1. Have you successfully trained a model and demonstrated its feasibility on a retrospective dataset (e.g., a curated public dataset)?
2. Has the model been validated on a completely independent, held-out test set from your local institution using the TRIPOD+AI checklist?
3. Have you conducted and published a formal Demographic Bias Audit to ensure equitable error rates across protected groups?
4. Has the model been externally validated on data from a completely independent clinical site or distinct geographic population (using STARD-AI)?
5. Has the model been deployed into a live clinical IT environment and evaluated by clinicians in "Shadow Mode" without influencing care?
6. Has the model passed a pre-registered prospective clinical trial (e.g., RCT) where it actively influenced patient care (using CONSORT-AI)?
7. Has the model obtained formal regulatory clearance (e.g., FDA 510(k), CE Mark) and is currently deployed in commercial/institutional practice?
8. Does the model demonstrate sustained, multi-site health economic benefits and long-term outcome improvements as a standard of care?
Your Maturity Level
CRL 1
Basic Principles Observed
Next Steps
To reach CRL 2, you must demonstrate algorithmic feasibility on a retrospective dataset and create a Model Card.