Artificial Intelligence Helps Make Value-Based Care Work at Scale

My two-year anniversary at Oracle was in April. I joined this extraordinary company because its founder, chair, and chief technology officer, Larry Ellison, is committed to transforming health and solving some of its biggest challenges. Having spent my entire professional life working in the healthcare sector, including the four years I spent leading the Centers for Medicare & Medicaid Services (CMS) under the first Trump administration, I know something about those challenges. I was excited to be part of an organization with the resources and capabilities to tackle them head on.

We can all rattle off long lists of problems that exist in the American healthcare system. But I firmly believe one of the fundamental root causes of many of these problems stems from the “fee-for-service” reimbursement system created to pay providers.

Fee-for-service healthcare is a payment model based on inputs (which and how many services were provided) versus outcomes (what impact did the service have on the patient’s health). It encourages the overutilization of resources, enables bad actors to bill for unnecessary services, upcoding, and overall does nothing to encourage or incentivize providers to improve clinical and financial outcomes.

We should be paying providers to keep people healthy, not just waiting to treat them when they are sick. We should pay them to achieve defined and agreed upon clinical and financial outcomes, and we should expect them to take on a reasonable amount of risk for their performance.

This payment model is called “value-based care (VBC).” It is not a new or novel concept. A growing share of public and private healthcare payments involves some form of VBC, though the term is used generously and broadly to describe an array of payment arrangements between payers and providers. This concept has broad appeal not only in the United States, but also in other countries. The UK and the NHS are discussing having providers take on more responsibility within their Integrated Care System.

During my tenure at CMS, we advanced several initiatives to expand and refine Medicare’s role in VBC, including aggressive expansion of alternative and new payment models and incentives for providers to embrace risk sharing. Medicare Advantage (MA), which has long been a vehicle for value-based models, gained further traction during this time, while commercial payers have slowly and cautiously followed.

However, progress has been incremental, uneven, and too often focused on the financial side of VBC. Hence the controversy over MA upcoding and the growth in prior authorization requirements and denials, result that defies the entire premise of VBC.

One of the key reasons for uneven and incremental growth of VBC is that healthcare systems have lacked interoperability which made access to timely, comprehensive data difficult, and made actionable insights, required to effectively manage the care of a patient population, impossible. Evidence-based insights, when applied to clinical, claims, and patient-generated data, can significantly improve how care is delivered and managed under value-based arrangements. This is no easy task, requiring constant review of patient medical records to identify and address changes in their conditions, determine next steps, which providers are meeting negotiated quality and cost-per-service standards, and develop effective patient activation strategies to help patients manage and improve their own health.

AI is the perfect answer to these and the many other processes that must be implemented to make VBC work—and to work at scale. It makes automation of clinical and operational workflows a realistic financial investment for VBC within health systems and private physician groups. AI’s growing ability to analyze massive amounts of data, interpret it, and even make predictions based on that analysis is what is needed to support the complex processes required to manage the care of patient populations intentionally and proactively.

At Oracle Health and Life Sciences, we are working to develop a set of AI-embedded solutions that work within our next-generation EHR that will allow hospitals, health systems, providers, and other value-based organizations to successfully manage the care of their patient populations without needing to add additional staff. Our EHR is being developed to help achieve negotiated (or required in the case of Medicare and Medicaid) improvements in clinical quality and cost reductions by reducing the utilization of the most expensive services and facilities whenever possible without negatively affecting patient outcomes.

Oracle is designing AI agents that can be deployed to continuously monitor a health system’s patient population—analyzing data from a wide range of sources including medical records from both current and past providers, wearables and other remote medical devices, and claims information—to proactively engage patients with higher likelihood of needing costly emergency visits or higher potential of developing a chronic condition or becoming more seriously ill.

This can enable early and personalized interventions to be developed and implemented across entire patient populations, with recommendations coming from the clinical decision tools embedded in the new Oracle Health EHR, access for such patients prioritized, care coordination supported, and patients provided with both information and regular engagement through the system’s patient portal.

At the same time, managers from the front line to the executive suite will have access to a wide array of clearly displayed and interpreted real-time information, including social determinant data, preferred referrals, and predictive assessments of the performance of individual providers, units, and the system. The tools will be able to optimize the clinical and financial outcomes included in any number of different VBC contracting arrangements to better facilitate collaboration between the system and its payers.

With AI-based tools like these, it will no longer be possible for health systems to argue, as they have in the past, that VBC is too hard, that they don’t have the data they need or the ability to process and analyze it quickly enough to prescribe individual-level interventions to address and even prevent illness among their patients.

We expect that this new generation of AI-based tools is finally going to make possible the adoption of VBC at scale. And I firmly believe that moving to a VBC-based payment model can drive meaningful transformation for health systems worldwide. Immediately, all the incentives are focused on keeping people out of the hospital whenever possible and appropriate. They are focused on disease prevention and health promotion. They are focused on using resources wisely and efficiently—on saving money rather than generating revenue.

Our commitment to VBC is to do what is right for patients. Every person should have access to care that is designed to keep communities healthy and increase access to care for those patients with chronic health conditions. This is why embedded AI can transform healthcare. By using intelligent AI to automate and optimize critical functions like claims, prior authorizations, and payments, we hope to build a seamless and accessible healthcare system that delivers quality care for all.