A Clinician's Guide to Understanding Bias in Critical Clinical Prediction Models

João Matos; Jack Gallifant; Anand Chowdhury; Nicoleta Economou-Zavlanos; Marie-Laure Charpignon; Judy Gichoya; Leo Anthony Celi; Lama Nazer; Heather King; An-Kwok Ian Wong

doi:10.1016/j.ccc.2024.05.011

A Clinician's Guide to Understanding Bias in Critical Clinical Prediction Models

Crit Care Clin. 2024 Oct;40(4):827-857. doi: 10.1016/j.ccc.2024.05.011.

Authors

Affiliations

¹ University of Porto (FEUP), Porto, Portugal; Institute for Systems and Computer Engineering, Technology and Science (INESC TEC), Porto, Portugal; Laboratory for Computational Physiology, Institute for Medical Engineering and Science, Massachusetts Institute of Technology, Cambridge, MA, USA.
² Laboratory for Computational Physiology, Institute for Medical Engineering and Science, Massachusetts Institute of Technology, Cambridge, MA, USA; Department of Critical Care, Guy's and St Thomas' NHS Trust, London, UK.
³ Division of Pulmonary, Allergy, and Critical Care Medicine, Department of Medicine, Duke University, Durham, NC, USA.
⁴ Duke Health, AI Health, Durham, NC, USA.
⁵ Institute for Data Systems and Society, Massachusetts Institute of Technology, Cambridge, MA, USA.
⁶ Department of Radiology, Emory University, Atlanta, GA, USA.
⁷ Laboratory for Computational Physiology, Institute for Medical Engineering and Science, Massachusetts Institute of Technology, Cambridge, MA, USA; Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA, USA; Department of Medicine, Beth Israel Deaconess Medical Center, Boston, MA, USA.
⁸ Department of Pharmacy, King Hussein Cancer Center, Amman, Jordan.
⁹ Durham VA Health Care System, Health Services Research and Development, Center of Innovation to Accelerate Discovery and Practice Transformation (ADAPT), Durham, NC, USA; Department of Population Health Sciences, Duke University, Durham, NC, USA; Division of General Internal Medicine, Duke University, Duke University School of Medicine, Durham, NC, USA.
¹⁰ Division of Pulmonary, Allergy, and Critical Care Medicine, Department of Medicine, Duke University, Durham, NC, USA; Department of Biostatistics and Bioinformatics, Duke University, Division of Translational Biomedical Informatics, Durham, NC, USA. Electronic address: med@aiwong.com.

PMID: 39218488
DOI: 10.1016/j.ccc.2024.05.011

Abstract

This narrative review focuses on the role of clinical prediction models in supporting informed decision-making in critical care, emphasizing their 2 forms: traditional scores and artificial intelligence (AI)-based models. Acknowledging the potential for both types to embed biases, the authors underscore the importance of critical appraisal to increase our trust in models. The authors outline recommendations and critical care examples to manage risk of bias in AI models. The authors advocate for enhanced interdisciplinary training for clinicians, who are encouraged to explore various resources (books, journals, news Web sites, and social media) and events (Datathons) to deepen their understanding of risk of bias.

Keywords: AI; Artificial intelligence; Bias; Machine learning; Prediction models.

Publication types

Review

MeSH terms

Artificial Intelligence*
Bias
Clinical Decision-Making
Critical Care* / standards
Humans