[[[SUMMARY_START]]]

Health systems and regulators are putting more weight on preventing illness and predicting risk before people get sick.
In the United States, Medicare payment experiments and broader accountable care programs are increasingly designed to reward prevention and better coordination.
At the same time, government agencies and medical groups are tightening expectations for how predictive AI tools should be tested, monitored, and governed.
Preventive screening guidance remains a key lever, and a recent Supreme Court decision preserved the Affordable Care Act’s pathway for no-cost coverage of many recommended services.

[[[SUMMARY_END]]]

Healthcare is gradually moving from “treating illness” to “anticipating risk and preventing disease.” The shift is visible in how care is paid for, how data and artificial intelligence are used, and how preventive services are defined and protected. Over the past year, U.S. policy signals have reinforced that direction, while also highlighting the governance challenges that come with greater reliance on prediction tools and population-level prevention.

## Payment incentives are increasingly tied to prevention

A major driver of predictive and preventive care is the way insurers and governments pay for health services. Traditional fee-for-service models pay more when more care is delivered. Value-based approaches try to pay more when outcomes improve and costs are avoided.

In Medicare, the Innovation Center continues to test and update accountable care models that push providers to manage populations, reduce avoidable hospital use, and invest in upstream care. The Accountable Care Organization Realizing Equity, Access, and Community Health (ACO REACH) model is one prominent example. It uses shared savings and losses and includes payment options such as monthly primary care payments. The model has also been updated for Performance Year 2026, including changes to financial methods aimed at long-term sustainability.

These models matter for prevention because they encourage primary care teams to identify people at higher risk earlier, close care gaps, and follow up more consistently. That can include better control of blood pressure and diabetes, support for smoking cessation, and more reliable cancer screening.

At the same time, the Innovation Center has also narrowed or ended some primary care payment tests earlier than planned, including steps affecting programs such as Primary Care First and Making Care Primary. The combined effect is a policy environment that still prioritizes prevention, but is actively reshaping which model designs it wants to scale.

## Predictive AI is spreading, but oversight expectations are rising

Predictive and preventive medicine increasingly depends on algorithms that flag risk: who is most likely to deteriorate, who may benefit from a screening test, or which patients need rapid follow-up after a hospital visit.

But the governance questions are growing louder. In January 2025, the U.S. Food and Drug Administration issued a proposed framework focused on the credibility of AI models used in drug and biological product submissions. The approach is risk-based and asks sponsors to show that an AI model is credible for a specific context of use.

Professional medical groups are also pushing for more disciplined evaluation and monitoring. In late 2025, an American Heart Association scientific advisory described pragmatic, risk-based approaches for evaluating and monitoring AI tools used in cardiovascular and stroke care, emphasizing real-world performance and the potential for bias.

Researchers have also warned that post-deployment monitoring is often missing from AI tools used in healthcare, raising concerns about how well models continue to work as clinical practice, patient populations, and data systems change.

In practice, this means healthcare organizations are being asked to treat predictive tools more like clinical products than one-time IT deployments: validate them, monitor them, and adjust when performance drifts.

## Preventive screening policy remains a cornerstone

Even as algorithms proliferate, classic preventive services—screenings and preventive medications—remain a core part of prevention policy.

The U.S. Preventive Services Task Force (USPSTF) continues to develop and update recommendations on topics ranging from cancer screening to pediatric vision screening. These recommendations are especially important because they are linked to insurance coverage rules that shape what patients can access without cost-sharing.

That link was tested in the Supreme Court case Kennedy v. Braidwood Management. The dispute focused on whether the structure and authority of the USPSTF could support the Affordable Care Act’s requirement that many USPSTF-recommended preventive services be covered without cost-sharing. In a decision issued on June 27, 2025, the Supreme Court upheld the authority of the USPSTF within that framework, preserving a central mechanism that supports broad access to preventive care.

## Governance now includes data, equity, and patient trust

The move toward prediction and prevention changes what “governance” means in healthcare.

Leaders are not only overseeing budgets and staffing. They are increasingly responsible for how risk is calculated, how people are targeted for outreach, and how fairness is tested when tools are trained on historical data. A risk score that helps one group may perform poorly in another. A care gap list may improve screening rates in one clinic but miss patients who change addresses, lack broadband, or use multiple health systems.

As preventive and predictive approaches expand—through accountable care incentives, digital tools, and screening programs—the policy trend is clear: systems are being pushed to prevent avoidable disease. The operational challenge is ensuring the methods are transparent, clinically sound, and continuously checked against real-world outcomes.

AI Perspective

Predictive and preventive care can reduce suffering when it is paired with strong primary care and timely follow-up. But prediction is not the same as prevention unless the health system can act on the signals with staff, access, and patient support. The next phase will likely be defined less by new algorithms and more by how well organizations monitor results and earn trust while using them.