HIMSS 2023 Impressions
Below are my takeaways from HIMSS 2023 that just wrapped up in Chicago.
More than 35,000 people attended the HIMSS conference in 2023. Source: HIMSS
HIMSS Conference Going Strong Post-COVID
The size of the crowd was noticeably and significantly bigger that 2022, by a lot. It felt like pre-COVID days with crowded hallways, information sessions and shuttle buses that were practically standing room only. Official attendance was 35,000 per HIMSS. The feedback from exhibitors was they met a fair number of customers, but most of these were pre-arranged appointments vs. random encounters in the exhibits. As I walked around, I saw more vendor badges than buyers, which makes me think HIMSS has become more of a vendor show than an end user conference that it once was.
HIMSS is not known to be the place to learn about innovative technology breakthroughs, but it is the place to learn about how healthcare delivery organizations (HDO) are executing programs to pilot and deploy them. The key information you can learn at HIMSS are:
How have HDOs approached these innovation programs structurally? In other words, who is the executive sponsor and how have they organized the team implementing the project?
What are their KPIs to measure success?
Which partners are they working with?
This year there were many solutions areas being discussed at HIMSS. There are two that stood out to me as key this year: Tools driven by large language model AI and Hospital at Home (H@H). Let’s start with H@H.
Hospital at Home
H@H strategy has been something I have been discussing for two years as a likely next move by health systems. Several forces combine to make this an imperative. First is the impact of COVID in 2020 when we saw massive adoption of telehealth along with the sudden drop in hospital-based procedures as many states shut down "elective care". The ability to push care outside the four walls of the hospital suddenly was not just a nice to have, or meet some kind of consumerism niche, it was actually strategic to continue the core mission of caring for the community. For those health systems with value based care contracts, virtual care allowed them to continue managing care for their patient panels.
The second force is the arrival of retail giants into health care (like Amazon and Walmart) and more generally the blurring of traditional industry lines in health delivery (like pharma companies building services marketed directly to consumers, and health insurers sponsoring their own telehealth services). It was logical that H@H should be a strategy for established health systems to beat these newcomers and sort of leap frog their ambitions.
The economics are compelling as H@H allows a health system to expand the number of patient days they can bill without adding physical beds. An interesting prediction was presented by Michael Capriotti from Virtua Health at his HIMSS session. He quoted Dr. Bruce Leff from Johns Hopkins who predicts in the future, hospitals will only offer "EDs, ORs, and ICUs. Everyone else will be treated at home".
An example of how one health system approached their hospital at home program presented at HIMSS 2023. (photo by the author)
Why are HDO's so interested in H@H right now? To unlock facility capacity. Just like electrical utilities who are happy to pay subsidies for homeowner solar electrical generation to avoid building new power plants, hospitals want to pursue H@H to avoid adding more inpatient bed towers. If they can add billable beds without actually building beds, that makes them money.
There was unfortunate mention of CMS' Omnibus Bill which temporarily guarantees parity for H@H stays to be paid at inpatient levels. I say unfortunate because what we want in the long term is the cost of providing care for equivalent hospital stays to go down. This is one way we are going to save healthcare from bankrupting the economy. However I admit in the short term it provides needed stimulus and incentive for hospitals to begin building their programs now with plenty of revenue support. Eventually these price parities will go away, and by then the more mature programs will have figured out the complex set of processes and support needed. If inpatient revenue will be cannibalized by H@H, there is a runway here to figure out how to deliver it profitably.
The final impression about H@H is that it is really hard to do. This is too much to cover here, but as an example the services these providers deliver include meals, durable medical equipment, continuous monitoring, medical transportation, in-home nursing and medications. It is an extension of a med/surg unit.
AI Tools for Healthcare
There was plenty of AI at HIMSS, and it was everywhere. It reminded me of RSNA of 6 years ago. Just like Cloud, we are seeing a lot of AI-washing. For example, one session I attended the presenters used AI in the title, but during the talk one realizes it was not about AI. What they really used were statistical modeling. When I asked details about bias and training data, they had no answer.
The biggest splash at HIMSS was the Epic and Microsoft joint announcement about bringing OpenAI's GPT-4 AI language model into health care for use in drafting message responses from health care workers to patients and for use in analyzing medical records while looking for trends. The Epic booth had an entire wall dedicated to future product ideas for how GPT-4 might be applied to health delivery. There was no promise as to when these features will be made available however.
There are more than enough blogs and news articles about AI in medicine, such that I don't feel compelled to write much more at this time. My conclusions from RSNA still stand today, so see that blog about the outlook of where we are going next. However, to restate one key takeaway from RSNA that I saw at HIMSS: the application of large language models to aid in reducing healthcare clinician toil, thus making their job easier, is the right product at the right time.
