10245
Blog
Josh Streit

Something to Break: 3 Things to Leave Behind to Embrace an AI-Empowered Health System

We really shouldn’t be surprised. But, collectively, we always are. Change comes slowly, creeping up like a shadow, and then all at once it’s everywhere. Molds, paradigms and businesses are all eventually shattered, broken to make way for the next great thing. 

Along the way the real innovators figure out how to break themselves as they grow their business in a kind of adaptive, malleable model which always improves, always moves forward and always has something new to offer. Those who do not get replaced with the kind of velocity that will make your head spin.

In this information age in which one can really learn anything on a whim, we really should be better at predicting major shifts. After all, the Sears to the Willis Tower transition came swiftly. Mobile computing became the default within the same decade of its creation. Are we really surprised that the internet has changed daily life for almost all? 

Of course not, and in nearly all cases, the invisible hand, a concept first described by Adam Smith in 1776 in “The Wealth of Nations,” drives this rapid pace of change constantly toward something better and often creatively destroying outdated systems. That’s the point. It’s invisible – at least at first. But after enough of its work has been made manifest, its fingerprints are everywhere.

The Change of Our Time: Clinical AI

That is what I was thinking about as I walked the showroom floor at HIMSS last year: Adam Smith, an invisible hand, and creative destruction. How could I not? All around me, literally in every booth, was AI, AI, AI. 

Apparently everyone has it now. If that is the case, this must be the most explosive technological revolution in the history of humanity. I mean, from one year to the next, every major and minor IT provider in the U.S. healthcare market has become an expert in AI driven solutions. Or….as Ray Dalio puts it so simply in “Principles,” is “this just another one of those?” 

Specifically, is AI another interoperability, enterprise imaging, deconstructed PACS, patient portal…(the list of past technological fads is exhausting, isn’t it)? Each were and are great ideas, but those which have also either fallen short of what was believed to be achievable or simply didn’t catch on in a real, system changing manner. It begs the question by contrast, is there something about today which is truly different? 

The last time an opportunity or shift of this apparent size occurred was 2009 with the implementation of HITECH in February of that year. The U.S. government created a source of funding to incentivize the adoption of electronic health records. That was different. 

Famously, Judy Fulkner had created Epic Systems in 1979 and electronic record keeping had been around since well before then, but in this case, things had truly shifted and there were many vendors in that space ready to make their run at the new paradigm of that decade. February of ‘09 was full of “shifts,” as one might recall, with the stock market bottoming less than a month following HITECH (not that the two are directly related), and the EHR and stock market have been off to the races ever since. 

Perhaps infamously, due to the outsized success – and its catalyst – relative to her competitors, Judy recently found herself the subject of a rather popular presentation on regulatory capture from Bill Gurley (a venture capitalist and central figure in Showtime’s “Super Pumped”). I suppose that story continues to unfold as the success Epic achieved within this technological era is inarguable. 

So if our time is truly different as it was in the era of “The Big Short,” then something else, an incumbent of some kind, has to step aside to make room. After all, budgets are only so big and each day presents us with the same 86,400 seconds. For every vendor to be able to do it and/or every health system to consume it, what was once a production process or product has to be broken, destroyed like paper charts used in the first decade of this millennium or all the other hopeful EHR vendors laid waste in the wake that is Epic Systems. So with respect to production, enterprise use of artificial intelligence in healthcare…what gets wasted? 

What We Have to Leave Behind

As vendors and technologists, we spend so much time presenting and demonstrating what has been made, I don’t think sufficient time is spent contemplating what gets broken in the process. 

