Cognivue vs Linus Health
The two governance leaders of the category, both with strong independent evidence, at opposite ends of the AI spectrum. Cognivue is the settled foundation: it holds the De Novo that created this regulatory category, its FOCUS study is a model bias investigation, and its scope as an adjunctive tool is drawn precisely, though it runs only on dedicated hardware with no EHR integration. Linus Health is the frontier: genuinely AI native, analysing drawing and speech behaviour to predict amyloid status, with peer reviewed method transparency and the most direct engagement with cultural bias in this sweep. The tension is regulatory certainty against capability, because Linus is cleared for cognitive assessment while its commercial claim has moved to biomarker prediction, a more consequential use much of which sits in conference abstracts. If your failure mode is a defensible, demographically validated screen on a settled clearance, start with Cognivue. If your failure mode is predicting amyloid pathology for treatment eligibility, start with Linus, and track which claims are cleared versus research.
- The regulatory foundation of the category: Cognivue holds the De Novo classification DEN130033, the first FDA cleared computerised cognitive test, graded A on regulatory status, where Linus is cleared for cognitive assessment but its headline claim has moved to amyloid prediction, a different intended use, holding it at B.
- A bias study designed and reported as one: the FOCUS study enrolled 452 participants across six US sites stratified by age, education, sex, race, and ethnicity, matching Linus at the top of the governance axis from empirical validation rather than design intent.
- Settled scope with independent evidence: validated as equally effective as the MoCA with superior test retest reliability, indicated precisely as an adjunctive tool, graded A on autonomy, with the reliability that matters for baselining and monitoring over time.
- Genuinely AI native: the DCTclock analyses the drawing process, stylus dynamics and speech acoustics with no paper equivalent, graded A on AI centrality where Cognivue's adaptive psychophysics grades B, and A on model transparency for peer reviewed method disclosure where Cognivue grades B.
- Biomarker prediction is the capability: DCTclock correlated with amyloid PET in cognitively normal people and the DCR predicting PET status in the 930 participant Bio-Hermes-001 study, addressing treatment eligibility rather than screening alone.
- Deployment without dedicated hardware: iPad administered with a remote option and EHR integrated health system workflows, graded B on both deployment and interoperability where Cognivue's dedicated hardware and one page report grade C on each.
Side-by-Side
| Axis | C Cognivue |
L Linus Health |
|---|---|---|
| AI Centrality | ||
| Autonomy and Oversight Model | ||
| Model and Technology Transparency | ||
| Clinical and Operational Evidence | ||
| AI Safety and PHI Stewardship | ||
| HIPAA and BAA Posture | ||
| Security Certifications and Trust Center | ||
| FDA and Regulatory Status | ||
| AI Governance and Bias Disclosure | ||
| EHR and Interoperability Depth | ||
| Deployment Model and Data Residency | ||
| Commercial Transparency | ||
| Setting and Specialty Coverage |
This is the strongest pairing in the category because both grade A on clinical evidence and both grade A on governance, the only two vendors here to do so, from opposite methods: Cognivue empirically through the stratified FOCUS study, Linus by naming the cultural and educational bias failure mode directly and publishing where its instrument is sensitive to it, including a South Asian cohort. The split is regulatory settledness against AI capability. Cognivue's De Novo is a finished, category creating clearance and its scope is precisely drawn; Linus is the more advanced model but its cleared indication is cognitive assessment while its commercial claim is amyloid and p-tau217 prediction, much of it in conference abstracts rather than full papers. Cognivue's weakness is deployment, dedicated hardware graded C with the weakest interoperability in the category. Neither publishes a security attestation or a BAA.