How it works
One install.
Continuous intelligence.
Emobot closes the monitoring gap between your clinical visits — the gap where silent churn happens. Below is exactly how, including how we protect patient privacy at every step.
STEP 01
Patient installs the app. Once.
3 minutes. Then nothing.
Your patient downloads EmoDTx on their iPhone and grants a few permissions. That's it — they'll never be prompted to do anything again.
- 3-minute setup. No account, no wearables, no habit change
- Any iPhone on iOS 16 or later
- 80% activate when a clinician proposes it and installs it with the patient during the session
.png&w=640&q=75)
3-min setup
then passive
Monitoring active
STEP 02
AI monitors continuously.
Multimodal. Passive. Private.
Emobot reads four signal streams from natural phone use: facial expression, voice, activity, and digital behavior. Facial analysis runs on-device; voice is held in a 30-minute window then discarded, leaving only the score.
- Facial microexpressions, correlated with depression severity
- Vocal biomarkers: prosody, rhythm, energy
- Actigraphy + digital behavior: movement, step count, screen time
.png&w=640&q=75)
Three pillars, one continuous score: emotional · physical · social.
STEP 03
The patient sees their own trend — and gets nudged when it dips.
A wellness app. Not a surveillance feed.
Inside their own app, patients see a daily wellness score and mood trend. When it starts to deteriorate, the app invites them to rebook with you — 48 hours before clinical signs appear. They rebook in-app, before they quietly drop off.
- The patient owns their data, inside their own app
- In-app nudge to rebook as soon as signals deteriorate
- 80% activation vs. 30–40% for PHQ-9 by email
The patient sees their own trend, and the in-app nudge when it dips.
STEP 04
When the patient opts in, you get a clean trend.
Like a lab result, not the blood sample.
If the patient shares, you open a web dashboard with their anonymized trend and the same 48-hour early signal they saw. No EHR integration. No audio, video, or raw biometrics — you see a trajectory, the patient still owns the data.
- Opt-in share; the clinic never sees raw data
- One anonymized depression index per day, plus breakdown
- Standalone dashboard, no EHR integration
portal.emobothealth.com — opt-in trend, never raw biometrics.
What 12 weeks actually looks like
The relapse PHQ-9 can’t catch in time.
A successful TMS course brings the patient into remission by week 6. Then the relapse: a quiet, abrupt rise back above PHQ-9 10. Emobot catches the trend break the moment it starts — weeks before the next scheduled PHQ-9 visit would have surfaced it.
12-week monitoring window · same patient
Patient enters at PHQ-9 ~18 (severe). TMS works: by W6 the patient is in remission, the PHQ-9 visit confirms it. Then the relapse: at W8 Emobot detects the upward trend break, triggers an in-app rebooking nudge — and the patient is recaptured weeks before the next scheduled PHQ-9 at W12 would have surfaced the deterioration.
Privacy & security
“Is it creepy?” No. Here’s why.
This is the first question every physician asks. We designed Emobot from the ground up to make the answer unambiguously “no.”
Data architecture · minimum footprint by design
Patient's phone
Raw video never leaves the phone.
Emobot cloud
voice in 30-min ephemeral window
scored, then discarded
HIPAA · BAA available
Clinician dashboard
On-device facial analysis
Raw video never leaves the phone. The neural network runs locally and the raw buffer is discarded.
Ephemeral voice processing
Voice audio is held in a 30-minute window, then discarded. Only the numerical emotion score is kept.
Only scores leave storage
The dashboard receives a daily depression index, like a lab result — never audio, video, or raw images.
HIPAA-ready architecture
TLS 1.3 in transit, AES-256 at rest. Patient records separated from EHR by design. BAAs available for US clinics.
Battery-optimized
~7–10% battery impact — less than most navigation apps. Reduced 30% in the past year.
Transparent to patients
Patients see their own trend. Monitoring is fully disclosed and consented.
Technical specifications
r=0.89
MADRS correlation
r=0.83
PHQ-9 correlation
7–10%
Battery impact (iPhone)
0
Raw video / images stored
3 min
Patient onboarding time
4
Signal streams
11
Clinical studies
Daily
Update frequency
See it live in your workflow.
Pick the path that fits, or book a demo and we’ll walk through a real patient case in 30 minutes.
For IP clinics & networks
The full ROI math + clinic deployment story.
Pricing, two-mechanism revenue model, security & compliance, the clinics who already deploy Emobot.
Open the clinic page →For patients
What the patient actually experiences.
The 50-second video, the screens, the safety net loop, and how to ask your clinic to enroll you.
See the patient experience →The live dashboard requires no login — a fast way to explore before you book.