Medical and Health AI Tools

Medical and health AI tools range from regulated devices and clinician-facing decision support to research, documentation, administration, and general wellness software. A category listing is not evidence that a product is clinically validated, authorized for a particular use, or appropriate for a patient.

Identify the intended use first

• Clinical diagnosis, triage, prognosis, or treatment: verify the exact intended use, jurisdiction, regulatory status, labeled users, contraindications, and required professional oversight.
• Medical imaging and measurement: review the authorized device record, validation population, acquisition requirements, false-positive and false-negative performance, and failure handling.
• Research and evidence review: trace every claim to the original paper or database and check retractions, dates, study design, population, and statistical limitations.
• Documentation and administration: measure correction time, coding or transcription errors, access controls, audit logs, retention, and integration failure modes.
• Consumer wellness: distinguish general information from medical advice and check whether escalation to a qualified professional is clear.

Safety and privacy checklist

Check independent evidence for the exact product version, performance across relevant demographic groups, human override, monitoring after updates, cybersecurity, incident reporting, protected-health-data terms, deletion, and data location. Do not upload identifiable health information during a casual trial.

The FDA's AI-enabled medical-device list links to authorization records and safety information, but the FDA says the list is not comprehensive. WHO recommends protecting autonomy, safety, transparency, accountability, inclusion, equity, and sustainability. For symptoms, diagnosis, medication, treatment, or emergencies, use a qualified health professional and local emergency services.

FDA AI-enabled medical devices · WHO ethics and governance guidance