CIFoundry

How to evaluate medical AI

Use these checklists to evaluate a medical AI tool before your committee signs off. Each guide splits its checklist by what the tool actually does, so an ambient scribe used in cardiology is judged like a scribe, not like an echo algorithm.

Start here

How to evaluate a medical AI tool

The 12 questions that apply to every clinical AI tool: intended use, evidence, FDA status, local validation, workflow, monitoring, data rights, liability and contracts.

Every tool-type checklist covers

  • What physicians look for
  • What administrators look for
  • Evidence to ask for
  • How to test it locally
  • Red flags
  • Questions for the vendor

Checklists by area

Pick the area, then check the line underneath matches what the tool does.

Used across every specialty

Tools any department might buy: ambient scribes, chart summaries and EHR assistants.

Specialty and care-setting clinical tools

Tools that read images, signals or slides, or support decisions in one specialty or care setting, such as the hospital ward and ICU.

Business, access and operations

Tools for revenue cycle, patient access and hospital capacity.

Rules & regulations

Plain-language explainers of the rules buyers ask about most. General information, not advice about a specific tool or patient.

Sources we use

The official databases and standards we check listings against. Use them to confirm a tool's regulatory status, evidence and claims yourself.

Regulatory & safety

Official records for checking the regulatory status and safety of medical AI.

Standards & assurance

Frameworks used to evaluate and govern clinical AI.

Evidence & reporting guidelines

Peer-reviewed reporting standards for AI in medicine.

Literature & databases

Where to find the primary research behind each tool.

Educational information, not medical, legal or regulatory advice. No vendor reviewed, sponsored or edited these guides.