is surprised.
Eventually I calmed down, we debriefed the situation, and I decided I could try again.
It took several repeated attempts, and constant at-my-elbow coaching, to even start a conversation with her. Wow. Turns out, there is a LOT to be humble about, in our line of work!
In hindsight, it was a perfect example of “limbic hijacking.” The limbic system in the brain regulates emotion. Limbic hijacking describes a moment when emotions run so high that the heart-rate spikes and the reptilian “fight or flight” hindbrain takes over. In the moment she threw the water bottle at me, my poor pre-frontal cortex (PFC), where problem-solving and effective communication skills are stored, got shoved aside by my protective reptile brain, and went completely silent.
Me: ‘Hello? Pre-frontal cortex? I’m stuck! I need a rescue! What constructive words can I say here? Help!’
PFC: =petrified silence=
Hindbrain: ‘I know! Get up and run away!’
Me: ‘Great. A lot of help you both are.’
I was hijacked.
Less correct image: OK, what? That fist is also not quite right. I’d look away, disturbed, too.
Over the course of a week, we practiced many communication strategies, all the ways to build a connection between doctor and patient, until I could survive that hostile interaction, and even bring back tools for teaching others.
is at the root of so many great things. One of the patient-interview frameworks we learned was E4: Engage, Empathize (of which, one crucial skill is reflective listening), Educate and Enlist. The E4 framework is a dramatic contrast to the way many physicians interview their patients, which could be called F2, or “Find It: Fix It.” In other words, focus on data. As a consequence, a physician following “F2” ends up ignoring ideas, feelings and values, which are crucial in establishing an effective therapeutic alliance between doctor and patient.
In the years since I took the course in West Haven, this material has become available in an online course through the Institute for Healthcare Communication and you can sign up for it here.
The E4 framework could be summed up in a quote from a physician colleague who told me:
“The more I listen, the more my patients learn.”
This beautiful, seemingly contradictory statement, has a lot to teach.
Less correct image: OK, this is not even close.
“Hey doc, those pills you prescribed didn’t work. I’m still gaining weight and my blood pressure is high. Maybe you should give me something that works for a meat-and-potatoes guy like me. I like my prime rib. And I hate when everyone lectures me about exercise. I just need to be able to see my daughter graduate.”
From the E4 framework, the second skill: empathize, includes reflective listening, which is active listening, non-verbal eye-contact and head-nodding, and saying back not only medical data, but also ideas, feelings and values, so that the patient can feel that their concerns are truly heard.
Less correct image: OK, what is even happening? Offering water to a combustible patient? Good idea on those gloves.
Potentially useful reflective statements:
“It sounds like your blood pressure is still high.” (Data)
“It sounds like you are a meat-and-potatoes guy.” (An idea)
“It sounds like you HATE getting the exercise lecture.” (A feeling)
“It sounds like you really want to be able to see your daughter’s graduation.” (A value)
In a medical interview, many things are happening; a patient is telling a story, a doctor is listening for medical clues. And now, the doctor also has to pay attention to these ideas and build a relationship. It seems like too much. However, with practice, it becomes easier.
Once you say these statements out loud, they become part of the conversation. The patient feels heard, and there is less of an argument over numbers and a resistance to behavior change. Instead, the two of you are having a conversation about the patient’s motivations, struggles and goals.
Before, I thought these statements were just “fluff” and not part of being a medical doctor. I ignored them and focused on symptoms and data like the blood pressure and weight.
Finally, I realized that patient care wasn’t just about book learning and medical knowledge, but also about caring for the human being across from me. Of course, I already knew that. But did I really?
After learning 10,000 new medical terms and studying dozens of textbooks and meeting thousands of patients over 11 years of training, this new framework was a revelation. The exam room interview was so much more complex and interesting than I had previously understood. It was not only a logical search for symptoms and signs of disease, but the more complicated task of building a human connection.
Huh!
With this revelation, I felt recharged! I could enjoy patient care again.
Image: I did not mean that the doctor would CATCH the water bottle…
Reflective listening works well with patients. Communications training makes an enormous difference in reducing physician/APP burnout AND ALSO increases patient satisfaction and adherence to therapy. What’s not to like?
Try this yourself: With patients, see if you can restrict yourself to speaking less than 50% of the time. Focus on reflective listening and asking to hear more. Then summarize the medical data, as well as the ideas, feelings, and values. This is difficult at first, but it is also increasingly rewarding for both patient and doctor or provider. Done well, it does not take more time.
