Outcome-Aligned Payment for AI-Detected Diagnostic Safety
A 36-month demonstration in which Aidoc's clinical AI, deployed at scale across Sentara Healthcare, generates measurable reductions in high-harm diagnostic errors; and Sentara Health Plans captures the downstream cost savings under an independently adjudicated, three-component payment model.
Diagnostic Error Kills at Scale. Five Pathways Are Now Solvable.
Newman-Toker et al. (2024) established the burden: 795,000 Americans suffer serious harm each year from diagnostic errors. The majority of that harm concentrates in a small number of high-acuity, time-sensitive conditions. What distinguishes those conditions today is not the absence of medical knowledge but the absence of timely detection.
Aidoc's AI algorithms, trained on millions of imaging studies and deployed across hundreds of U.S. health systems, detect five of those high-harm conditions in real time; before clinical deterioration, before the admission that follows the missed diagnosis. Sentara Healthcare, with its breadth of imaging volume and its integrated payor structure, is the natural first site at which the clinical detection story and the payment model story are told simultaneously.
Five High-Harm Conditions. One Imaging AI. Real-Time Detection.
Each pathway was selected on two criteria: (1) the condition causes serious harm when missed, and (2) Aidoc's AI detects it on imaging studies already performed as standard of care. No new testing. No workflow disruption. The algorithm reads every relevant scan and flags the finding before the radiologist closes the study.
Five Actors. One Closed Loop.
The model works because each actor occupies a structurally distinct role. Aidoc detects. Sentara Healthcare deploys and treats. Sentara Health Plans captures the downstream savings. FHG measures independently so both sides trust the reconciliation. SOLV provides the coding substrate that makes the measurement machine-readable and nationally auditable.
No role is redundant. Remove any one actor and the model does not close.
Three Components. One Arc. Aligned from Day One.
Jesse's proposal to Sentara Health inverts the standard vendor contract. Rather than a flat subscription fee disconnected from outcome, the payment model is structured in three time-bound components that progressively transfer risk from Aidoc to the demonstrated value of the deployment. FHG adjudicates each gate independently; neither Aidoc nor Sentara sets the measurement methodology.
Independent Adjudication Is Not Optional. It Is the Product.
Outcome-aligned payment requires one thing that neither Aidoc nor Sentara can provide for themselves: a trusted neutral party whose measurement both sides accept. Without independent measurement, the shared savings negotiation becomes a dispute over attribution methodology. With it, the 36-month reconciliation is an audit, not a negotiation.
FHG occupies that role. Our methodology uses commercial claims data, Medicare Advantage encounter data, and the SOLV coding substrate to construct a pre-agreed attribution framework that identifies, for each Aidoc-flagged case, the downstream care path and its cost. The difference between the actual downstream cost and the expected downstream cost in the absence of AI detection is the basis of the shared savings calculation.
- Claims-Based AttributionCommercial claims and Medicare Advantage encounter data; not hospital billing; as the authoritative cost source for the shared savings calculation.
- Attribution Agreed in AdvanceThe methodology is specified in the ISOW before deployment begins. It cannot be rewritten after the data are known.
- 36-Month Reconciliation CycleA single reconciliation at month 36 closes the demonstration. Interim reporting at 12 and 24 months provides early verification without reopening the attribution framework.
- Independence from Both PartiesFHG has no financial stake in the shared savings calculation. We are compensated for measurement, not for a share of the savings. Neither Aidoc nor Sentara can direct the methodology.
- CMMI-Ready DocumentationThe measurement framework is designed from the outset to satisfy CMMI Innovation Center evidentiary requirements; enabling the Sentara demonstration to serve as the foundation for a national CMMI model application.
- SOLV IntegrationThe SOLV coding layer ensures that diagnostic codes generated by Aidoc's AI are machine-readable within the national CRG architecture, making attribution portable across payors and geographies.
SOLV Makes the Reconciliation Machine-Readable. Everywhere.
SOLV's CRG (Clinically-Related Group) architecture is the national coding standard for downstream care attribution. SOLV's autonomous coding and ambient AI platform is contracted with virtually every major health plan, Medicare, Medicaid, and numerous VA and international health systems.
What this means for the Sentara model: when FHG constructs the attribution methodology at month 36, the diagnostic codes flowing from Aidoc's AI detections map directly into the CRG system that SOLV already maintains. There is no bespoke coding reconciliation required. The shared savings calculation is machine-readable, auditable; and critically, portable.
Portability is the key word. When the Sentara model moves to CMMI, HCSC, or any subsequent replication site, the coding infrastructure that governed the Sentara demonstration already exists at those sites. SOLV is the coding substrate not just for Sentara; it is the substrate for the national model.
The Integrated Payor/Provider Structure Is the Load-Bearing Architecture.
Most health systems cannot be the first site for this model. The model requires that the downstream cost savings accrue to an entity that also controls the payment to Aidoc. In most health systems, the hospital and the health plan are owned by different entities. The cost savings flow to the insurer. The hospital has no claim on them. The model does not close.
Sentara is different. Sentara Healthcare (health system) and Sentara Health Plans (commercial and Medicare Advantage payor) are both subsidiaries of Sentara Health: the same parent board, the same consolidated financial statement. Every dollar of avoidable downstream cost reduction that Aidoc detection enables accrues to the Sentara Health parent.
