Value-Based Care AI: Real MLR Reductions or Hype?

The hype around artificial intelligence in value-based care has been a gold rush for investors, all chasing projections of slashed medical loss ratios (MLRs) and perfectly optimized risk coding. Billions of dollars have poured into AI platforms that promise to change how providers handle patient populations and make money from shared savings models. Now that the initial excitement is wearing off, value-based care investors and healthcare private equity partners have to ask the hard questions. Are these AI tools actually producing verifiable clinical savings and bigger shared savings checks, or are we just looking at a pile of theoretical efficiencies with no real-world impact?

The Commercial Imperative of AI-Driven Risk Adjustment

The business case for putting AI into value-based care is pretty simple: by correctly identifying and documenting every patient condition, especially the Hierarchical Condition Categories (HCCs) that matter most, AI can get the risk adjustment right. That, in turn, means payers and providers working under CMS Medicare Shared Savings Program (MSSP) rules get paid appropriately for how sick their patients actually are. The money is what drives this: better HCC coding means higher risk-adjusted payments, which can make or break an Accountable Care Organization’s (ACO) shot at earning shared savings. Naturally, the market is now flooded with AI-powered solutions. These tools all promise to pull more data from electronic health records (EHRs), find gaps in coding, and deliver insights for managing patients better. For investors, the dream is that these platforms become the essential plumbing for any provider group trying to survive value-based care. The real question isn’t just whether the AI can technically do these things, but whether it’s producing measurable financial and clinical results out in messy, real-world clinics. We’ve seen some AI-enabled risk adjustment tools that claim to make clinical review 3X more productive, and they hit HCC discovery accuracy rates of more than 95% on actual clinical data. Some of these AI tools have been shown to lift the average Risk Adjustment Factor (RAF) score by about 0.2 for each patient they touch.

Aledade and Pearl Health: Working through the ACO Field

Aledade is one of the biggest names in the ACO management business, having built a huge network of physician-led ACOs that work within different CMS programs like the MSSP. Their playbook combines their tech platform with hands-on practice transformation help and data analytics, all designed to help independent primary care practices win at value-based care. Pearl Health is in the same crowded space, also focused on giving primary care docs the tech and services they need to succeed in value-based contracts. Both companies sell themselves on their ability to generate shared savings by improving care coordination, focusing on prevention, and, most critically, nailing risk adjustment. But when you’re evaluating these companies, you have to look past the top-line shared savings numbers. While the CMS MSSP shared savings data is public and gives you a bird’s-eye view of how an ACO is doing, figuring out how much of that success came directly from an AI tool requires some real digging. The official numbers from the Centers for Medicare & Medicaid Services (CMS) show just how big the pie is: in 2023, the Medicare Shared Savings Program created over $2.1 billion in net savings while paying out $3.1 billion to the ACOs that did the work. For Performance Year 2024, ACOs in the program hit their best-ever shared savings rates, with 75% of them earning a total of $4.1 billion in performance payments as Medicare itself saved $2.4 billion. The real work is trying to separate the AI’s impact from everything else, like better care management routines, getting patients more involved, or just running the practice more efficiently. For example, Aledade has consistently posted huge MSSP results, with its ACOs generating massive shared savings. In 2023, Aledade ACOs were responsible for $801 million in savings for the MSSP and took home nearly $538 million in shared savings payments, with an impressive 93% of their partners earning a check. Then in 2024, their ACOs saved Medicare over $1 billion and earned $775 million in shared savings, with that same 93% success rate. Since they started in 2014, Aledade’s partners have generated over $3 billion in healthcare savings. The company also locked down a $500 million senior secured credit facility in late 2025. Aledade MSSP performance reports Pearl Health’s rapid growth and move into new markets also shows that investors are buying the story. Pearl raised a $75 million Series B in early 2023. By July 2026, it had pulled in another $110 million in debt and equity, which included a $50 million Series C round. The company now manages about $3.6 billion in medical spending per year and claims it hit profitability in 2025. Its network grew to over 3,500 primary care providers in more than 40 states by 2024. For both companies, the specific, provable reductions in MLRs that came directly from their AI platforms are the key thing to investigate during due diligence. The goal for any investor is to verify the average HCC score improvements they claim from AI-assisted coding and then connect those dots to actual money in the bank.