From Aidoc’s perspective, we certainly have needed to pivot and break from our original conceptions of what AI in healthcare was going to be about – what was going to be its most important contributions, functions and lessons. Mostly, as you can see from the list below, what needed the most corrective adjustment was our own perspectives. Here are some of those aspects of ourselves and our customers we were forced to address if the potential of our field was to ever materialize: 

  1. The notion that current ways – those unassisted by AI workflows and assessments – of doing things are – because they have been accepted up to now – good enough. As the cliche goes, ‘because that’s how we’ve always done it’ is not a good defense for not exploring the means by which any physician, to highlight just one use case, of any level of competency can’t be augmented. For instance, in the field of medical imaging in which Aidoc has spent the last nine years doing this kind of work, our ‘second set of eyes’ – which never takes a day off or gets sick or carries healthcare benefit expenses – has been the engine behind +150 published papers, abstracts, posters and global speaking engagements. The feedback is positive, too. Physicians take extra time out of their day to let us know how our assessment and triage tools made a positive contribution to their patients. While great, it is just a start. We must all continue to drive to more and deeper exploration of the workflows which can be augmented in this manner if the wave of new patients flowing into our systems are to be accommodated. 
  1. The assumption that existing investments, relationships and infrastructure are adequate for the implementation of a technology the space, with few exceptions, has never seen. As Aidoc has done its part to try and pave a reasonable onramp for the AI field, we have found that narrow is that path of solutions. So narrow that we could not find a ready made solution adequate to meet the needs of this new frontier. We falsely assumed we could get by producing lots and lots of algorithms (which we’ve done) and they could ‘live’ in the existing healthcare space. Granted there have been enormous investments to date on complex infrastructure for data center, electronic records, revenue cycle, etc. These are all critical parts of today’s healthcare IT infrastructure. We have even had a handful of FDA approved algorithms since shortly after the time, pun intended, of the Great Financial Crisis. But none, literally not one, of the organizations providing today’s current, critical IT infrastructure is rooted in artificial intelligence automation. Like all the vendors on the showroom floor of your favorite trade show, they’re now trying to catch up and retrofit a new era of technology into their legacy systems, business practices and most notably and materially, marketing. 
  1. The belief that a health system can’t get started today. Aidoc has actually done most of the initial heavy lifting of what it requires to take one’s first steps on this journey. Granted, it took the creation of a new kind of clinical IT infrastructure we trademarked as the AI Operating System (aiOS™), but that work is in the past already. If one follows the imaging informatics space closely, you may recall that it was in the Fall of 2022 in which Aunt Minnie recognized this technological accomplishment as its choice for best new radiology software that year. 


In fact, here is a list of the kind of out of the box capabilities the aiOS™ accomplishes for an enterprise:

  • Robust suite of multimodal use cases
  • Largest suite of FDA cleared solutions
  • Enterprise integrations to all relevant systems
  • Integrated desktop, mobile and web portal environments
  • Subacute workflow for ongoing patient engagement
  • Proprietary, automatic, agentic workflow orchestration
  • Dockerized, partner ecosystem of solutions
  • Modular, enterprise visualization of all solutions
  • HIPPA, SOC 2/Type 2 compliance
  • +150 customer publications, making their results public
  • ROI model based on those publications
  • Literally days’ worth of customer testimonials 
  • Accompanying drift monitoring and mitigation practice for all solutions

We took the steps to create this kind of environment because it was necessary. There was no alternative system/software/platform which could accomplish the above. Had we found one, it would certainly have been faster/easier to leverage the creativity of others to our advantage. When what one needs isn’t available from others, there is only one other avenue to pursue – build it. 

Through this process of creation, we learned that what needed to be broken and destroyed first was our original notions, assumptions and beliefs regarding what the future of healthcare looks like, what it will require, and who will be its deliverer. 

Through the much lamented, layered bureaucracies of healthcare, ours is a clear and direct path through the sea of marketing and trade show aspirations, in which everyone is an expert, to the material, actual benefit designed and directed to your next patient. It’s real, and it works. 

We are honored and humbled by the 1,200+ global medical centers which have already joined Aidoc on this journey of learning into our future. At the same time, we are equally excited for the new relationships being established concurrent to this writing. Provided we are all continuing to break what’s not helpful and adapt to what is, it’s going to be a brighter, safer and more productive tomorrow. 

Explore the Latest AI Insights, Trends and Research

Josh Streit