Sandy
Sandy Author at Health Tech Bytes. Cloud Architect, data science and AI practitioner, health informatics strategist

Can healthcare embrace the AI revolution?

Published on | 5 min read

Can healthcare embrace the AI revolution?

“Healthcare loves a new new thing and generative AI is now it”, stated Robert Galvin, MD, an executive advisor of The Blackstone Group, in this Forbes article predicting the top 2024 healthcare trends. In following the latest announcements from VIVE to HIMSS, it was refreshing to see emerging technologies take center stage to great fanfare. However, as much as new technology is revolutionary, healthcare has always been a laggard. Given this unique time of technological progress, can healthcare embrace the AI revolution effectively?

Generative AI is having its moment in the clinic

AI is nothing new, and I’ve noted in recent conversations that this wave seems reminiscent of IBM Watson’s heyday. Has anything changed since then? To a degree, I would say yes. One could consider this moment in AI as the “perfect storm”, as hospitals look to regain financial stability post-pandemic and computing has become more powerful, cheap, and smart. As highlighted in Harvard Business Review’s Why the Tech Industry Won’t Disrupt Health Care, healthcare organizations must transform themselves to stay relevant. Failure to reduce costs, improve quality, and expand access will render them obsolete.

Technology transformation is nothing new for healthcare. Providers have evolved consistently but slowly, including reimbursement changes, EHR adoption, and increased focus on whole person health. I would hardly call any of these changes revolutionary though, unlike Big Tech which has completely overhauled other industries. The HBR authors argue that this is by design, and any new entrants will find it difficult to uproot existing healthcare entities.

At the same time, we should not cast doubt that generative AI’s capabilities are impressive. While the machine learning models underpinning generative AI have been around for decades, its popularity has soared due to the complexity of tasks (images, audio, etc.) that can be performed and the amount of data (scale) on which these tasks are trained. This technology is being seen as a huge boon in the clinic, with use cases ranging from patient case summarization to drafting responses to patient messages. I am bullish on these types of administrative use cases that can save time and cost.

Revolutionary vs. evolutionary technology

But could healthcare eventually be disrupted in the same ways that Big Tech has done with other industries?

Dr. Galvin goes on to say that “CHAT GPT 4 and other large learning and generative models are truly revolutionary, but change in clinical medical is always evolutionary”. In a 2019 interview with Dr. David Blumenthal and Dr. Namita Seth Mohta, they went on to highlight some of the challenges with new tech in healthcare:

“Galvin: Sometimes the unintended consequences exceed the benefits, to be honest. When you get into a system as big as health care, as resistant to change in health care, and inherently much more complicated, this is not buying goods and services over Amazon; this is not getting an Uber or using Lyft. These are in many cases very sick people with complicated diseases in a system that’s already very complicated.

Blumenthal: The 5% of patients who account for 50% of expenditures need continuous longitudinal care, comprehensive care, and care that’s knowledgeable about their complicated problems. Their simple problems are often complicated by their other conditions. So it’s hard to imagine solving our major health care issues, especially around cost, without developing effective systems for caring for that 5%.”

This made make me think a bit about what constitutes revolutionary vs. evolutionary technology. In another HBR article, author Jeffrey Stibel writes about the automobile as a revolutionary innovation, while building better engines for car as an evolutionary innovation. Revolutionary innovations are based on novel ideas, while evolutionary innovations build on existing solutions and limitations.

One of the challenges I see with revolutionary technology in healthcare is that it may not significantly impact the 5% that Dr. Blumenthal cites. If we’re interested in reducing cost for the 5%, I think generative AI can only go so far. In contrast, the 95% of patients who have everyday health concerns will likely be revolutionized by generative AI. However, it may not move the needle very far in reducing cost.

Moving healthcare beyond evolutionary technology

How do we manage such discrepancies? As Dr. Blumenthal notes, there needs to be effective systems (beyond technology) to care for the 5%. If we want healthcare to transform as an industry, it is imperative to involve all stakeholders across the ecosystem. Although there have been well-established partnerships between technology companies and health systems, new collaborations have recently sprung up among a more diverse set of players. These include providers, payers, technology companies, government entities, community health organizations, and patients. Recent groups include the Coalition for Health AI (CHAI), Trustworthy & Responsible AI Network (TRAIN), and VALID AI, among others. These are all exciting developments to address healthcare challenges using AI.

Conclusion

It is clear that AI in healthcare is making significant headway in the last year, but in my view, it is still an open question of how much change it will affect. On the one hand, generative AI may handle simple healthcare problems more swiftly, but have limited impact for complex problems that are inherently more costly. It is interesting to see concerted efforts to make AI accessible for all healthcare stakeholders, and I will be following these groups over the upcoming months.

References

comments powered by Disqus