I just returned home from attending Epic’s XGM 2024. While there, I had a discussion with an Epic physician colleague while we were standing in one of the auditoriums watching the end of a presentation. I commented that it seemed that the volume and speed of development had accelerated. She affirmed that. What people may not have known is that their coding conversion to a web-based platform (Hyperdrive) had sucked up enormous resources over the past decade. That conversion is now done, and the full weight of their R & D is plowing forward, and it shows. That firehose of development was opened up in Verona, Wisconsin last week and here are some of my random thoughts:
The big documentation news that got the most applause on the first day is that Diagnosis Aware Notes is finally here for widespread implementation. I think it will finally convince a critical mass of our colleagues to use problem-oriented charting. People have always found value in problem-oriented charting, but the process was clunky and painful. Now, it will be easier and more intuitive. I recall bugging Janet Campbell (Epic’s VP, R & D Relations) in 2019 and posting this diagram on Epic's UserWeb (Ideas - A new take on problem oriented charting (epic.com). I have no idea how much of my ideas were incorporated into the design, but I am grateful that they finally have it designed and are available for widespread use.
There was, of course, tons of AI. Automation of In Basket messages has consumed most of the early external attention, but there are quite a few other things that I think will make huge impacts:
Ambient documentation continued its forward progress. There was a serious emphasis on getting these tools to nurses.
There are 70+ AI projects in active development. A few have already made it to production. Here are some that stuck out: AI coding query suggestions; chart abstractions and form prepopulation; lots of versions of automated summarizations; AI generated SQL queries.
The AI team will increasingly open-source their tools.
As more and more healthcare systems develop their own data science skillsets, Nebula will allow them to localize and customize models based on local data.
The AI Trust & Assurance Suite will facilitate tracking complex models without requiring organizations to create their own monitoring systems.
Analytics (Epic's Cogito suite)
Epic has continued to accelerate Slicer Dicer's development. There are now 145 data models and the list is increasing. I remember when there was just one. Epic has continued to increase this self-service tool's capabilities, allowing end users and nonanalyst informaticists to get sophisticated answers to questions. It is Epic's ad hoc, self-service, no-code analytic solution. Every Epic organization should increasingly use it as an analytic first pass. The feature that I personally am most excited about is the ability to create custom filters (aka columns or features) without coding. It is an analytics informaticist's dream.
In addition to Cosmos querying data in the larger Epic community, a "Research Data Enclave” will help researchers create data sets from local data.
Epic is partnering with Microsoft and leveraging MS's Azure Fabric to seamlessly combine native Epic data with external data like ERP, traffic, weather and supply chain.
Population Health:
All sorts of tools to facilitate risk adjustment, from relatively simple descriptive analytics to advanced AI-driven solutions and prepopulation.
Suspected population health metrics will automatically teed up to be added to the problem list.
Pharmacy nonadherence dashboards will be available.
There were increasing tools to address and record care gaps and engage patients.
Of course, I think most of you will know that I couldn’t resist engaging in as many Cosmos sessions as I could:
I attended a Cosmos-specific forum, where we heard how organizations are already starting to use Cosmos to publish studies. We got tastes of R and Python coding. We were also introduced to the new Cosmos GitLab, which will increasingly be a key tool in helping the Cosmos community come up with new insights.
Best Care Choices and Community Antibiograms cull data outside of an organization’s local Epic system to help guide treatment orders of magnitude more nimbly than waiting for similar information in traditional analog information delivery workflows.
Condition-specific (e.g., sickle cell) growth charts will allow pediatricians to track growth more appropriately and meaningfully.
(Rare) "Patients like mine" is real and live. Epic presented a real anecdote of a patient with unusual GI symptoms refractory to "usual" treatments. The physician used "patients like mine" to connect with another physician who had used the unique treatment of gastric Botox. The patient received the treatment and dramatically improved. Examples like this reinforce to me that information is the key to solving our healthcare problems.
Study sponsors can ping local Epic organizations that they have patients eligible for studies by querying logical parameters and pushing the list to the local organization.
Ultimately, Epic is well on its way to creating the learning healthcare system that the IOM conceptualized almost 20 years ago.
This is by no means an exhaustive summary. It is just the items that I could remember and decipher my notes. Epic's spring conference used to take a very distant 2nd fiddle to its bigger sibling UGM. I think this is no longer the case. When Epic created XGM out of what was then widely considered less prestigious Spring Forum content a few years ago, I recall thinking and hearing from others, "What? Where are we doing to see and talk about the technical, cool s&*t?" Well, after a bit of a speedbump from the pandemic, that has been answered. IMO, XGM is a full peer to UGM. Different, but still just as important.
It wasn't all nerdy seriousness. We continued our tradition of a night at the Ivory Room piano bar. For all the firehose-drinking hardcore IT work, the most satisfying part of the week was reconnecting in person with colleagues and friends.
Very insightful summary. It was so awesome to connect with you once again in person, John. Let's continue to rock/roll or shake/bake!