In Part 4, we stated our belief that the future of healthcare belongs to the consumer. How AI is used in healthcare should also align with the consumer’s needs.
Between our blogs, there was a big announcement that headlined, “Microsoft® and Epic® expand strategic collaboration with integration of Azure OpenAI Service.”[1]
The news was fascinating, but I wonder what it can ultimately achieve besides the marketing hyperbole for Epic over AI? We recognize that both Microsoft (MSFT) and Epic command recognition in their respective technology specialty space; however, this does not mean that combining two leading companies can produce a viable ‘offspring.’ We also are focusing on AI opportunities in healthcare; from a slightly different direction. At the crux of the question regarding this combination is the acronym HIPAA, Health Insurance Portability and Accountability Act which swings a big sword in protection of patient medical records and data. More on this observation later.
The formal press release noted:
This co-innovation is focused on delivering a comprehensive array of generative AI-powered solutions integrated with Epic’s EHR to increase productivity, enhance patient care and improve financial integrity of health systems globally. One of the initial solutions is already underway, with UC San Diego Health, UW Health in Madison, Wisconsin, and Stanford Health Care among the first organizations starting to deploy enhancements to automatically draft message responses.
One of the key elements in this statement was the term “generative AI-powered solutions,” Let’s unpack this for just a moment with the definition of “generative AI.” One common definition is “Generative AI is a type of artificial intelligence technology that can produce various types of content including text, imagery, audio, and synthetic data[2]. The recent buzz around generative AI has been driven by the simplicity of new user interfaces for creating high-quality text, graphics, and videos in a matter of seconds[3].” The interesting part noted in the footnotes of the article provided 10 use cases for generative AI, but none of these align with healthcare or the apparent direction of the Microsoft – Epic announcement other than the drafting of messages. I’ll just leave this right here.
In previous blogs, we have noted some of the possible applications for AI in healthcare, clearly not all, and some of them have to do with decision support for physician and possible triage for patients. Further, we noted our view and our path as a company that puts the patient/consumer at the center and in control of the data. What we are very cautious about is the possible direct, indirect, or unexpected access of Protected Health Information or PHI [4]of individual patients.
In order to have real AI opportunities, the natural language AI will need access to real-time medical information not just from open sources, but also from secure or HIPAA-protected sources. In the last few years, we have witnessed many healthcare IT companies have been fined because they shared or allowed non-healthcare entities to have access to patient information. Some were thought to be de-identified, but it was later determined that with the access to deep data, the information from many locations, for example social media companies, could be used to re-engineer and then re-identify patients.
This deal is going to be tricky for industry behemoths in their individual space. Though the plan seems to be very reasonable, to reduce cost and improve quality of care, those are very difficult to achieve at the highest level. MSFT also noted the industry pressures in their press release:
Leading industry experts have highlighted the urgent need for health systems and hospitals to address intense pressures on costs and margins. Approximately half of U.S. hospitals finished 2022 with negative margins as widespread workforce shortages and increased labor expenses, as well as supply disruptions and inflationary effects, caused expenses to meaningfully outpace revenue increases. Industry participants recognize that achieving long-term financial sustainability through increased productivity and technological efficiency is a mission-critical strategic priority.
Essentially, we see the same problem, but a materially different solution. Hospitals are the highest single-cost of healthcare, and as much as we all need them, real innovation in patient health is not found in hospitals, but way before things become acute. The innovation of care is much closer to the primary care providers, the systems like Management Services Organizations (MSOs) that provide some of the most innovative care to Medicare patients and Medicare Advantage Beneficiaries in the USA.
Yet, we hold steadfast to the belief that most innovation will occur with the patient or consumer, who want to exercise and have more involvement in their own care. People are becoming more use to technology around them, on their body, in their homes, in their cars in a variety of non-invasive information gathering from Internet of Things (IoT), and their smartphones or exercise monitors, etc..
Once the consumer is the control point of their medical data, with a Centralized Medical Record combining and then analyzing all their data, THEN AI can be a total change agent. Most of the information that the Epic system has is acute, in part because they are so integrated into the biggest hospital chains in the USA, along with Cerner® and Meditech®.
Once the consumer has direct access to most, if not all their medical records outside of the biggest Acute an Ambulatory EHR, then we will really see consumer directed efficiency and cost reductions. Until then, these are much like the much-expected partnership with Amazon®, JP Morgan®, and Berkshire Hathaway®, that created Haven®, that was eventually “disbanded.”[5] Healthcare continues to prove that it is more complicated than even the largest, well-funded companies America can fully comprehend.
[1] https://news.microsoft.com/2023/04/17/microsoft-and-epic-expand-strategic-collaboration-with-integration-of-azure-openai-service/
[2] https://www.techtarget.com/searchcio/definition/synthetic-data
[3] https://www.techtarget.com/searchenterpriseai/definition/generative-AI#
[4] by HIPAA regulations
[5] https://www.cnbc.com/2021/01/04/haven-the-amazon-berkshire-jpmorgan-venture-to-disrupt-healthcare-is-disbanding-after-3-years.html