A three-level digital screening funnel for ADHD, autism, specific learning disabilities and more — for children aged 8–10, in English, Hindi and Kannada, on any device a school already owns.
Each level raises or lowers the probability using likelihood ratios; only the few who need it reach clinical assessment.
NeuroApex screens and refers; it does not diagnose. Accuracy figures are shown only where published evidence supports them, and our own sensitivity/specificity will be reported only after a validation study.
Every test is mapped to peer-reviewed validation evidence with the PDF one click away.
Constructs, scoring, limits and linked papers.
Open-access PDFs, searchable, with the exact findings each supports.
Norming, test–retest, language fairness, STARD reporting.
RPwD Act 2016, NEP 2020, Samagra Shiksha, Start-up India, BIRAC, CSR and DPDP Act 2023 — with sources.
What each scheme provides, fit score, documents, route.
DPDP verifiable parental consent, ICMR ethics.
0–6, 6–18 and 18–36 months.
Levels 1 and 2 playable. Instant child, parent and teacher reports; a compressed research bundle for future ML.
Installable PWA; phone, tablet, laptop; works offline.
English, Hindi and Kannada in the demo; more are a translation task, not an engineering one.
Granular sensor consent, pseudonymous IDs, on-device processing.
The prototype runs its signal processing on the device. Open Settings in the demo and switch on Developer mode to see it live.
Face distance and head angle from on-device face landmarks, plus noise, light and a baseline reaction time before any test.
Camera, motion, microphone, light, coarse opt-in location and touch. Frames and audio are never stored, only derived features.
Ex-Gaussian reaction-time fits, EZ-diffusion, d′, tremor band power (FFT) and Bayesian scoring with credible intervals.
Live sensor HUD and a data inspector showing the exact features collected, for technical review.
Child, parent and teacher reports with confidence intervals, charts and citations. Printable.
Compressed, checksummed bundle (schema v2) with a feature dictionary and CSV, for later ML.
Norms, priors and likelihood ratios in the prototype are placeholders until a validation study. Hindi and Kannada text awaits native-speaker review.