MOVO-X
Loading MOVO-X…
An AI self check-in kiosk does what a touchscreen form cannot: it identifies the patient by face and ID chip, understands their chief complaint, scores acuity, and routes them to the right doctor — in under 90 seconds, with no staff involvement.
Most "AI check-in kiosks" are touchscreen forms wearing an AI badge. Genuine AI runs inside the kiosk hardware: computer-vision ID verification, on-device NLP symptom capture, acuity classification, and smart priority routing — all executing locally without a round-trip to the cloud. The difference is patient data never leaves the device by default.
Face match against the captured ID document photo. Liveness detection blocks spoofs. The entire pipeline runs on-device — no patient image touches a remote server unless the clinic explicitly opts in.
Patient speaks or types their reason for visit in any of 12+ languages. On-device NLP maps their words to ICD-10/SNOMED-CT clinical codes delivered to the clinician before the patient sits down.
Lightweight on-device classifier scores the intake in real-time and flags high-priority patients. Score plus contributing factors are surfaced to triage staff as decision-support — never overriding clinician judgment.
Per-patient wait time calculated from live queue load, doctor availability, and historical consultation-time data. Sent to the patient via WhatsApp the moment they finish check-in.
Per-kiosk anomaly detection watches patient-flow patterns, hardware health, and ID-verification confidence in real-time. Alerts ops staff before a patient hits a dead-end.
Acuity and NLP models calibrate per clinic on operational data. After 30 days the acuity model stabilises above 90% agreement with triage nurses. Federated learning keeps data in-facility.
90-second end-to-end check-in including ID verification
Acuity agreement with triage nurse above 90% after 30-day calibration
Zero raw biometric data transmitted to cloud by default
ID false-reject rate below 0.5% across 174 national ID formats
MOVO-X AI self check-in runs as a native Android Kotlin application on RK3566 hardware with on-device TensorFlow Lite and ONNX Runtime models. APDU NFC reads MyKad and biometric passports in under 2 seconds. The AI stack is deliberately lean — task-specific models, not a general LLM — so inference completes in under 500 ms on the ARM Cortex-A55. Every model has a documented model card, evaluation harness, and human-override path.
No. Computer vision, NLP, and acuity scoring are discrete production models with documented governance — not wrapper prompts around GPT. Each runs independently in its own inference container on the kiosk hardware.
On-device face match compares the live face to the ID document photo. The match result (pass/fail + confidence score) is stored, not raw images — unless the clinic opts into biometric retention for repeat-patient recognition.
Every decision is logged with confidence score. Below configurable thresholds the kiosk flags for staff-assist intake. Clinical records always capture both the AI signal and any human override.
MyKad (Malaysia), biometric passports, NRIC (Singapore), NIK (Indonesia), Aadhaar (India), Cédula (Colombia, Ecuador), and 170+ other national IDs via MRZ OCR and NFC APDU.
Yes. Per-clinic fine-tuning is available on enterprise tier. Acuity models calibrate automatically within 30 days; NLP models can be tuned for specialty vocabulary on request.
Tell us about your operation. We'll send a tailored proposal — architecture, integration scope, deployment timeline, and total investment — within hours.