121 Study design considerations for post-deployment monitoring of AI/ML models
Momin Malik, Dave Watson, Madison J Beenken, Shauna M Overgaard, Hanyin Wang, Imad Absah, Chung Il Wi, Young J Juhn · 2026
Read 121 Study design considerations for post-deployment monitoring of AI/ML models by Momin Malik, Dave Watson, Madison J Beenken, Shauna M Overgaard, Hanyin Wang, Imad Absah, Chung Il Wi, Young J Juhn (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13173101.txt.
Description
Momin Malik and seven colleagues set out guidance for checking AI models after they enter real clinical use. A model that tests well in the lab can drift or act unfairly once deployed. Yet little guidance exists on how to design the checks. The authors link machine learning practice to older ideas from medical testing.
They show that the sampling style used to build a model cannot give valid measures of how often its alarms are right. Under real use, some key measures can only be found if a random share of negative calls gets checked too. Fairness measures face the same trap. Their guidance aims to set a clear, strict standard for watching models in practice. That should help hospitals trust what these tools report.
Catalog record
| CREATOR | Momin Malik, Dave Watson, Madison J Beenken, Shauna M Overgaard, Hanyin Wang, Imad Absah, Chung Il Wi, Young J Juhn |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 2 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |