The Evolving Reality of AI in Healthcare: A CMIO‘s Perspective
The hype around Artificial Intelligence in healthcare has been…intense. After attending recent industry discussions and reflecting on the past year, it’s clear we’re navigating a complex landscape. It’s a world of genuine potential, frustrating roadblocks, and a healthy dose of vendor-driven realities. As a seasoned CMIO, I want to share my take on some common pronouncements about AI, and more importantly, spark a conversation about your experiences.
Let’s break down some of the predictions made recently, and where we stand now.
Evaluating the Promises: A Year Later
Here’s a look at some statements circulating last year, and my assessment of their accuracy today:
* “AI will fundamentally change healthcare as we know it.” Mostly true, but the change is proving more nuanced than initially predicted.The foundational shift is happening, but adoption and impact are uneven.
* “I’m cautiously optimistic about generative AI in clinical applications; it seems like just one more thing.” Spot on. We’ve seen explosive growth, but also a growing skepticism. The initial excitement is settling into a more realistic assessment of benefits and limitations.
* “AI is going to bring back the humanity in medicine. We will actually have time with patients rather than just taking a bill-and-go approach.” A 50/50 proposition. While AI can reduce administrative burden, studies haven’t consistently shown that time savings translate to more patient-facing time. More robust research is crucial.
* “By 2025, this is totally going to bring the joy back into medicine.” Regrettably, a thumbs down. We’re past 2025, and the core challenges of medical practise - administrative overhead, burnout, and systemic issues – remain.
* “Data quality isn’t attractive. It’s not going to wind up on a movie poster.” Subjective! Many of us do find beauty in clean, normalized data. It’s the foundation of reliable AI, and frankly, a well-maintained dataset is a thing of pride.
* “Vendors kind of care about health,but really want to make money.” Sadly,all too true. It’s a spectrum, but the profit motive is undeniably a driving force.
* “Just because it has AI in the name doesn’t mean it’s useful.” Absolutely. AI is a buzzword, and many products leverage it without delivering meaningful value.
* “I hate the subscription model. You used to be able to just buy stuff.” A sentiment shared by many. The shift to perpetual subscriptions continues to be a pain point.
* “I’m tired of hearing about ‘move fast and break things.’ Vendors need to move fast, but also heal their broken things just like hospitals do.” A worldwide truth among CMIOs. Stability and reliability are paramount in healthcare.
* “AI is just giving us an escalating arms race of appeals and denials. They say we’re diagnosing too much sepsis even though they wanted us to find sepsis sooner.” The “Spy vs. Spy” dynamic is real. Payers and providers are constantly reacting to each other’s AI-driven strategies.
* “Ambient documentation adoption will be limited because the operations people want a tangible ROI. How do you put a dollar amount on physician wellbeing? Our arguments about turnover and recruitment fall on deaf ears. they’ll probably just pass the cost on to clinicians.” Unfortunately, this is playing out in many organizations. Demonstrating ROI beyond cost savings remains a notable hurdle.
The Challenge: Automating the Odious
One of the most compelling ideas I heard recently was a speaker’s challenge to the audience: “Go play with ChatGPT and try to make it do the part of your profession that you hate.”
It’s a powerful concept. But where are the real wins? what tasks are you successfully offloading to AI?
Here are some areas where I’m seeing potential, and where I’m still searching for solutions:
* Successful Automations:
* Summarizing lengthy documents: AI excels at condensing complex details, saving valuable time.
* **Drafting
