MOVO-X
Loading MOVO-X…
A clinic in Kuala Lumpur serves patients who speak Malay, Mandarin, Tamil, English, and Bengali in the same morning session. A multilingual check-in kiosk means every patient completes their own intake in their own language — without a staff interpreter.
Language barriers at check-in cause two clinical problems: patients under-report symptoms because they cannot express them accurately in a second language, and staff over-simplify intake to compensate — creating incomplete clinical records. Multilingual capture solves both.
Patient selects their language at the first screen. All form fields, prompts, error messages, and guidance render in that language from selection onward. Language preference is stored for return visits.
Chief complaint NLP runs in the patient's chosen language. On-device models for Malay, English, Mandarin (Simplified), Tamil, Arabic, Thai, Indonesian, and more. ICD-10 output is always in standard English regardless of input language.
Voice prompts in 12+ languages using on-device neural TTS — not pre-recorded clips. Natural speech rhythm, locally-appropriate phrasing, and real-time generation for dynamic content like wait times.
Arabic and Persian UI renders right-to-left correctly. Input validation, error messages, and the keyboard layout all adapt to the script direction. Thai and Tamil scripts render accurately without font-swap artefacts.
The TV queue display renders patient names in their native script — Chinese characters, Arabic, Tamil, Devanagari. Patients recognise their own name on the board without Romanisation.
Queue position, doctor ready, and appointment confirmation WhatsApp messages are sent in the patient's language of choice captured at check-in.
Zero language-barrier-related incomplete intakes after multilingual deployment
Patient self-reported symptom completeness improves in non-English-speaking cohorts
Staff interpreter requests drop by 70%
Patient satisfaction among non-English-speaking demographics increases significantly
Multilingual support uses a combination of on-device NLP models per language (for NLP intake), a localisation framework (for UI strings), and neural TTS (for voice). UI strings are stored in JSON locale files with 100% coverage — no language has missing translations. RTL languages use CSS logical properties throughout the UI; no separate RTL stylesheet is maintained. Adding a new language takes 2–3 weeks: NLP model fine-tuning, UI string translation, and voice model validation.
English, Bahasa Malaysia, Mandarin (Simplified), Tamil, Thai, Bahasa Indonesia, Arabic, Spanish, German, Japanese, French, and Portuguese. Korean, Vietnamese, and Hindi are in active development.
On return visits, the kiosk defaults to the patient's stored language preference. New patients see the language selection screen. Auto-detection via accent recognition is available as a beta feature.
Yes. The clinical coding output (ICD-10, SNOMED-CT) is always in English standard codes regardless of what language the patient used for their chief complaint description. The original patient words are preserved as a separate field.
Malay, Mandarin, and Tamil models achieve similar accuracy to English after 30 days of per-clinic calibration. RTL language models are slightly lower accuracy initially but improve with operational data.
Yes. MOH requirements for Bahasa Malaysia as the primary clinical documentation language are met. Patients can complete intake in any language; clinical records are maintained in Bahasa Malaysia and English per MOH guidelines.
Tell us about your operation. We'll send a tailored proposal — architecture, integration scope, deployment timeline, and total investment — within hours.