Neurology · Autism spectrum disorder diagnosis aid
EarliPoint Assessment
EarliTec Diagnostics, Inc.
EarliPoint is the eye-tracking autism diagnostic aid validated in two double-blind prospective studies published side by side in JAMA and JAMA Network Open in September 2023. The honest headline sits in the primary paper: against expert clinical diagnosis in all comers, sensitivity was 71.0 per cent and specificity 80.7 per cent, rising to 78.0 and 85.4 per cent when the expert diagnosis was certain. A second pivotal study in children aged 31 to 95 months (sensitivity 77.1 per cent, specificity 78.6 per cent) underpins the March 2026 clearance that widened the indication; the footprint remains US-only across a four-clearance FDA lineage.
Performance Metrics
Clinical Evidence
The evidence base is unusual for this category: two large double-blind prospective diagnostic-accuracy studies published simultaneously (September 2023) in JAMA and JAMA Network Open, both manufacturer-affiliated — Warren Jones, the senior investigator, is a co-founder of EarliTec, and the device descends from the Marcus Autism Center and Emory University research programme. The JAMA study (Jones, Klaiman, Richardson et al., JAMA 2023;330(9):854-865; NCT03469986) enrolled 499 children aged 16 to 30 months at six US specialty centres from April 2018 through May 2019; measurement succeeded in 475 (95.2 per cent, mean age 24.1 months). Against best-practice expert clinical diagnosis made blind to the index test, eye-tracking measurement of social visual engagement showed [[sensitivity]] of 71.0 per cent (95 per cent CI 64.7 to 76.6) and [[specificity]] of 80.7 per cent (75.4 to 85.1). In the 335 children whose expert diagnosis was made with certainty, sensitivity rose to 78.0 per cent (70.7 to 83.9) and specificity to 85.4 per cent (79.5 to 89.8). Test results also correlated with expert assessments of individual social disability (r = -0.75), verbal ability (r = 0.65) and nonverbal cognitive ability (r = 0.65) — the basis for the device's assessment indices, which reach beyond a binary call. The companion paper (JAMA Network Open 2023;6(9):e2330145) reported the discovery and replication studies behind the classifier: 1,089 children aged 16 to 30 months in two prospective double-blind cohorts (719 discovery, 370 replication), with AUC 0.90 and 0.89, sensitivity 81.9 and 80.6 per cent, and specificity 89.9 and 82.3 per cent respectively. The scientific lineage runs to Jones and Klin's 2013 Nature paper showing that attention to others' eyes declines from two to six months of age in infants later diagnosed with autism — foundational science, not device validation. The K253442 510(k) summary (March 2026) reports a second prospective, double-blind, multi-centre, within-subject pivotal study in children aged 31 to under 96 months, run against the same expert-clinician reference standard. With the new borderline range (diagnostic index scores above 0 and up to 3) reported as a separate output, sensitivity was 77.1 per cent (155 of 201; 95 per cent CI 70.7 to 82.7) and specificity 78.6 per cent (132 of 168; 71.6 to 84.5) among 369 certain diagnoses; counting borderline results as negative, sensitivity fell to 63.5 per cent (155 of 244) with specificity 84.5 per cent (197 of 233) in 477 children. Pooled across both studies (16 to under 96 months, n=676 with borderline separate), sensitivity was 80.2 per cent and specificity 81.3 per cent. Uncertain reference diagnoses (112 in the older cohort) were excluded from the performance analysis. This study is reported in the regulatory summary rather than a peer-reviewed publication as of this review. No randomised controlled trial of the device has been published, and independent (non-manufacturer-affiliated) validation at scale has not yet appeared.
