Ever wondered about CT’s biggest failures? At CHIME 2024 Fall Forum, be prepared to come and laugh at CT’s expense. And then learn how to write your own failure resumé, embracing your own vulnerability. Let’s build a fault-tolerant future together.
In my experience, the CHIME Fall Forum is a terrific place to rub shoulders with those doing practical, innovative work in information technology, in informatics and improving patient care with EHR-enabled workflows.
Come join the conversation! See you there.
And, if you’re coming, comment to this post, or reply to this message and help vote on which ukulele song CT plays:
An EHR parody of Blank Space by Taylor Swift or
An EHR parody of Sweet Caroline by Neil Diamond?
Or perhaps an older EHR parody from my youtube channel catalog:
Don’t you wish there were a step-by-step guide to fixing that darn Inbasket problem for your burned out EHR colleagues and for yourself? Surprise! Here you go. #AMAstepsforward
Updated for August of 2024, Drs. Jane Fogg, Christine Sinsky, Jill Jin and I collaborated on our respective organizations’ efforts at Inbox reduction and collated our methods into one best-practices guide.
I love the AMA approach with a step-by-step guide for organizations getting into this work. Committee structure, which leaders to involve, how to tackle each specific method. There are 8 easy steps here, take any one of them, to make a significant dent in the Inbox universe and reduce the burden of this important work.
A preview:
Develop an EHR Inbox Task Force
Measure Your Current State Using Audit Log Data
Adopt a Strategic Framework: Eliminate, Automate, Delegate, Collaborate
Begin With a “Great Purge”
Eliminate Low-Value and Preventable Messages
Automate Protocols and Pathways for Routine Tasks
Delegate Message Handling to Upskilled and Empowered Team Members
Collaborate to Fully Cover the Inbox During Physician Time Off
CMIO’s take: Are you doing this work? Are you doing something not described here? Have you recently had some success? This is too big and too important not to work together. Let me know!
We are all struggling with information transparency as we build and deploy AI models in healthcare. How did you build that? How can I trust it? Why should I trust you? Maybe the FDA nutrition label can teach us something.
The FDA nutrition label is a paragon of simplicity. Established in 1990, there have been 30 years of standardization and improvements and is now a mandatory part of food labeling.
Similarly, we are in the wild West days of artificial intelligence. With ChatGPT and similar generative AI tools, along with a rapidly growing cohort of predictive AI tools, there is growing public interest, and skepticism about the value of AI, particularly in healthcare.
We face questions from patients as well as healthcare providers:
Who built this?
Where was this model trained? On what population of patients?
How are you accounting for potential bias?
Who is running this program?
How are you analyzing performance and working to improve the tool?
These are difficult questions, and even more difficult to explain in an easily digestible way.
The FDA label is small, very concise, highly readable and information-dense.
Here is my attempt to recreate this label for AI projects.
I included Purpose, Used by, Used with, Developed by, Monitored by, Risks, Benefits, HIPAA status, Local Fine-tuning status, Bias review status.
I am quite certain this will not be enough, and is also not simple and clear enough. Won’t you help suggest improvements? This needs to be simpler, more readable, more comprehensive, and address both patient and healthcare provider concerns. Easy, right?
Here are more examples based on this model: our generative AI projects we’re working on with partners:
Our predictive AI models we have developed:
CMIO’s take? What is YOUR take? Use make great old ideas new again. Those who don’t know the past are doomed to repeat it. On the other hand, I’m doing it on purpose.
Ever wanted to deconstruct what CT is thinking? Here’s your chance. CT’s anxieties and lessons about predictive AI and generative AI, the last slide of one of his current talks, up to date as of 5/23/2024. Out of date tomorrow. And some books he read to Steal Like an Artist.
April 19, 2024: In this episode, Dr. CT Lin, CMIO of UC Health, engages in a thought-provoking discussion about the integration of generative AI in healthcare workflows, sharing insights from his conversations with industry experts. He explores the nuances of clinicians’ perceptions of AI-generated notes, the potential impact on patient care, and the delicate balance between automation and human expertise. CT delves into the challenges of customization, the importance of teamwork in optimizing EHR usage, and strategies for mitigating clinician burnout. As the conversation unfolds, listeners are prompted to consider the evolving role of technology in healthcare and its implications for clinician well-being and patient outcomes. How can AI enhance efficiency without compromising quality? What strategies can healthcare organizations implement to foster effective collaboration and innovation? And how can clinicians navigate the complexities of automation while preserving their expertise and autonomy?
Key Points:
AI Voice
AI Acceptance
Patient Transparency
Clinician Burnout
Here’s the YouTube recording of the our Keynote interview:
Immediate release of test results, discussed with a radiologist audience: a pro and con debate. What could go wrong? And, of course, the ukulele.
Thanks to Dr. Jennifer Kemp who designed and invited me to present a panel called: “Information Blocking Pro and Con: A Debate.”
