Performance Drift in a Nationally Deployed Population Health Risk Algorithm in the US Veterans Health Administration
This cohort study found that CAN algorithm performance declined from 2016 to 2021 due to shifts in outcome prevalence and distributions of health care utilization and demographic covariates. Close surveillance of clinical risk algorithms and quality metrics derived from algorithm-generated risk scores could mitigate suboptimal resource allocation or decision-making.
Authors
- Craig Kreisler
- Kristin A. Linn (ORCID: https://orcid.org/0000-0001-8463-8084)
- Ravi B. Parikh (ORCID: https://orcid.org/0000-0003-2692-6306)
- Jean Feng (ORCID: https://orcid.org/0000-0003-2041-3104)
- Amol S. Navathe (ORCID: https://orcid.org/0000-0001-7182-4988)
- Likhitha Kolla (ORCID: https://orcid.org/0000-0002-1169-906X)
- Harvineet Singh (ORCID: https://orcid.org/0000-0003-0166-9331)
- Sungkyun Park (ORCID: https://orcid.org/0000-0003-1816-310X)
- Jinbo Chen (ORCID: https://orcid.org/0000-0003-3174-2552)
- Christopher Brent Roberts (ORCID: https://orcid.org/0000-0002-6526-6067)
Institutions
- Emory University (US)
- University of California, San Francisco (US)
- Piedmont Cancer Institute (US)
- Office of Patient Care Services (US)
- Philadelphia VA Medical Center (US)
- Winship Cancer Institute
- University of Pennsylvania (US)
- National Patient Safety Foundation (US)
Publication Details
- Journal
- JAMA Health Forum
- Published
- 2025-08-15
- DOI
- https://doi.org/10.1001/jamahealthforum.2025.2717
- Citations
- 2
- Primary Topic
- Sepsis Diagnosis and Treatment
- Type
- article
- Field-Weighted Citation Impact
- 2.71