The intellectual foundation — patent-pending, architected, and ready to build.
Every CAR-T program building AI infrastructure hits the same wall — in five places simultaneously. Mednos AI was the first system designed to remove all five.
The same analyte reads differently on different instruments. The failure is not in YOUR model — it is in the measurements the model receives.
Mednos AI turns inconsistent measurements into comparable, reliable data across platforms — while preserving full traceability and control.
As CAR-T scales, no computable standard exists to benchmark a facility's execution risk against regional peers before a payer authorizes a $400K+ therapy.
Mednos Compares™ →Serious adverse events are best projected on temporal changes — not single lab values at one point in time. Delayed multi-organ toxicities across multiple organ systems are not addressed by any currently marketed clinical decision support product.
Mednos is Temporal™ →True de-identification cannot be guaranteed — a legal blocker that stops multi-institutional deployment cold. What leaves each Mednos institution is a governed, de-identified summary of measurement environment characteristics. Not patient data in any form.
Mednos AI was designed so this blocker never exists.
Your EHR vendor cannot build this. Clinical platforms govern patient workflows — not instrument-level measurement control, GMP environments, or PCCP-compatible model update controls.
The right vendor for CAR-T AI infrastructure is not your incumbent system. It has to be purpose-built.
PCCP (Predetermined Change Control Plan): The FDA framework that allows a cleared AI system to update its parameters post-market without a new regulatory submission, provided governance conditions are defined and approved in advance.
Mednos is Use-Case Integrated™ →The safety infrastructure that REMS provided does not disappear with the requirement. Regional hospitals now have the authorization to treat — but not the AI infrastructure to do it safely and defensibly.
REMS is gone. The safety infrastructure gap it leaves at regional hospitals is not. Mednos AI fills it.
REMS (Risk Evaluation and Mitigation Strategy): An FDA-mandated safety and access control program for high-risk therapies. CAR-T therapies historically required REMS certification — mandating specific treatment centers and strict monitoring protocols. These requirements were removed in June 2025 as clinical management of toxicities matured.
Day +7 · CRS 68% · Three orders pending signature
CRS (Cytokine Release Syndrome): Acute inflammatory toxicity triggered by CAR-T cell activation — the primary serious adverse event in the first days post-infusion. ICANS (Immune Effector Cell-Associated Neurotoxicity Syndrome): Neurologic toxicity occurring in the acute post-infusion window. CirAE (CAR-T-associated immune-related Adverse Event): Delayed multi-organ toxicities emerging weeks to months post-infusion — distinct from CRS and ICANS.
Risk score 1.15 · Care agreement unlocks approval
Same patient · Four platforms · Four different answers
Ken Harris is the Founder and CEO of Mednos AI. He brings more than three decades of experience in healthcare and seven years in healthcare AI, having led the development and deployment of more than 100 clinical and clinical research AI tools — working alongside teams of MD and PhD subject matter experts across the full healthcare enterprise.
Prior to founding Mednos AI, he held senior commercial AI leadership roles at a global cloud and AI platform and at a leading cell and gene therapy organization. Earlier in his career, he founded a cell therapy company that reached Nasdaq — and witnessed firsthand the computational and data infrastructure failures that limited the field's potential. Those failures became the founding thesis of Mednos AI.
He has direct operational experience across clinical decision support, regulated manufacturing environments, complex multi-modal precision medicine AI models, and outcomes-based reimbursement — the four domains the Mednos AI patent portfolio addresses.
Mednos AI, Inc. is incorporated in Delaware and headquartered in Boston, Massachusetts.
Mednos AI operates on a federated computation model. What leaves your institution depends on which capability is in use — but in every case, it is a governed, de-identified summary, never patient data in any form.
For lab measurement harmonization: a de-identified characterization of your measurement environment. No patient data in any form — no patient identifiers, no patient measurements, no information from which a patient record could be reconstructed.
For pre-authorization governance: de-identified operational performance indicators specific to your facility. No patient records, no patient-level data in any form.
On-premises deployment is supported for payers and health systems with FedRAMP or data residency requirements.
"The privacy guarantee is an architectural property of the system. It is not a contractual commitment that depends on vendor compliance — it is a structural impossibility for PHI to leave your network perimeter."
Each institution processes its own data locally. Only a governed, de-identified summary transmits centrally. Raw patient data never moves.