Navina: AI at the Point of Care for Risk Adjustment

Navina’s angle is different. It puts its AI right at the point of care to help primary care docs find HCC codes during the patient visit. Their platform plugs into the EHR and gives the doctor a clean patient summary that highlights potential HCCs that might have been missed or not fully documented. The idea is that if you make this information easy to see and act on during the appointment, you can dramatically improve how accurate and complete your risk coding is. For Navina’s model to work, a few things have to be true: its AI must be accurate in finding the right codes, it has to fit into a doctor’s workflow without causing a headache, and doctors have to actually use it. This is where you need to see peer-reviewed studies on HCC coding accuracy and clinical outcomes. As an investor, you should be asking for proof that practices using Navina see a real, measurable increase in HCC capture rates compared to practices that don’t, and that this increase leads to higher risk adjustment factors without just driving up useless tests and procedures. A good number to watch is the average HCC score improvements that are attributed to AI-assisted coding. If Navina’s AI is consistently helping doctors find previously undiagnosed conditions and that leads to a higher average HCC score per patient, that flows directly to the risk-adjusted payments from the payer. The trick is making sure these improvements reflect real patient sickness and aren’t just aggressive coding that will get you audited (which is a fast way to get in trouble). The impact on MLRs should be a more accurate financial picture of the patient panel’s health, leading to the right premium payments and, hopefully, more predictable results from value-based contracts.

Auditing AI-Driven Clinical Savings Claims

If you’re a value-based care investor or a private equity partner, you have to audit these AI savings claims. No excuses. The “Core Thesis / What Happened” investigation has to go way beyond the marketing slides and get to verifiable impact. Here’s a checklist for auditing AI’s effect on MLR reductions:

• Get the Baseline: Demand the data from before the AI was turned on. What was the average HCC score, MLR, or shared savings performance? Without a solid baseline, any claims of improvement are just noise.
• Question the Attribution: How exactly is the company proving the AI caused the savings? Is it a direct, causal link they can show, or just a correlation that happened while a bunch of other things were changing in the practice? You need a methodology that isolates the AI’s contribution.
• Use the Public Data: CMS publishes MSSP performance datasets for anyone to see. Use them. You can cross-reference the numbers with claims made by the AI vendors who work with those specific ACOs. CMS MSSP public use files
• Look for Peer Review: Give more weight to companies that have published their results in peer-reviewed journals. It shows they’re willing to stand behind their methods and gives their claims an external stamp of approval. You want studies that look specifically at AI’s impact on risk adjustment accuracy and MLR.
• Talk to the Payers: Call the health plans and ask what they think. Do they accept the higher HCC coding as a true reflection of patient health, or are they flagging these providers for audits? Payer confidence is a powerful sign that the impact is real and sustainable.
• Do the Real Math: What’s the total cost to own the AI solution (think integration, training, support) when compared against the verifiable savings it generated? A positive ROI is the only thing that matters.
• Check for Model Decay: AI models can get stale and less accurate as they get retrained on new data. Investors need to ask how these companies monitor for this “algorithmic drift.” An AI model whose performance degrades over time won’t be able to spot HCCs correctly, and the whole value proposition falls apart. The story of AI-driven optimization is a good one, but the reality on the ground requires a healthy dose of skepticism. While companies like Aledade, Pearl Health, and Navina are making real progress using AI in value-based care, investors have to apply a critical eye to their claims and focus on hard data that proves actual MLR reductions and verified shared savings.
  

  Methodology and Source Note

  This report isn’t just opinion. It’s a data-driven market assessment based on a quantitative look at ACO performance and risk coding accuracy. Our work involved a critical review of publicly available CMS MSSP performance data and an examination of peer-reviewed studies on HCC coding accuracy and clinical outcomes. The organizations mentioned here, including Aledade, Pearl Health, Navina, and the Centers for Medicare and Medicaid Services, are central to the analysis. The conclusions here come from an independent evaluation of these sources, with the goal of giving value-based care investors and private equity partners a clear-eyed view on the real impact of AI in lowering medical loss ratios in these networks. Peer-reviewed studies on AI and HCC coding