By Mark Watts
If a health system applies AI to accelerate patient access but provides the same care delivery model built upon perverse incentives, you’ll accelerate all the broken consequences after that. More patients seen, same suboptimal outcomes.
One thing that’s always been true about technology, and that remains true in the age of AI, is that tech mostly makes the existing system move faster. This will happen even faster with AI. And, on the rare occasions tech restructures a system, it takes decades. However, we don’t have that kind of runway here.
The problem is that if a current system is full of fundamentally flawed incentive models and dumb processes, the AI and tech just churn out more of it.
We see it already with clinical documentation: AI scribes don’t eliminate having to document low ROI/nonsensical things. It just hides the work from clinicians. Which is OK, it’s not like that will impact patient outcomes in an obvious way.
But when we look beyond administrative/operational workflows, and into actual clinical care, we start to see all kinds of ways that AI-ing the current world accelerates stuff that doesn’t make sense.
In a fee-for-service world, maybe fully optimized AI scribes and co-pilots let more patients have primary care visits in a day. Cool, I like that. Yet, it doesn’t magically make the healthcare system care about population health.
AI will get good at screening surgical patients and predicting risk, but that also means those organizations who are cherry picking healthier patients will just get much better at that – and make it more likely that complex, sicker patients have a harder time getting care.
Even in a sort of value-based payment world where readmissions are penalized, AI will help hospitals even further optimize keeping readmissions at the lowest possible rational rate. At some point, the marginal cost of preventing the next readmission exceeds the penalty you’d avoid, and AI will be used in real-time to even better toe that line.
Whatever rules you put in place, whatever system you design, whatever incentive models you create – AI will help health systems optimize the heck out of it. While most people entered healthcare to help patients, the reality is that health system boards carry lots of fiduciary obligations. If a CXO is not willing to stay in line to hit the KPIs, the board will find someone who will.
Which means short of fixing its own perverse incentives, what healthcare will always need more of is courage to keep doing what’s right for patients even when it’s not the most profitable decision.
Yes, we will need payment models, regulation, etc. to change too. We all know how long that will take.
So, in the age of AI, where courage is already in short supply and the ability to optimize the existing system gets only better, a culture of courage is increasingly important.
Mark Watts is an experienced imaging professional who founded an AI company called Zenlike.ai.
