FlagshipHealthcare AISafety review required

MedAI

An owner-confirmed priority in healthcare AI, presented within a safety-first frame. Nothing here implies diagnosis, treatment, validation, or approval.

Public source snapshot

medaivn.com

Source-backed context

What the official site says

The official site presents MedAI as a Vietnamese traditional-medicine AI assistant for herb lookup, folk remedies, interactions, references, and safety warnings.

Its About page lists 5,190+ herb records, symptom analysis with suggested herbs, drug-interaction checks, source references, and medicinal-plant image recognition.
The project pages identify HSB, Vietnam National University, Hanoi, and a 2024-present research phase.

Publication boundary: The official terms say MedAI is not a doctor or medical facility and must not be used for diagnosis, prescribing, or treatment. This page makes no clinical, outcome, or regulatory claim.

Read the official source

◐ Draft structure — owner input required

This page is a verified scaffold, not final copy. Its sections follow the required questionnaire in docs/07_MEDAI_REQUIREMENTS.md and will be filled only with owner-approved facts. No claims, metrics, or statuses are asserted here yet.

01Intended use

Awaiting owner inputThe care setting, country, users, and whether the system is for information, administration, decision support, triage, or another owner-approved purpose.

02Explicit exclusions

Awaiting owner inputNon-intended uses and the boundaries the system will not cross. The safety frame is defined as much by refusal as capability.

03Human oversight & escalation

Sketch only — not a process claimWho reviews what, how urgent situations transfer to people, and how clinical responsibility remains with qualified professionals.

04Data, privacy & evidence

Awaiting owner inputData sources, controller, consent, retention, deletion, access controls, provider boundaries, and evaluation evidence — including a clear statement if none exists.

05Regulation & limitations

Awaiting owner inputClassification, regulatory status, bias and accessibility testing, known limitations, and the safety-incident process.