FDA Clears Neurora AI, but Its Evidence Is Unpublished

NeuroraWhat to know

30-second read

For Pediatrics · Psychiatry

Who

Children 18 months to 12 years already undergoing a clinician-administered autism spectrum disorder (ASD) evaluation.

What it detects

AI scores 100+ vocal, facial and gaze biometrics from the recorded exam, benchmarked against children with and without ASD.

Evidence

None published. Argus Cognitive has disclosed no sensitivity, specificity or study-size figure since the 8 Oct 2026 clearance.

Where it fits

After the behavioral assessment is recorded; a clinician uploads the file to Neurora’s portal for a score alongside their own judgment.

Key consideration

Cleared by 510(k) substantial equivalence, not an efficacy trial; both prior cleared rivals published pivotal-study data at their own clearance.

Availability

Cleared 8 October 2026; no price, billing code or launch timeline disclosed.

Practice impactNo change to practice yet

Until Argus Cognitive publishes validation data, a Neurora score is information of unknown reliability — document it as AI-assisted, not as evidence, and weigh it far more lightly than a competitor's published figures.

Read the full physician analysis

Full analysis · 8 min readArgus Cognitive press release, 8 Oct 2026 · Cognoa and EarliTec FDA clearance records

Argus Cognitive has won FDA 510(k) clearance for Neurora, AI software that scores more than 100 vocal, facial and gaze biometrics from a recorded behavioral assessment to support an autism spectrum disorder evaluation in children 18 months to 12 years old. The Hanover, New Hampshire company announced the 8 October 2026 clearance as a second, quantitative opinion for clinicians, not a replacement for one — the same framing every digital autism-scoring tool uses. What Argus Cognitive has not done, unlike the two FDA-cleared tools Neurora now competes against, is publish a single figure showing how often that opinion agrees with an actual clinician's diagnosis.

A hundred biometrics, scored after the exam is already done

Argus Cognitive did not build a new exam; it built a scoring layer for one clinicians already run. A clinician records video and audio during a standard behavioral assessment, uploads it to Neurora's portal, and the software measures speech pacing, the direction and duration of a child's gaze, facial expression dynamics and more than a hundred other cues, comparing the pattern against children with and without an ASD diagnosis. That design choice is Neurora's real point of difference: it adds no hardware and no new session to a visit that is already scheduled, where Cognoa's Canvas Dx asks a parent to complete a questionnaire and submit a short home video, and EarliTec's EarliPoint runs its own structured eye-tracking session with dedicated equipment in the office.

Two rivals published their numbers. Neurora has not.

Cognoa's Canvas Dx won FDA De Novo authorization in 2021 on a pivotal study of 425 patients aged 18 months to 5 years across 14 clinical sites: its output matched an expert diagnostic panel on 81% of children who the panel called ASD-positive and 98% of children the panel called ASD-negative. EarliTec's EarliPoint, cleared in 2022 for children 16 to 30 months, rested on published eye-tracking validation studies that collectively covered more than 500 children. Argus Cognitive has disclosed no equivalent number for Neurora since its own clearance — no sensitivity, no specificity, no study population, no comparator. This article also could not independently confirm Neurora's cited 510(k) number against FDA's public database, which can lag a fresh clearance by days; that gap is flagged here, not resolved.

Three cleared autism-scoring aids, by the evidence published at clearance

425 pts · 81% / 98%Cognoa Canvas Dx (De Novo, 2021): pivotal study across 14 sites, matched an expert panel's ASD-positive and ASD-negative calls

500+ childrenEarliTec EarliPoint (510(k), 2022): published eye-tracking validation studies across multiple sites

Not disclosedArgus Cognitive Neurora (510(k), 2026): no sensitivity, specificity or study size published since clearance

Figures as reported in each company's own FDA clearance summary or contemporaneous trade coverage at the time of that clearance; see refs below.

Capacity, not confidence, is still the bottleneck

Autism evaluation already strains clinical capacity: long waits for a developmental-pediatrics or child-psychiatry appointment are a well-documented bottleneck to early diagnosis in the United States, and every digital scoring aid is implicitly built to help against that wait. Neurora does not shorten the wait for the underlying exam — a clinician still has to administer and record it. What a published validation study would add is not speed but confidence: a number a clinician can set against their own uncertainty on a borderline case. Without one, Neurora's score adds information, but not yet confidence.

Three FDA clearances for autism-evaluation AI

	Neurora (Argus Cognitive)	Canvas Dx (Cognoa)	EarliPoint (EarliTec)
Clearance	510(k), 8 Oct 2026	De Novo, June 2021	510(k), June 2022
Age range	18 months–12 years	18 months–5 years	16–30 months at original clearance
Input	Clinician-recorded video/audio of a standard behavioral exam, scored after the fact	Parent/caregiver questionnaire plus a short home video, scored against a clinician checklist	In-office eye-tracking during a structured video-viewing session
Published accuracy at clearance	None disclosed	81% / 98% match to an expert panel (425 patients, 14 sites)	Validation studies across 500+ children
Standalone use	Not intended to stand alone	Not intended to stand alone	Not intended to stand alone

Neurora and Canvas Dx figures from each company's own clearance announcement; EarliPoint from published coverage of its validation studies. Prescribe and diagnose from the current FDA clearance letters, not this table.

What to do with a second opinion you can't grade yet

A pediatrician or child psychiatrist already using Canvas Dx or EarliPoint has a published number to anchor how much weight its output deserves. For Neurora, until Argus Cognitive publishes a validation study, the practical approach is to treat its score as an unweighted data point: note it in the chart as AI-assisted scoring, not as corroborating evidence, and do not let it tip a borderline clinical call. Nothing here says Neurora's underlying method is unsound — only that neither this article nor any public source can yet say how sound it is.

What's next

Neurora is cleared but not yet priced, coded or scheduled for a wider launch; Argus Cognitive has not said when, or whether, it will publish validation data. How clinics weigh a third biometric-scoring entrant against two with published numbers — and whether Argus Cognitive closes that gap — is the open question this article will revisit.

Frequently asked questions

Does Neurora replace a clinician's autism evaluation?

No. Argus Cognitive and FDA's clearance both frame Neurora's output as a quantitative second opinion, meant to sit alongside a clinician's own findings and judgment — the same framing Cognoa and EarliTec use for their own cleared tools. A clinician still has to administer and record the underlying behavioral assessment.

Why does unpublished accuracy data matter if Neurora is only a second opinion?

Because a clinician cannot weigh an opinion of unknown reliability against their own clinical judgment. Cognoa disclosed that Canvas Dx matched an expert panel's calls 81% of the time on ASD-positive children and 98% of the time on ASD-negative children, across 425 patients; EarliTec's EarliPoint rested on published eye-tracking studies across more than 500 children. Argus Cognitive has disclosed no equivalent number for Neurora.

Is Neurora covered by insurance?

Unknown. Argus Cognitive has not announced a price, a billing or CPT code, or any payer coverage decision since the 8 October 2026 clearance.