I am at RSNA today. My first. Did you know the Radiological Society of North America is the largest medical conference in the world? I did not know this until yesterday. 40,000 attendees, over 4000 speakers. That works out to about 80 speakers PER HOUR. Geez.
I was one.
Disappointingly, not all 40,000 attendees came to see our panel presentation. 🙁
Nevertheless, of the 80 attendees, we had an excellent discussion in the context of releasing complex radiology images to patients, including MRI CT PET etc. and the resulting problems if/when patients find out about cancer or other devastating result by viewing their results online.
There are exceptions to the federal rule:
If the patient prefers not to see the results
If releasing the results may result in Physical Harm to the patient or other person (note that anxiety or psychological harm does NOT qualify)
Systematic embargo or delay of release of result is forbidden based on this federal rule.
The proposed penalty for violating this rule is $1 million. However, we are aware of only about 400 registered complaints of information blocking, 3/4 of which are organization to organization blocking, and only about 100 of patients registering because of not receiving notes or results. And we are not aware of any successful complaints resulting in actual penalties, as yet.
Question from the UK
We had a question from a colleague from the United Kingdom, as they are just now about to formulate a similar law. ‘Would it be reasonable given US experience to establish a national standard for embargo: let’s say all providers uniformly delay a high risk result for 3 days?’
Our reply: probably not. Airline passengers are now aware of every moment of their luggage transport, and every moment of their arriving Uber driver. Why would radiology reports be any different? The consumerism movement is unidirectionally toward more transparency. Maybe 5 years ago, if the UK was considering a standard, that would have been standard of care in the US, but no longer.
indicate that among 8000 patients who had already received test results from a patient portal, 96% indicated they wished to continue to receive results immediately. Even among those receiving abnormal results, 95% still wished to receive results immediately.
It is also true, however, that 8% of patients receiving immediately released results did worry more. However, we believe this worry is based on getting “bad news” more than it is about getting “bad news immediately.” These are the patients we need to focus on, and more details we need to study.
Our suggested plan: that ordering physicians use anticipatory guidance: ordering physicians will eventually need to explain the result to the patient. Why not spend one more minute at the time of ordering to dramatically reduce the anxiety of the patient when they view the result later at home?
3 easy steps:
1. We are ordering a test. You may see the result before me. Best case, this is normal and I will contact you this way…
2. Worst case it could be … there is X% chance of this. If that is the case this is how I would reach you…
3. You have a choice: look immediately or wait to hear from me. What Q do you have?
In our experience this works very well and doesn’t take much time at all.
And for radiologists, publishing a contact number for patients to call if they have questions is very reassuring to patients and, guess what: they rarely use that number: in a busy multi-radiologist practice over the course of years reading hundreds of thousands of studies, their office has received 1-3 phone calls A YEAR from patients. And most of the time it is about factual errors in the report, and rarely is it to ask about the medical impact of the findings. It is quite minimal work.
CMIO’s take? The time for immediate release is here. There are great solutions to the anticipated problems. It also happens to be the law in the US.
If you’re still not convinced, or even if you are, here is a song for you, fresh from Chicago’s RSNA 2023:
For the 63rd episode of the CIO podcast hosted by Healthcare IT Today, we are joined by CT Lin, MD, CMIO at UCHealth-Colorado to talk about patient messaging. To kick off the episode we dive into his work in sharing patient results and the efforts to keep it private. Next, we talk about how the sudden increase in patient messaging has led to some practices charging for the service to get Lin’s experience and thoughts on the topic. Then we take a look into Lin’s past projects to get his insights on what project he felt was the most successful and what made it successful. Looking forward, we also discuss what projects Lin wants to work on but hasn’t had the time. We then talk about AI and where we think it’s heading. Finally, Lin shares the best career advice he’s been given and how playing the ukulele has impacted his career. Plus, he finishes off the episode with a health IT ukulele song.
Here’s a look at the questions and topics we discuss in this episode:
You were ahead of the curve with sharing results with patients. Where are you at today with Information Blocking and sharing data with patients?
Patient messages are overwhelming doctor’s Epic inboxes. Many are starting to charge for these messages. What’s been your experience with this and how is UCHealth approaching it?
What’s the project you’ve worked on that’s brought you the most personal satisfaction and feeling of success and what made it successful?
What’s a project you want to work on, but just haven’t had time to yet?
Where is all this AI headed?
What’s the best piece of career advice you’ve been given?
Where did you learn to play the Ukulele and how’s that impacted your professional career?
Come for the Informatics discussion, stay for the ukulele. Or don’t. Topics: Big Data / AI in practical use; Blowing up the Classroom for EHR Training; Inbasket Hyperobject: what is that, exactly? And can we deconstruct it?
And maybe, a ukulele song. Transcript available for those who aren’t patient enough to listen (but then, song lyrics and melodies don’t translate well with an auto-transcriber).