Less correct image: Lego version of my story
In the years since West Haven, I have found many other places to apply these communications skills. In fact, we move so fast in our worlds that in meetings, you regularly see people talking past each other, trying to get their own points across, and not listening.
Have you heard THESE phrases in meetings?
Every one of these sentences illustrates a mindset of trying to convince others and NOT a mindset of listening or seeking first to understand.
There are terrific books on this subject:
Here’s an idea:
In your next conversation or meeting, actively suppress your urge to jump in and make your own point. Instead, listen to truly understand, and then summarize before contributing to the conversation.
“Seek first to understand, then to be understood.” — Stephen Covey
listening to all the voices, encouraging the less vocal participants to speak up, and hearing as many viewpoints as possible. I listen for participants talking past each other, repeating themselves, and not listening to others’ positions.
Importantly, participants rarely summarize the positions or statements of others, especially those with whom they disagree. Consequently, no one feels heard. As a result, people start repeating themselves.
Usually the meeting is about half over when I speak up and make a brief summary. I call this my “buckets” summary. I say something like: “I have heard three different threads in this discussion so far.”
Then I summarize: “Bucket 1 is …. Bucket 2 is … Bucket 3 is …”
If I can, I also try to give the discussion a new direction. “Can we agree that for Bucket 1, we don’t have the right participants, and also that this is less crucial, so we can defer? Then, for Bucket 2, we can grant that participant A has a useful idea that seems straightforward. I did not hear any disagreement. The big discussion, then, is Bucket 3, where … Do we agree on this as our priority? If so, may I propose we try X? What do we think of this?”
I find that after a statement like this, if all the parties feel like they were heard, then I did a good enough job listening and summarizing. Then, we can use the remaining minutes to focus on actual problem solving.
“If you want to go fast, go alone. If you want to go far, go together.” —African Proverb
Image: I don’t understand what is happening here, but it is so entertaining. The more you look, the less sense it makes. Is she cold? Why is that one finger on the bottle so long? Is that a bed or a desk? What happened to the bottom of the glass door? Who threw the bottle? What is he pointing at? Is it a magic trick?
Try this yourself: In meetings, see if you can listen hard enough to serve as a summarizer, an organizer and a clarifier. Or, listen for someone who does this well, and practice listening and summarizing like they do.
To borrow my colleague’s quote:
“The more I listen, the more effective my meetings are.”
The psychedelic scientist who sends brains back to childhood.
https://www.wired.com/story/the-psychedelic-scientist-who-sends-brains-back-to-childhood/
This is a long read. Totally worth it. I have been reading about psychedelics in the past few years and the brain science that is catching up to what LSD and psylocibin users have been saying since the 1960’s.
Michael Polan’s work “How to Change Your Mind” also speaks to this rich history. Here, the article describes how a ‘critical period’ of learning in the youthful brain in mice, which closes and makes the adult mouse untrainable, can be ‘re-opened’ with psychedelics.
Wat.
I found this a very hopeful avenue of research for our mental health future.
Printing a network of neurons: super cool and a harbinger of things to come …
https://www.wired.com/story/a-lab-just-3d-printed-a-neural-network-of-living-brain-cells/
In high school I wrote an article (in the 1970’s!) for homework on computers and how one day amazingly complex, hi-powered supercomputers would be THE SIZE OF BASKETBALLS and NOT THE SIZE OF ENTIRE 2-STORY HOMES. Wow.
Now, we walk around with supercomputers in our pockets. There is more computing power on an iPhone than there was for all of Cape Canaveral at the time of the Apollo moon landings, with all the 3-body orbital mechanics calculations.
And now, we have neural-networks in software mimicking how reasoning brains work. Our next step is creating the physical layout of how brains are structured, which will give us another layer of insight into brain function, growth and remodeling.
Right now, the lab is printing 8 mm x 8 mm x 0.4 mm structures (about 1/3 of an inch square, and maybe a sheet of paper in thickness) with 100,000 cells (wow). Just wait ’till we scale this up.
A character, in a situation, facing choices. That’s what it’s all about folks.
Beyond giving gentle feedback, this Story Skills workshop via Story Republic taught me to see story structure separately and simultaneously to story content.
A story, at its root, is a Character, in a Situation, facing Choices. Broken down further, it comprises the 5 C’s: Context. Catalyst. Complication. Change. Consequence.
