A clinical intelligence layer for the mental status exam
Digital MSE™ (dMSE) is Attunio's clinical intelligence layer for the mental status exam, transforming movement and speech-timing signals — never facial emotion — into structured, clinician-reviewable observations during telehealth visits. The clinician confirms what enters the record. Try confirming and dismissing below.
Motion & timing only
No facial-emotion or expression analysis. Movement energy and speech timing, nothing more.
Nothing is stored
Derived numbers only — no raw video or audio frames persist. Opt-in consent, separately given.
Clinician decides
Decision support, not diagnosis. Nothing enters the note until the clinician confirms it.
Digital MSE™
Clinical intelligence · confirm to include in note
- Psychomotor↑ vs. baseline
Mild psychomotor agitation
3 fidget episodes observed
- Speech latency↑ vs. baseline
Delayed response latency
Avg 2.4s response delay
- Speech rate↓ vs. baseline
Reduced speech rate
~95 words per minute
- Attention↓ vs. baseline
Reduced sustained attention
On-screen gaze 68% of visit
Model card
What Digital MSE™ measures, and what it deliberately does not
What it measures
- Movement energy and fidget episodes (motion magnitude)
- Response latency, speech rate, and pause structure (audio timing)
- On-screen gaze ratio as an attention proxy
What it does not claim
- No facial-emotion or affect recognition
- No diagnosis, risk score, or autonomous assessment
- Not a substitute for clinical judgment
Known limitations
- Network lag can mimic response latency
- Off-camera motion can skew movement signals
- Requires an established personal baseline to be meaningful
Bias & validation posture
- Population is neurodivergent by definition — signals read against personal baseline, not population norms
- Every observation carries a confidence level and provenance
- Confirm/dismiss actions are audit-logged
Illustrative concept only. Inferring clinical state from video may constitute regulated software and requires validation for accuracy and bias — especially in neurodivergent populations. Signals shown here are mock data to explore the clinician experience, not a clinical claim.