TY - JOUR
T1 - AI as an intervention
T2 - improving clinical outcomes relies on a causal approach to AI development and validation
AU - Joshi, Shalmali
AU - Urteaga, Iñigo
AU - Van Amsterdam, Wouter A.C.
AU - Hripcsak, George
AU - Elias, Pierre
AU - Recht, Benjamin
AU - Elhadad, Noémie
AU - Fackler, James
AU - Sendak, Mark P.
AU - Wiens, Jenna
AU - Deshpande, Kaivalya
AU - Wald, Yoav
AU - Fiterau, Madalina
AU - Lipton, Zachary
AU - Malinsky, Daniel
AU - Nayan, Madhur
AU - Namkoong, Hongseok
AU - Park, Soojin
AU - Vogt, Julia E.
AU - Ranganath, Rajesh
N1 - Publisher Copyright:
© 2025 The Author(s).
PY - 2025/3/1
Y1 - 2025/3/1
N2 - The primary practice of healthcare artificial intelligence (AI) starts with model development, often using state-of-the-art AI, retrospectively evaluated using metrics lifted from the AI literature like AUROC and DICE score. However, good performance on these metrics may not translate to improved clinical outcomes. Instead, we argue for a better development pipeline constructed by working backward from the end goal of positively impacting clinically relevant outcomes using AI, leading to considerations of causality in model development and validation, and subsequently a better development pipeline. Healthcare AI should be "actionable,"and the change in actions induced by AI should improve outcomes. Quantifying the effect of changes in actions on outcomes is causal inference. The development, evaluation, and validation of healthcare AI should therefore account for the causal effect of intervening with the AI on clinically relevant outcomes. Using a causal lens, we make recommendations for key stakeholders at various stages of the healthcare AI pipeline. Our recommendations aim to increase the positive impact of AI on clinical outcomes.
AB - The primary practice of healthcare artificial intelligence (AI) starts with model development, often using state-of-the-art AI, retrospectively evaluated using metrics lifted from the AI literature like AUROC and DICE score. However, good performance on these metrics may not translate to improved clinical outcomes. Instead, we argue for a better development pipeline constructed by working backward from the end goal of positively impacting clinically relevant outcomes using AI, leading to considerations of causality in model development and validation, and subsequently a better development pipeline. Healthcare AI should be "actionable,"and the change in actions induced by AI should improve outcomes. Quantifying the effect of changes in actions on outcomes is causal inference. The development, evaluation, and validation of healthcare AI should therefore account for the causal effect of intervening with the AI on clinically relevant outcomes. Using a causal lens, we make recommendations for key stakeholders at various stages of the healthcare AI pipeline. Our recommendations aim to increase the positive impact of AI on clinical outcomes.
KW - artificial intelligence
KW - causal inference
KW - healthcare
UR - http://www.scopus.com/inward/record.url?scp=85218602910&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85218602910&partnerID=8YFLogxK
U2 - 10.1093/jamia/ocae301
DO - 10.1093/jamia/ocae301
M3 - Article
C2 - 39775871
AN - SCOPUS:85218602910
SN - 1067-5027
VL - 32
SP - 589
EP - 594
JO - Journal of the American Medical Informatics Association
JF - Journal of the American Medical Informatics Association
IS - 3
ER -