The brilliance of this workshop is the amazing community it builds, and that it delicately distributes feedback by encouraging folks to comment on participant stories that have fewer comments, and connects people from around the world. For example, my accountability group included 2 folks from Australia, 2 from England, a guy from Chicago and me. Furthermore, I found a ‘storytelling in a healthcare group’ with a couple of pharmacists, a physical therapist, a chiropractor, and an ICU nurse. It is amazing how strongly you can connect online in a short time. Many of us plan to stay in touch.
I thought I was a pretty good storyteller before the workshop. I regularly got nice comments about my funny stories and people would laugh and smile as I told them. As I started writing these down and using the 5 C framework, I began to realize that I was telling Anecdotes (a funny or interesting moment) and not necessarily a Story (where a character faces a choice and is thus changed by it). As Bernadette taught us: a good story entertains, a great story moves us. I wanted to tell Great Stories.
Every part of the 5 C’s were hard for me. It was easy to put my best stories into this framework, but with the feedback from my co-participants, I saw that:
CMIO’s take? Thank you Seth and Bernadette for a wonderful thing you’ve built. Heck with Reading, wRiting and ‘Rithmetic. We all need this course.
Tell stories. P-values alone don’t change minds. Learn gentle feedback. Feedback alone doesn’t prompt improvement.
In 2020, I took a Story Skills Workshop (by Seth Godin and Bernadette Jiwa). I have to say that I learned quite a lot, and not what I was expecting to learn. This was the first session in what is to be a workshop offered repeatedly. I highly, highly recommend it. Seth and Bernadette offer a series of online lessons, released over time. There are about 6 expert coaches, and the instruction is to sign up for an interest group or ‘accountability group’. You’re given a story structure (the 5 C’s: Context, Catalyst, Complication, Consequence) and then specific lessons to write and polish specific elements of your own story in this framework. The cool part is the instruction to ‘first write your own story, and then go comment on at least 5 others.’
CMIO’s take? Story telling: cool. Gentle, effective feedback: cooler. Don’t we all need to get better at this?
Dr. Casey Greene of the Center for Health Artificial Intelligence at CU describes the business of Serendipity
I very much enjoyed the presentation by Dr. Greene recently at the University of Colorado’s AB Nexus event on AI, Advanced Computing, Quantum & Human Health.
His take on AI: it is a way to engineer serendipity. Using pharmacogenomics (drug-gene interactions) as an inspiration, he described how information can be introduced into care pathways to engineer serendipitous moments. We can now isolate DNA from a patient’s blood, saliva, or other samples and detect genomic variants. Many of these variants affect an individual patient’s response to certain medications.
The trouble is that while many research biobanks can know a lot about a patient, getting this in front of patients and their physicians, APP’s and other clinicians at just the right moment, is the moment of engineered serendipity that makes a difference in improving care.
A common strategy in pharmacogenomics is testing by free-standing genomic companies that will run similar genomic analyses, and then generate a PDF or printout or even a wallet card and say “Be sure to show this to your doctor the next time you receive a prescription. Who knows, it may affect which drug you should take!” This is often impractical. Can you image a patient bringing her/his wallet card or PDF to every future doctor visit (urgent care, ER, clinic, hospital, virtual visit) on the off chance that their genomic variant and the associated tens to hundreds of affected drugs, is one that she or he will be prescribed in the future?
In our UCHealth system, we have incorporated key genomic variants, generated for greater than 30,000 patients in our biobank project, into our Epic Electronic Health Record. In total, this includes more than 200,000 actionable genetic data points. Now, Just-In-Time, if a prescriber plans to prescribe Plavix (for example), a common anti-platelet medication, to someone with a variant for which it is contraindicated, the EHR will put up an alert: “CAUTION: Pharmacogenomic alert. This patient’s genomic variant means that this medication may not be effective. Consider Ticagrelor instead. Click here to switch.”
Another example of serendipity is our ability to detect a confluence of a hospital patient’s symptoms to spot sepsis. This blood infection is subtle and deadly if not treated quickly enough. We use AI to spot patterns and notify the right person at the right time. As a result, we have been able to save 800 more lives from sepsis and related deteriorations in-hospital.
CMIO’ take? I like the idea that AI, or pharmacogenomics, is not just a tool, it is a way to engineer serendipity.
A rock, a hard place and lateral thinking. Clear some space on your desk and in your head and come along!
Generative AI response to
“a physician informaticist caught between a rock and a hard place”
YOU are a physician informaticist. You are board certified in informatics, you are an effective physician builder, and enjoy an excellent relationship with your department chair and clinical colleagues. You have protected time to this work. You have excellent IT analyst colleagues who will collaborate and help get your projects over the finish line so that they’ll be used. You have built scoring tools, you have built smartforms to better capture clinical findings, you have built analytics reports that clearly demonstrate that you have improved the care of the patients in your clinics.
