Mendel

Clinical reasoning over unstructured records is the entire product. Hypercube abstracts, structures, and answers questions across pathology reports, genomic data, physician notes, claims, and electronic data capture records, and the company's stated design premise is that generic language models are insufficient for this task. There is no services or data brokerage layer; the platform is sold to operate on the customer's own data.

Unusually well specified because the company submitted the autonomy question itself to prospective study. The Penn evaluation is structured around three explicit arms, AI alone as an autonomous algorithm, human alone as current practice, and human plus AI where the algorithm supplies a rank ordered candidate list and abstracted elements to a research coordinator who decides.

Designing a trial that isolates the autonomous configuration from the assisted one, rather than asserting that a human is in the loop, is the most rigorous treatment of this axis among trial vendors here. In deployment the tool ranks and abstracts; the coordinator determines eligibility.

The architectural claim is specific, unusual, and motivated by a stated failure mode: Hypercube pairs large language modeling with symbolic reasoning over a clinical hypergraph, adopted because the company argues pure neural approaches cannot discern clinical nuance and are susceptible to hallucination. This is the same neuro symbolic reasoning that RAAPID applies in risk adjustment.

Critically, every answer is tied to discrete highlighted evidence in the original record, which the company frames as making the algorithm's failure mode immediately visible rather than obscured. Publishing the reason for an architecture choice and building for inspectable failure is a materially higher standard than describing capabilities.

Two concrete mechanisms sit here rather than assurances, and one route is answered while the others are not. The platform is cloud agnostic and can be hosted inside the customer's own environment, so an institution can process clinical records without them leaving infrastructure it controls, and a dedicated capability de identifies protected health information from clinical records, which is a control operating on the data rather than a statement about handling it.

Evidence linking functions here too, since a system that records which source text produced each answer leaves an audit path through what was read. What is absent is everything about the routes a customer might actually take. No model or model family is named, no foundation model provider is identified, no hosting arrangement is described for the vendor operated option, and no sub processor list was located, so the customer hosted path is bounded by the customer's own perimeter while the alternative paths are undescribed.

Nothing states retention, or whether customer records contribute to model development, which matters for a system whose reasoning layer sits over a clinical hypergraph that presumably improves as it sees more clinical language. Ask which deployment applies to your contract, and for the hosted route ask for the model provider, hosting arrangement, sub processor list, retention and the training position.

Independent prospective evaluation is underway at a major academic center, which is more than most vendors in this category can show. Two University of Pennsylvania registered studies assess AI augmented record abstraction for trial prescreening, one leveraging the EA8191 INDICATE Phase III prostate cancer trial as a historical control comparison, with hypotheses covering efficiency, accuracy, and diversity of prescreening.

Earlier work with Penn oncologists benchmarked the platform against GPT-4 and Llama2-7b, reporting domain specific models outperforming generic ones. The benchmark comparisons are vendor initiated, and full published results of the prospective studies were not located at the time of review, so the evidence is credible and in progress rather than settled.

Two concrete mechanisms rather than assurances. The platform is cloud agnostic and can be hosted inside the customer's own environment so patient data never leaves, and a dedicated capability de identifies PHI from clinical records. Evidence linking, where each answer traces to highlighted source text, functions as a safety control as well as a transparency one, since an unsupported assertion is visibly unsupported. What is absent is published hallucination rate measurement or formal failure mode documentation.

Converted from Not Rated. The prior reading was right and is now confirmed on the company's own pages rather than inferred: the architecture is built for regulated data and the contractual posture is not published.

What is confirmed. The product documentation states the platform is cloud agnostic and can be hosted in the customer's own environment so that data never leaves, with vendor hosted options offered as an alternative. Redaction is listed among the platform's stated use cases alongside abstraction, chart review and cohort building. Native availability on a major data cloud means it can operate inside a customer's existing warehouse tenancy.

That matters here more than it would elsewhere, because it changes which legal question applies. Where the platform runs entirely inside the customer's own environment and no records are disclosed to the vendor, the business associate framework is not engaged in the ordinary way, and the customer's own existing safeguards govern. Where the vendor hosts, it plainly is engaged, and nothing addresses it: no business associate agreement availability statement, no characterisation of the company's role, and no identification of which entity would sign. The answer to this axis therefore differs by deployment, and a buyer must know which configuration is being sold before the question can be answered at all.

A second legal question sits alongside the first and is specific to this lane. For the trial prescreening use, screening records to identify candidates happens before anyone has consented to anything, so the operative instruments are institutional review board approval and an authorisation or waiver rather than a business associate agreement. A vendor can be correctly positioned on one and silent on the other. Establish under whose authority the screening runs, who holds the approval, and whether the vendor acts as the institution's agent or on its own account.

Converted from Not Rated. No SOC 2, ISO 27001, HITRUST or equivalent attestation was located and there is no trust centre. Two searches were run; one failed to reach the company at all, returning unrelated vendors' trust pages and generic framework material, which this index treats as weak evidence rather than as confirmation.

What the company does publish is architecture rather than assurance, and it is substantive enough to change how the gap should be read. Its own product documentation states the platform is cloud agnostic and can be hosted in the customer's own environment so that data never leaves, with vendor hosted options offered as an alternative. It is available natively on a major data cloud, meaning it can run inside a customer's existing warehouse tenancy, and it is listed on a major cloud marketplace under a first party seller profile.

