How AI can help clinicians devote more time to patient care

There’s a famous British painting of a country doctor sitting in a cottage holding a sick child’s hand during a house call, which was common in those days. He’s in intense thought. All these years later we still have those cultural expectations of physicians. But we fall far short in the age of modern medicine. Doctors know it. Patients know it. And so do employers and their advisers.

The irony of that painting is there was far less that we could do for patients before the arrival of antibiotics. That child could have had scarlet fever or been bled with leeches. The tools available to country doctors then were minimal, but what they could do was be more present for patients in their home, serving as a comfort to the family when clinical science had a ways to go. 

Fast forwarding to today, clinicians now have an arsenal of technology at their disposal, such as AI. My objective for AI is to help modern doctors get reacquainted with what it means to be a country doctor in terms of helping return their focus to the bedside. Can AI help us bring back more personal touch with patients? 

Read more: AT&T and Maven expand fertility benefits to support 125,000 employees

There is a lot of talk about precision medicine, but the bulk of medicine is still cookie-cutter care. When I was a member of Duke University Medical School’s faculty, they’d often bring professors to Washington, D.C. to speak with policymakers and legislators. On one occasion, when I was giving a talk about AI in healthcare, a legislative aide asked, “What should we be doing to facilitate bringing these technologies closer to bear for health care?” My answer was that it’s not a technology problem. It’s an incentives problem (more about that later). 

The way AI operates is with an objective function. If, say, the goal were remuneration, then you can build an AI model that maximizes reimbursement, which may not benefit the patient. If we align reimbursement with beneficial outcomes for patients, then we can mitigate perverse incentives and AI won’t have a “conflict of interest.”

An ability to mine operational efficiencies on the clinical side through AI could improve patient care by freeing up busy physicians who not only have limited time for office visits, but also for their own families. After a busy day of performing far too many mundane tasks, many physicians end up firing up our laptops in the evening to finish charting clinic notes during what we like to call “pajama time.” 

This vicious cycle, often referred to as “checkbox medicine,” makes clinicians feel like highly paid data entry technicians. It’s not unusual for doctors to spend as much, if not more, time furiously typing notes on their electronic health record during patient visits as they do on actual examinations. This isn’t what anyone signed up for at medical school.

Read more: 2024 health benefits strategies: A guide to reducing costs while supporting employee well-being

I’m hopeful that AI and machine learning will create opportunities that allow physicians to spend less time on labor-intensive tasks and more time strategically thinking about their patients. Before anyone jumps to conclusions about the power and promise of AI, however, it’s imperative to first manage expectations. 

AI isn’t a magic bullet. Like any other tool, it can be used expertly or poorly. While I’ve been focusing on operational applications of AI as we contemplate applications with diagnostic or therapeutic implications, as with any innovation we must consider the risks alongside the benefits. This is why it’s critical to use our existing frameworks for clinical evidence and improvement to integrate AI into healthcare delivery in a safe and responsible way, just as we have with other technologies. 

No doubt, AI will help improve clinical outcomes and reduce wasteful spending, but there are larger systemic problems that must be considered as we continue to leverage this technology. Incentives play a role in achieving good outcomes, but it’s difficult to properly align them with fee-for-service medicine, which delivers more reimbursement to physicians at the expense of patients. We need to embrace value-based care, which aligns incentives so that physicians are rewarded for good outcomes rather than volume. AI, in turn, can be the engine that helps realign a flawed system for greater efficiency alongside better health and safety.

Read more: Autoimmune disease support might be the next big benefit in healthcare

To some extent, there will need to be regulatory and legislative solutions to help value-based care work at a national level and uniformly for everyone. One important step in this direction is the 21st Century Cures Act, which includes mandates for data interoperability for exchanging health information. 

Relying on algorithms will not magically make everything better. There needs to be more thought given to equalizing access to AI and other useful technologies. Key stakeholders in the employer-provided group health insurance market can play an important role in helping bring the promise of this vision to realization.