That closed loop is the structural prerequisite. Sentara has it. Most systems do not.
- Health System Footprint12 hospitals across Virginia and North Carolina; major imaging volume in Hampton Roads, Richmond, and Northern Virginia.
- Clinical DeploymentAidoc's AI across imaging workflows: ED CT, inpatient CTA, elective imaging; flagging all five pathways in real time, every study, every site.
- Outcome TrackingClinical outcomes tracked at the encounter level and linked to downstream claims via FHG's pre-agreed attribution methodology.
- Commercial + Medicare Advantage PrimaryNot Medicaid-centric. Commercial and Medicare Advantage are the tracks on which downstream avoidable cost is most cleanly measurable and most materially recoverable.
- Closed-Loop Cost CaptureAs both the health plan and the health system parent, Sentara Health Plans captures every avoidable re-presentation, readmission, or downstream complication that Aidoc detection prevents.
- CMMI CandidacyCombined health system and health plan under integrated governance satisfies CMMI's preferred implementation partner criteria for a payment innovation model.
Sentara Proves It. HCSC Operationalizes It at Scale.
The Sentara demonstration is not an end in itself. It is the evidence base for the national replication. The entity best positioned to take that replication to scale; and the one with the deepest strategic alignment to do so; is Health Care Service Corporation (HCSC).
HCSC operates BlueCross BlueShield plans in five states: Illinois, Texas, Oklahoma, Montana, and New Mexico; covering approximately 16 million members. Unlike Sentara, HCSC is a pure payor. But the diagnostic safety model does not require integrated governance once the CMMI national model is in place. Under a CMMI model, the shared savings structure is federally sanctioned and measurable against any CMS beneficiary population. HCSC's Medicare Advantage and commercial plan volume across five states represents one of the largest single-organization replication opportunities in the country.
The bridge from Sentara to HCSC runs through a name both FHG and Aidoc know well.
The largest mutual health insurer in the United States. On a per-member diagnostic safety burden analysis; using the Newman-Toker (2024) rates as a population-level input; HCSC's five-state footprint represents a high-harm diagnostic event volume that dwarfs any single health system demonstration. The Sentara model, once CMMI-sanctioned, maps directly onto HCSC's plan infrastructure without requiring integrated health-system ownership.
Barbara McAneny, MD; hematologist, oncologist, and former President of the American Medical Association (2018–2019); is the physician-executive who connects the Aidoc/Jesse Ehrenfeld orbit to HCSC/BCBS-NM. Jesse Ehrenfeld, MD, MPH caught up with Barbara as recently as June 2026. The AMA alumni connection, the shared AMA-President credential, and Barbara's roots in New Mexico's health care community position her as the natural warm introduction from Jesse to HCSC's strategic leadership.
The sequencing: Sentara demonstration, then CMMI application, then Barbara/HCSC engagement, then national replication. Each step accelerates the one after it.
Healthsperien. The Intellectual Infrastructure Beneath the Model.
Healthsperien is not one of the five actors in the Sentara partnership structure. It does not hold a seat at the payment table. But the analytical foundation on which the five-pathway selection rests; the diagnostic error burden quantification, the total-cost-of-care attribution framework, the downstream avoidable-cost modeling; reflects precisely the kind of rigorous health analytics work that Healthsperien has built its practice around.
The Newman-Toker et al. (2024) research that anchors the 795,000-serious-harms figure is the lineal descendant of years of diagnostic error burden quantification by the clinical epidemiology and health economics community. Healthsperien's capabilities in total-cost-of-care analytics; the ability to model, by condition, what a missed or delayed diagnosis costs a payor across the downstream care trajectory; are the specific analytical tools that make the shared savings model's measurement credible to a sophisticated payor like Sentara Health Plans.
The selection of the five pathways was not arbitrary. Each was chosen because: (1) its serious-harm burden is quantifiable from published epidemiology; (2) its downstream cost is attributable in claims data; (3) Aidoc's AI has demonstrated detection accuracy; and (4) the avoidable-cost calculation survives scrutiny from a sophisticated actuary. Healthsperien's analytical DNA is visible in that selection logic, even if Healthsperien does not appear in the proposal's payment architecture.
Sentara Is the Proof. CMMI Is the Prize.
The Center for Medicare and Medicaid Innovation (CMMI) exists to test alternative payment models and scale the ones that work. The Sentara diagnostic safety demonstration is designed, from its first day, to generate the evidentiary record that a CMMI Innovation Center application requires: a pre-specified attribution methodology, independently adjudicated, across a defined patient population, with a documented cost and outcome trajectory.
A CMMI "Diagnostic Safety Payment Innovation" model would sanction the three-component payment structure at the federal level; making it replicable at any health system or health plan in the country without a bespoke negotiation. The Sentara plus HCSC combination; one integrated health system and one major national payor, connected by FHG measurement and SOLV coding; provides the clinical and payor-side validation that CMMI requires to advance a model from demonstration to national adoption.
Four Moves. This Week.
The thread is warm. Jesse replied within minutes to the CRGs preview. The Sentara proposal is in hand. The ask is a conversation, not a contract.