That produces two different buyers with two different exposures, and the missing attestation does not affect them equally. A customer running the platform inside its own environment has substituted architecture for assurance: nothing transits, so the question narrows to what the software does inside the boundary, what it transmits outbound, what telemetry leaves, and what the vendor can see during support. A customer taking the vendor hosted option has none of that and no attestation either, which is the weaker position by some distance. Establish which configuration is being quoted before assessing this row.

One discrepancy to resolve directly. The company's own product page states that trial environments use synthetic data. Earlier trade coverage of the same feature described it as using a real world clinical dataset. A demonstration environment is a recognised place for real patient data to end up inside a sales process, so ask which is accurate today and what governs the environment either way.

Converted from Not Rated. This is a scoping determination rather than an absence finding, and the distinction is the whole of the change.

The prior note's reasoning was correct and only its conclusion moves. It observed that applying a clearance standard to a product outside device regulation would misrepresent both, which is right, and is precisely why this axis grades a scoping determination on whether the question is answered rather than on whether a clearance exists. Here the question is answered.

The product abstracts and reasons over unstructured records for chart review, cohort building and trial prescreening. It does not diagnose, does not direct treatment, and does not make a recommendation about the care of an identified patient. Candidates it surfaces are confirmed by a research team before anyone is approached. That places it outside device regulation, and no pathway is claimed.

The evidence here is stronger than a vendor assertion, which is why this sits at B rather than lower. The platform has itself been the subject of registered prospective evaluation, and those registration records state that the intervention is neither an FDA regulated drug nor an FDA regulated device. A registration record is completed by the study sponsor and held in a public registry, so it is documentary confirmation of regulatory status from outside the company rather than a marketing claim about itself. Very few records in this index can point to anything comparable on this axis.

What does govern this product is worth naming so the grade is not misread as an absence of oversight. Prescreening patients against trial criteria happens before anyone has consented to anything, so the operative instruments are institutional review board approval and an authorisation or waiver covering the screening itself, not a device clearance. Establish under whose authority the screening runs and who holds that approval.

Notable for treating diversity as a measured study endpoint rather than a marketing claim. The Penn prospective study states a hypothesis that the human plus AI workflow improves the efficiency, accuracy, and diversity of trial prescreening, which puts equity of enrollment under formal evaluation. Given that standard prescreening is known to introduce demographic bias, testing whether AI augmentation reduces or amplifies it is the right question. Results were not available at review, and no separate governance framework or subgroup performance disclosure was located.

This vendor publishes the reason for its architecture choice and builds so that its failure mode is visible, which is a materially higher standard than describing capabilities. The design pairs language modelling with symbolic reasoning over a clinical hypergraph, and the company states why: it argues that purely neural approaches cannot discern clinical nuance and are susceptible to hallucination.

A stated rationale is contestable in a way a capability list is not, because a reader can disagree with the premise and evaluate whether the design follows from it. The second half matters more. Every answer is tied to discrete highlighted evidence in the original record, which the company frames explicitly as making the algorithm's failure mode immediately visible rather than obscured.

That is designing for inspectable failure rather than for the appearance of correctness: an unsupported assertion shows as unsupported at the moment a clinician reads it, rather than surviving until an audit. Most products in this index are built so that a wrong answer looks exactly like a right one. Held below the top grade because nothing is measured or promised.

No hallucination rate, no formal failure mode documentation, no accuracy figure and no warranty, indemnity or remediation commitment was located, so the mechanism is present and its effectiveness is unquantified. Ask for the measured rate at which answers carry insufficient or wrong evidence, and what a reviewer is expected to do with a weakly evidenced answer.

Integration is oriented toward the analytics stack rather than the clinical one, and is named specifically: upstream and downstream connectivity with data warehouses including Databricks and Snowflake, and business intelligence tools including Tableau and Qlik. The platform ingests structured and unstructured sources spanning notes, pathology, genomics, claims, and electronic data capture.

Named connectors are a stronger disclosure than asserted interoperability, though direct EHR integration is not enumerated, which fits a product operating on warehoused clinical data rather than at the point of care.

The clearest deployment answer among the trial vendors added here, and the strongest residency position. The platform is explicitly cloud agnostic and can be hosted in the customer's own environment so that data never leaves, with vendor hosting available as an alternative rather than a requirement. For an institution unwilling to move patient records to a vendor cloud, that removes the primary objection outright. Try Me environments with synthetic data are offered for evaluation, which lets a buyer test before exposing real records.

No pricing is published. The company does make a comparative cost argument, that domain specific models can outperform generic large language models at a scalable cost, but no rates, licensing structure, or per record economics are disclosed. Evaluation access through synthetic data environments lowers the cost of assessing fit before committing, which is a partial offset.

Broader than trial matching alone. Distinct product configurations address chart review, cohort building and analysis, and trial eligibility mapping against templates that can incorporate prior authorization criteria and NCCN guidelines, serving clinical care, pharmaceutical, and diagnostics customers. Published evaluation work is concentrated in oncology, which is where unstructured record complexity is greatest, but the underlying capability is not disease specific.