| Study | Design | n | Sensitivity | Specificity | AUC | Published |
|---|---|---|---|---|---|---|
| Jones W, Klaiman C, Richardson S, et al. (Marcus Autism Center / Emory; manufacturer-affiliated; NCT03469986) | ProspectiveProspective | 475 | 71.0% all comers / 78.0% certain-diagnosis subgroup | 80.7% all comers / 85.4% certain-diagnosis subgroup | — | JAMA 2023;330(9):854-865; 499 enrolled, 475 measured (95.2%), 6 US specialty centres, double-blind |
| Jones W, Klaiman C, Richardson S, et al. (discovery cohort; manufacturer-affiliated) | ProspectiveProspective | 719 | 81.9% | 89.9% | 0.90 | JAMA Netw Open 2023;6(9):e2330145; double-blind discovery study |
| Jones W, Klaiman C, Richardson S, et al. (replication cohort; manufacturer-affiliated) | ProspectiveProspective | 370 | 80.6% | 82.3% | 0.89 | JAMA Netw Open 2023;6(9):e2330145; double-blind replication study |
| EarliTec Diagnostics (second pivotal study, ages 31 to <96 months; manufacturer-sponsored) | ProspectiveProspective | 369 | 77.1% (155/201) borderline as separate output / 63.5% (155/244) borderline counted negative | 78.6% (132/168) borderline as separate output / 84.5% (197/233) borderline counted negative | — | FDA 510(k) summary K253442 (March 2026), https://www.accessdata.fda.gov/cdrh_docs/pdf25/K253442.pdf; prospective, double-blind, multi-centre, within-subject comparison vs expert clinical diagnosis; 369 certain diagnoses with borderline separate (477 with borderline as negative), 112 uncertain diagnoses excluded; combined 16 to <96 months: 80.2% sens / 81.3% spec (n=676). Not peer-reviewed |
Inside the algorithm
Editorial featureHow EarliPoint reaches its indices.
Five stages — from raw input to verdict — drawn from manufacturer documentation and the public regulatory record.
- INGEST
- NORMALISE
- DETECT
- LOCALISE
- VERDICT
Stage 01 · INGEST
The child watches videos of social interaction while gaze is recorded.
The child is seated in front of a portable tablet-based device showing short video scenes of everyday childhood social interaction — children playing, caregivers talking. An integrated eye tracker samples where the child is looking, moment by moment, for the duration of each scene. There is nothing invasive and no task to perform; the data is the child's natural viewing behaviour.
Input
Eye-tracking samples over social video scenes
Inference
Per assessment session
Inside the Auris+ Listing
Five more sections complete this device’s Auris+ Listing.
Decision Ledger
ProPro unlocks a private, cross-vendor log of every case you read on this device — what the AI called, what you concluded, and a one-line reason if you overrode it. Ready for the EU AI Act's deployer logging obligations when they land in 2028.
Clinical Evidence Deep Dive
ProPro unlocks the structured clinical-evidence summary — study count, target patient population, and a tabular accuracy-metrics view drawn from peer-reviewed sources.
Peer-Reviewed Publications
ProPro unlocks the curated peer-reviewed publication list with PubMed cross-links — the citation backbone of every editorial verdict.
Post-Market & Regulatory Conditions
ProPro unlocks the post-market surveillance summary, recall record, and the conditions of approval that bound real-world use.
AI Algorithm Version History
ProPro unlocks the chronological record of algorithm version changes — what changed when, drawn from manufacturer changelogs and regulatory filings.
Regulatory Approvals
K253442
Class II (21 CFR 882.1491, product code QPF)
Safety Record
No recall, safety communication or adverse-event report specific to the EarliPoint family was found in public reporting as of August 2026. FDA MAUDE and the enforcement database could not be queried directly from this environment (register access blocked), so this rests on search-surfaced sources rather than a direct register read; the 2022 510(k) review reported no serious adverse events in the supporting studies. The device is non-invasive — the child watches videos while gaze is recorded. The clinically relevant risk is misclassification: in all-comers the pivotal sensitivity of 71.0 per cent means roughly three in ten autistic children were not flagged by the index test alone, which is why the indication frames the device strictly as an aid to a qualified clinician's diagnosis, not a standalone screen, within a population already referred for concerns.
Intended Use & Indications
The EarliPoint Assessment is indicated as a tool to aid qualified clinicians in the diagnosis and assessment of autism spectrum disorder in children aged 16 through 95 months who are at risk based on concerns identified by a parent, caregiver, or healthcare provider. During the assessment the child watches short video scenes of social interaction on a portable tablet-based device while an eye tracker records their gaze. The measurements quantify the child's social visual engagement — attention to people, faces and socially salient moments — against age-expected reference data, and the resulting indices support, rather than replace, the clinician's diagnostic judgment. The original 2022 clearance restricted use to specialised developmental-disabilities centres and to children aged 16 through 30 months; subsequent clearances moved the device onto a portable tablet (2023), dropped the specialised-centre restriction and framed the population as children at risk based on concerns identified by a parent, caregiver or healthcare provider (2025), and extended the age indication from 31 through 95 months with a new borderline output (2026). Results are interpreted by a qualified clinician.