1. Intended use
What exact decision or task is the tool designed to support—and what is outside scope?
Responsible innovation
AI in medicine
AI can analyze images, identify patterns, assist research and support clinical workflows. Safe use depends on intended purpose, representative data, validation, transparency, monitoring and human oversight.
Where AI is used
From pixels to pathways
- Medical imaging acquisition and interpretation
- Risk prediction and early-detection support
- Clinical documentation and workflow automation
- Drug discovery and trial design
- Personalized diagnostics and treatment support
- Remote monitoring and signal detection
Evaluation checklist
Before trusting a health algorithm
2. Population
Was it tested in people and settings similar to where it will be used?
3. Performance
Which metric is reported, against what comparison, and with what uncertainty?
4. Human oversight
Who reviews the output, handles disagreement and communicates limitations?
5. Monitoring
How are errors, data drift, updates and real-world performance tracked?
6. Privacy
What data is collected, retained, shared, sold or used to improve the model?
FDA context: The FDA maintains an AI-enabled medical-device list and evaluates devices for their intended use. Inclusion does not mean an AI system is perfect, appropriate for every patient or free from the need for clinical judgment.
FDA AI-Enabled Devices
Browse authorized AI-enabled medical devices and public decision summaries.
Open FDA list ↗AI and Medical Products
FDA resources covering drugs, biologics, devices and cross-center work.
Open FDA hub ↗ClinicalTrials.gov
Search registered studies and learn whether a claim is backed by completed human research.
Open trials database ↗