Table 1 MSK Model Information Sheet questions (2024 version)

Category	Requested Information
General	MSK Business Sponsor
	MSK Business Lead
	Strategic Alignment Category
	Impacted Users
	Primary Function/Objective
	Expected Benefits (Use/Impact metrics)
	Development Team
	Development Staff Members
	Link(s) to model documentation
	Key words (tags for search within Registry)
	Is this an externally acquired solution (vendor)?
Technical	System of Use/Record (i.e., where users will experience it)
	Data Sources Used
	Does it have PHI?
	Does it have PII?
	Is it subject to GDPR?
	Model Training Methods/Model Type
	Model Performance Metrics
	Development Platform
	Github Link to code
	IRB link (if applicable)
	Fairness Assessment Method
FDA SaMD Screening	Info Type: Communicable in normal clinical conversation?
	Output: Gives clinician directive or recommendation?
	Disclosure to Clinician: Gives basis for recommendation? Dynamic or “learning” algorithm?
	FDA SaMD Determination
	Other Regulatory Determination TBD (e.g., State regs, ONC, NIST, etc.)
Approval Status	Silent Pilot Approval
	Full Production Launch Approval
	Anticipatable AI-induced Adverse Events (aiAE’s)
	How often is performance re-evaluation required?
	Last post-launch date/status of review of relevant research and evidence

IRB institutional review board, PHI personal health information, PII personal identifying information, GDPR general data protection regulation, FDA SaMD Federal Drug Administration Software as a Medical Device, ONC Office of the National Coordinator, NIST National Institute of Standards and Technology, aiAE’s AI-induced adverse events.
Source: Memorial Sloan Kettering.
Required questions to complete a model’s registration by AI developers (both built and acquired models) are shown below, and are completed in collaboration with the AI Governance Committee staff.

Quick links

Search