AI is transforming healthcare at a breathtaking pace — and it’s just getting started. Costly administrative and provider workflows are being transformed as AI automates time-consuming tasks such as charting and coding. Cancer detection is getting faster and more accurate — allowing earlier diagnosis and treatment, and improving access. And scalable mental health and behavioral support are within view.
But AI’s potential to fill our healthcare system’s gaping prevention hole — given that up to 80% of chronic disease is preventable — is hampered in the current policy landscape.
Our fee-for-service payment model incentivizes visits, tests and procedures, not stopping poor health trajectories before they require medical attention. Increasingly, brief provider visits crowd out deeper discussions of behavior change. Prevention has long-term horizons; our current approach is short-term. Prevention also needs follow-up and coordination among providers, as no one “owns” it end-to-end. These require unreimbursed time that overworked providers lack.
Next, there are the artificial supply constraints we self-inflict. Amidst a shortage of primary care providers, federal policy caps residency training spots under limits set in the 1990s, pulling many qualified applicants from the pipeline. The old-guard medical school accreditation system prevents medical schools from opening or expanding, leaving thousands of eager would-be students with no path toward an MD. Meanwhile, skilled international doctors are hampered from practicing by visa and licensing hurdles, and a requirement that they repeat their residencies.
Laws that prevent nurse practitioners, physician assistants and pharmacists from independently providing routine, preventive care worsen the bottleneck. Countries that have relaxed such laws show improved access to preventive care and greater capacity — and, contrary to the predictions of some in the medical establishment, no meaningful reduction in safety and quality. States are coming around on this, but slowly.
Then, there is the U.S. health insurance system. Because coverage is tied to jobs and eligibility thresholds, insurance is inherently unstable. Insurance “churn” breaks continuity; people change plans and providers, so no one oversees prevention over time. The insurer paying for prevention today may not be the one benefiting from lower costs later on, creating misaligned incentives. Finally, there is administrative friction. Getting care can require navigating approvals, networks and hidden costs, so many patients delay or skip it.
In recent decades, the public health conversation has centered on expanding insurance coverage. Yet, clinical care accounts for only about 20% of health.
There has also been intense focus on reducing health disparities due to the “social determinants of health,” such as income and housing. While negative health outcomes cluster in poor ZIP codes and regions, chronic disease is not confined to poverty. Large numbers of middle- and upper-income Americans are overweight, sedentary, prediabetic or hypertensive despite having good insurance and access to care. Poor metabolic health — now recognized as a major, largely preventable driver of chronic disease — affects more than 90% of U.S. adults.
This is the “prevention paradox.” As former CDC Director Tom Frieden has written in The Formula for Better Health, “threats we can’t see kill more people than dramatic disasters, and invisible improvements save more lives than high-profile cures.” Improved blood-pressure control, for example, could save more lives than any other clinical intervention. And still, “we can’t get the most important outcome right even half the time.”
Is the promise of AI to fix all of this?
Alas, no. AI is an extraordinary high-tech tool, but at the end of the day, it’s just a tool. Addressing the prevention paradox, on the other hand, will be mostly boring and low-tech.
In a recent LinkedIn post, Frieden pointed to the success of a South Korean initiative to reduce hypertension that began in the 1990s, which led to an 87% drop in strokes by 2023. It was no tech miracle. The effort focused on widespread, regular primary care, consistent treatment, affordable drugs and a cross-sector effort involving the food industry to reduce sodium in the Korean diet. Unglamorous, durable work. AI can assist in planning, coordinating and scaling these efforts, and it will help develop better preventive drugs, but the rest is up to us.
The real problem we face is not know-how, but that a shift toward prevention requires political, economic and behavioral costs no one is willing to bear. Politicians operate on election cycles, whereas prevention pays off over years. Hospitals, specialists and pharma make money by treating disease, not by preventing it. This isn’t finger-pointing; it’s political and economic reality. And Americans say they want better health, yet we struggle to commit to even basic preventive measures — from nutrition education in schools to the mundane daily habits that actually drive long-term health.
AI may become one of the most powerful preventive tools ever developed. But unless we realign incentives toward keeping people healthy and early intervention, we risk using revolutionary technology mainly to manage diseases we failed to prevent in the first place.
Kelly Rogers Victor, Ph.D., MPH, MPP is a writer and consultant on nutrition, health policy and public health. Her columns appear regularly in The Detroit News. Reach her at Kelly@upstreamhealthconsulting.com.
This article originally appeared on The Detroit News: Why AI is not the magic fix for healthcare | Rogers Victor
Reporting by Kelly Rogers Victor / The Detroit News
USA TODAY Network via Reuters Connect
By Kelly Rogers Victor | USA TODAY Network
