Healthcare Intelligence
AI that helps hospital staff — without diagnosing in place of doctors. A concept TTNC is designing with hospitals — a screen prototype for discussion, not a system in use. All data is fictional.
“For blood sugar and lipid tests, patients should take nothing but water for 8–12 hours.”
- Docs searched
- 2
- tokens
- 126 / 88
- log
- #5a1c3e
AI answers the general questions. Symptoms go to people.
Every patient message is sorted by type. General questions are answered from the hospital's service guide; symptoms, medicines and personal data go to the right staff member. AI does not diagnose, recommend treatment or change medication.
Demo dataAll fictional — people, documents and figures are not real.AI does not diagnose, recommend treatment or change medication — that work belongs to doctors, nurses and pharmacists. Every message and hand-off is logged for review.
What HCi is being designed to do
General questions answered around the clock; questions about symptoms or medicines go to a nurse or pharmacist; emergency words bring up 1669 at once.
OPD nurses
Requests from chat or web go into a queue, reminders go out, and reminder calls are ordered. Staff confirm every booking; the score only orders calls and never removes a patient's right to care.
Appointments desk
Alerts for low and expiring stock, usage forecasts and draft purchase requests. Pharmacists decide; AI never orders by itself.
Pharmacists
Draft summaries that show where every sentence came from, hide data by role and make allergies stand out. A doctor reviews every summary.
Doctors
Answers from the hospital's own procedures and manuals, with sources. If the documents don't say, it says so — no web search.
All staff
Health data is sensitive data
Health data is sensitive personal data under Thai personal-data protection law, so we recommend Private AI (Full), with every layer in systems the hospital controls. Links to the HIS, pharmacy stock and chat channels are designed with each hospital.
- Patient data stays in
- Your organisation
- What crosses the line
- Every layer installed on site — no need to send data out for processing
- Processed on
- GPUs in your organisation
- You can start with
- A 32–192 GB GPU server
