Side A
For clinicians
Subtype review with the confidence surface exposed — see what the classifier decided and why. In build.
A four-subtype ADHD classifier — with an indeterminate path.
A deterministic four-subtype classification engine using eye-tracking and cognitive features. Confidence-gated with an explicit indeterminate output — no false certainty.
Subtype classifiers in psychiatry commonly overreach. Trained on ambiguous ground truth and deployed against patients whose signal is weaker than expected, they output a class every time — even when confidence is low. That confidence gap doesn't disappear just because a system refuses to show it.
Facet uses four subtypes plus an explicit indeterminate path. When the eye-tracking and cognitive signal isn't strong enough to pick a subtype confidently, the output is "indeterminate — collect more data," with the reason shown to the clinician. That's where the model's uncertainty belongs — and it's what makes the other three answers trustworthy.
Deterministic classification into one of four subtypes — or an explicit indeterminate output when confidence is below threshold.
Gaze dynamics as the primary signal — a modality with more discriminating power than self-report at the subtype level.
Complementary features from a validated cognitive battery — used to disambiguate cases the eye-tracking alone can't resolve.
Every decision has a receipt — the inputs, the model version, and the confidence. Built with Software-as-a-Medical-Device review in mind, though that classification itself is still pending.
Facet's classifier is implemented and tested against synthetic data; validation on a real clinical cohort — and the clinical UI — are both in build.
Subtype review with the confidence surface exposed — see what the classifier decided and why. In build.
Batch classification and audit — run the classifier over a study population and audit every decision. In build.
Facet is one of the research tools built inside the Synapse Spark Foundation. Everything we make is part of the same mission — understanding neurodivergent minds and supporting the researchers advancing the science.
← Back to the Foundation