The governed summary transmitted describes your measurement environment — not patient biology. It contains no patient identifiers, no patient measurements, and no information from which a patient record could be reconstructed.
For payers, government contractors, and health systems with strict data residency requirements, the platform deploys entirely within your own infrastructure. No patient-derived data, model outputs, or authorization decisions are transmitted outside your network perimeter.
This is not a theoretical guarantee. It is a network already built to prove it.
Each node transmits only a governed, de-identified summary to the governance core — never patient data in any form. The core continuously refines calibration and returns governed parameters to every participant.
Each family is independently licensable and architecturally interconnected. The Lab Harmonization family is the data foundation. Pre-Authorization and Clinical AI sit above it, sharing the same governance architecture.
Every major healthcare AI platform was built for breadth. None was built for the regulatory and data infrastructure of engineered cell therapy. These six provisionals cover that gap — as an integrated, governed stack. Not replicable by a horizontal platform. Not on any timeline.
Early licensees gain a calibration advantage that compounds. The first to license does not just acquire a technology — they acquire a head start that grows.
You can build the model.
You cannot replicate the network.
The first to license shapes the standard.
Every participant after strengthens it.
That advantage is not available to a late entrant — at any price, on any timeline.
The portfolio serves the drug manufacturer's clinical affairs team, the hospital and provider, the clinical research statistician, and the manufacturing quality assurance department — across six distinct buyer segments. Licensing can be structured as field-of-use by segment, exclusive or non-exclusive, or as full portfolio acquisition. View the interactive buyer map →
CAR-T clinical decision support at the point of care. EHR-integrated CDS with clinician review workflow. Payer authorization artifact production for value-based payment.
Primary: Family 01 · Family 02GCP clinical trial lab harmonization with CDISC-formatted output. GLP nonclinical study harmonization for IND/BLA regulatory submissions. GMP batch-release governance.
Primary: Family 03Pre-authorization governance for CAR-T commercial launch. Toxicity prediction for Phase 2/3 clinical programs. Full portfolio acquisition available.
Primary: Family 01 · Family 02Computable facility execution risk scoring for prior authorization — replacing subjective criteria with a benchmarked, auditable standard. Immutable audit trail per authorization decision.
Primary: Family 02GMP lot-release intelligence with versioned, audit-traced control per batch. Multi-site comparability packages for regulatory submissions.
Primary: Family 03GCP central lab multi-instrument and reagent-lot normalization. CDISC-formatted harmonized datasets for sponsor eCTD submissions. A single CRO central lab is covered independently.
Primary: Family 03Every patent family was designed for FDA's current and emerging AI guidance — and the governance framework is designed to align with the EU ATMP regulatory framework for global programs. Not adapted after the fact.
For global CAR-T programs, the governance design aligns with Regulation (EC) No 1394/2007, EMA's 2025 guideline on quality, non-clinical and clinical requirements for investigational ATMPs, and is positioned to support forthcoming EU pharmaceutical legislation reforms emphasizing digitalized regulatory processes and advanced therapy evaluation.
Designed for FDA regulatory compliance from inception. Compatible with the Predetermined Change Control Plan framework — governance updates without re-submission.
Human-approved model updates, electronic signature workflows, and complete audit records across GCP, GMP, and GLP. Every decision is traceable to a specific governed version state — permanently and without modification.
The lab harmonization family produces harmonized datasets with versioned governance identifiers carried directly into CDISC-formatted regulatory submission packages — no manual reformatting required.
The system enables pre-authorized parameter improvements under a Predetermined Change Control Plan — the key FDA regulatory barrier this portfolio structurally resolves for every buyer segment.
FDA's 2025 draft guidance on the use of AI to support regulatory decision-making for drug and biological products specifically addresses AI use in the manufacturing phase for product quality decisions — including model credibility, audit traceability, and data integrity. The Mednos AI lab harmonization and GMP governance architecture was designed in direct alignment with this emerging framework. In regulated environments, draft guidance is directionally authoritative. We build to it.
This portfolio is available for exclusive or non-exclusive licensing, field-of-use licensing by buyer segment, and outright acquisition. Every provisional was built for regulatory compliance from the ground up — PCCP, 21 CFR Part 11, and FDA submission-ready throughout. Request the portfolio overview or contact us directly to begin.