AND YET. Your organization is part of a large health system with byzantine socio-political dynamics. You have multiple bosses, and each boss has very different ideas on what you should and should not work on, and they don’t like each other. Projects that you initiate and introduce to your leaders disappear from view, only to resurface later as some one else’s idea.
A project that you were not initially involved with is now YOUR problem. Some project team built a Best Practice Alert that interrupts all the doctors, and no one is happy: it doesn’t have the right information to help with decision-making, the doctors are bypassing it, the intervention is not working, but HEY. YOU’RE the expert on workflow, so can you help us? Tonight? We need to fix this tonight.
Why don’t they keep you involved on projects you initiated? Why don’t they involve you earlier on projects where you have expertise? Despite your best efforts to describe your dilemma, it seems your bosses don’t really get you and aren’t listening.
And, you have made accommodations for quite a while. Maybe it’s okay, maybe it will be better next time. Maybe you’ll finally get be publicly recognized. Oh, not this year? Okay, maybe next year.
What if you heard this from a colleague? What would you do?
Multiple choice:
Reader, I have used all of these responses in my long and confusing career. What a disappointment I have been to my mentees and colleagues.
This time, though, in a recent conversation, I went a different way. This is something I vaguely learned during a leadership exercise, I’m not sure to whom to attribute this.
Given the scenario above, I said instead: Here is a task for next time we chat.
Take a time machine 2 years into the future. You have made some choices, advanced your career, and now you can’t WAIT to talk about the amazing things you’re working on.
How did you get here?
Set aside some time. Get a bunch of blank pieces of paper, some color pens, sticky notes, scissors, glue, whatever you have to be creative. TURN OFF your laptop, phone, notifications. No interruptions. No electronics. No one around to judge you.
Your mindset: Blue sky thinking. Post-cards from your future. Turn off your editor brain that says “You can’t do that.” Everything is possible.
Task 1: Quantity not quality. Fill an entire page with ideas of what you might be doing in 2 years. If you fill it, keep going.
* One column of ideas are “Redesigning my current role”.
* One column is “Stay in my current field but make a big leap”.
* One column says “NO limits. BUT, I cannot stay in my current field”
Fill the page. Most of the ideas will be terrible. Keep going. Dedicate at least 30 minutes to this. Your brain is not permitted to edit or cross out.
Task 2: Highlight the ideas that appeal to you in all 3 columns. Pick a few from each column and write a few sentences filling out the idea, just enough to develop some feelings around it. Does it excite you? Surprise you? Disgust you? Bore you? Scribble some notes. Spend at least 30 minutes on this.
Task 3: Get up. Go outside for a walk. Think purposefully about NOTHING. You have just prompted your brain to jump out of its rut and be open to completely new directions. Don’t listen to music or an audio-book or podcast. Be silent with your thoughts. Let your mind wander. At least 30 minutes. Some would even argue that you should sleep on it and conduct task 4 tomorrow. This is to allow system 2, your subconscious and non-directed brain to explore.
Task 4: Were there more ideas to scribble down? Pick your favorite idea from columns 1, 2 and 3 and write a page about each one. Scenarios you see yourself in. Things you are working on in each parallel universe. Projects you’ve accomplished. The joys of that work. Draw a picture. Use scissors and glue to build an artifact. Envision yourself in that role. Feel it. Close your eyes and look around in that world. Spend at least 30 minutes.
As you conclude your time, look at your 3 papers. One of them speaks more to you than the others. In one of them your inner child rejoices. In one of them, your energetic younger self cannot wait to get going. What is it?
At the least, it can be fun to see: a) how badly your handwriting has degenerated as we handwrite less, and express more thoughts by typing or speech-to-text, b) what it feels like to be in kindergarten again, scribbling, coloring, drawing, c) see how your right-brain creativity can take some time to come out of hiding.
Who knows? There might be something really useful or interesting there. Is it time to act? If not right now, should we remind ourselves to do this exercise again in a year?
This is an exercise in lateral thinking. This is a way of jumping your brain out of the usual well-worn cobblestone paths with deeply-grooved ruts from horse and oxen-drawn carts rumbling along them for millenia. Sometimes going “off-road” can spark an insight.
CMIO’s take? What did you come up with? Are there fragments of these dreams that you can put into action now? Where did this journey take you? Take a picture, send it to me! Let me see!
