Abstract:Early recognition and treatment of sepsis are linked to improved patient outcomes. Machine learning-based early warning systems may reduce the time to recognition, but few systems have undergone clinical evaluation. In this prospective, multi-site cohort study, we examined the association between patient outcomes and provider interaction with a deployed sepsis alert system called the Targeted Real-time Early Warning System (TREWS). During the study, 590,736 patients were monitored by TREWS across five hospital… Show more
“…As one might expect, across the two patient cohorts analyzed in our study, the observed associations between the membership in the study arm and outcomes only goes away once adjusted for the presence of lactate (i.e., backdoor adjustment 9 ). Notably, Adams et al 1 do not adjust for such confounders in their analysis, which is likely to significantly bias their findings. We recommend all observational studies of ML-based predictive models to include directed acyclic graph diagrams to explicitly lay out assumptions surrounding the data generation processes, and to use pertinent causal inference techniques to mitigate the problem of biased effect estimates.…”
Section: A Causal Inference Perspectivementioning
confidence: 99%
“…Causal Inference and Backdoor Adjustment Figure A1. A directed acyclic graphical model representing the experimental setup of Adams et al 1 (top) and our proposed thought experiment (bottom). Our thought experiment was aimed at demonstrating the confounding effect of the backdoor path induced by node A, when assessing the causal effect of node S on E, T, and O. In our thought experiment, a fair coin is flipped repeatedly, starting from the time of ED triage, and an alert is fired with the landing of the first head.…”
Section: Figure A1mentioning
confidence: 99%
“…A directed acyclic graphical model representing the experimental setup of Adams et al 1 (top) and our proposed thought experiment (bottom).…”
Section: Figure A1mentioning
confidence: 99%
“…Directed acyclic graph (DAG) representations of the Standard Process of Care (A), the experimental setup of Adams et al 1 (B) and our proposed thought experiment (C). Note, ‘processes of care ‘-related factors, as well as confounders such as incoming patient volume, case complexity and frailty 8 or distractions due to false alarms produced by the TREWS algorithm can affect patient evaluation, timing of antibiotics, and outcomes.…”
Section: Figurementioning
confidence: 99%
“…
We read with interest the Adams et al 1 report of the TREWS machine learning (ML)-based sepsis early warning system. The authors conclude that large-scale randomized trials are needed to confirm their observations, but assert that their findings indicate the potential for the TREWS system to identify sepsis patients early and improve patient outcomes, including a significant decrease in mortality.
We read with interest the Adams et al. (doi: https://doi.org/10.1038/s41591-022-01894-0) report of the TREWS machine learning (ML)-based sepsis early warning system. The authors conclude that large-scale randomized trials are needed to confirm their observations, but assert that their findings indicate the potential for the TREWS system to identify sepsis patients early and improve patient outcomes, including a significant decrease in mortality. However, this conclusion is based upon a comparison of those whose alert was confirmed vs. not confirmed within 3 hours, rather than random allocation to TREWS vs. no TREWS. Using data from over 650,000 patient encounters across two distinct healthcare systems, we show that the findings of Adams et al. are likely to be severely biased due to the failure to adjust for 'processes of care'-related confounding factors.
“…As one might expect, across the two patient cohorts analyzed in our study, the observed associations between the membership in the study arm and outcomes only goes away once adjusted for the presence of lactate (i.e., backdoor adjustment 9 ). Notably, Adams et al 1 do not adjust for such confounders in their analysis, which is likely to significantly bias their findings. We recommend all observational studies of ML-based predictive models to include directed acyclic graph diagrams to explicitly lay out assumptions surrounding the data generation processes, and to use pertinent causal inference techniques to mitigate the problem of biased effect estimates.…”
Section: A Causal Inference Perspectivementioning
confidence: 99%
“…Causal Inference and Backdoor Adjustment Figure A1. A directed acyclic graphical model representing the experimental setup of Adams et al 1 (top) and our proposed thought experiment (bottom). Our thought experiment was aimed at demonstrating the confounding effect of the backdoor path induced by node A, when assessing the causal effect of node S on E, T, and O. In our thought experiment, a fair coin is flipped repeatedly, starting from the time of ED triage, and an alert is fired with the landing of the first head.…”
Section: Figure A1mentioning
confidence: 99%
“…A directed acyclic graphical model representing the experimental setup of Adams et al 1 (top) and our proposed thought experiment (bottom).…”
Section: Figure A1mentioning
confidence: 99%
“…Directed acyclic graph (DAG) representations of the Standard Process of Care (A), the experimental setup of Adams et al 1 (B) and our proposed thought experiment (C). Note, ‘processes of care ‘-related factors, as well as confounders such as incoming patient volume, case complexity and frailty 8 or distractions due to false alarms produced by the TREWS algorithm can affect patient evaluation, timing of antibiotics, and outcomes.…”
Section: Figurementioning
confidence: 99%
“…
We read with interest the Adams et al 1 report of the TREWS machine learning (ML)-based sepsis early warning system. The authors conclude that large-scale randomized trials are needed to confirm their observations, but assert that their findings indicate the potential for the TREWS system to identify sepsis patients early and improve patient outcomes, including a significant decrease in mortality.
We read with interest the Adams et al. (doi: https://doi.org/10.1038/s41591-022-01894-0) report of the TREWS machine learning (ML)-based sepsis early warning system. The authors conclude that large-scale randomized trials are needed to confirm their observations, but assert that their findings indicate the potential for the TREWS system to identify sepsis patients early and improve patient outcomes, including a significant decrease in mortality. However, this conclusion is based upon a comparison of those whose alert was confirmed vs. not confirmed within 3 hours, rather than random allocation to TREWS vs. no TREWS. Using data from over 650,000 patient encounters across two distinct healthcare systems, we show that the findings of Adams et al. are likely to be severely biased due to the failure to adjust for 'processes of care'-related confounding factors.
This Viewpoint discusses recent legal directives by the DHHS and FDA that could increase health care entities’ liability for possible discriminatory biases of clinical algorithms and the need for additional legal clarity to avoid adverse effects on algorithm development and use.
ImportanceGiven the importance of rigorous development and evaluation standards needed of artificial intelligence (AI) models used in health care, nationwide accepted procedures to provide assurance that the use of AI is fair, appropriate, valid, effective, and safe are urgently needed.ObservationsWhile there are several efforts to develop standards and best practices to evaluate AI, there is a gap between having such guidance and the application of such guidance to both existing and new AI models being developed. As of now, there is no publicly available, nationwide mechanism that enables objective evaluation and ongoing assessment of the consequences of using health AI models in clinical care settings.Conclusion and RelevanceThe need to create a public-private partnership to support a nationwide health AI assurance labs network is outlined here. In this network, community best practices could be applied for testing health AI models to produce reports on their performance that can be widely shared for managing the lifecycle of AI models over time and across populations and sites where these models are deployed.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
