AI and machine learning in resuscitation: Ongoing research, new concepts, and key challenges

Okada, Yohei; Mertens, Mayli; Liu, Nan; Lam, Sean Shao Wei; Ong, Marcus Eng Hock

doi:10.1016/j.resplu.2023.100435

Cited by 12 publications

(3 citation statements)

References 90 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This information will allow for further recommendations to be made to improve guidelines adherence and patients' outcomes. Additionally, with the development of Artificial Intelligence (AI) and machine learning algorithms, the need for accurate documentation is even more important [17]. The use of AI and machine learning can introduce prediction models and the analysis of large-scale data to provide insights on how to optimize treatment, improving workflow during code events.…”

Section: Discussionmentioning

confidence: 99%

“…The use of AI and machine learning can introduce prediction models and the analysis of large-scale data to provide insights on how to optimize treatment, improving workflow during code events. However, the deployment of AI and machine learning algorithms would require a rigorous validation process, of which accurate documentation is essential to avoid erroneous prediction models, of the guidelines for clinical practice [17].…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Discrepancies between Retrospective Review of “Real-Time” Electronic Health Record Documentation and Prospective Observer Documentation of In-Hospital Cardiac Arrest Quality Metrics in an Academic Cardiac Intensive Care Unit

Morris,

Couperus,

Jasani

et al. 2023

JCM

View full text Add to dashboard Cite

Background: Every year, approximately 200,000 patients will experience in-hospital cardiac arrest (IHCA) in the United States. Survival has been shown to be greatest with the prompt initiation of CPR and early interventions, leading to the development of time-based quality measures. It is uncertain how documentation practices affect reports of compliance with time-based quality measures in IHCA. Methods: A retrospective review of all cases of IHCA that occurred in the Cardiac Intensive Care Unit (CICU) at an academic quaternary hospital was conducted. For each case, a member of the code team (observer) documented performance measures as part of a prospective cardiac arrest quality improvement database. We compared those data to those abstracted in the retrospective review of “real-time” documentation in a Resuscitation Narrator module within electronic health records (EHRs) to investigate for discrepancies. Results: We identified 52 cases of IHCA, all of which were witnessed events. In total, 47 (90%) cases were reviewed by observers as receiving epinephrine within 5 min, but only 42 (81%) were documented as such in the EHR review (p = 0.04), meaning that the interrater agreement for this metric was low (Kappa = 0.27, 95% CI 0.16–0.36). Four (27%) eligible patients were reported as having defibrillation within 2 min by observers, compared to five (33%) reported by the EHR review (p = 0.90), and with substantial agreement (Kappa = 0.73, 95% CI 0.66–0.79). There was almost perfect agreement (Kappa = 0.82, 95% CI 0.76–0.88) for the initial rhythm of cardiac arrest (25% shockable rhythm by observers vs. 29% for EHR review, p = 0.31). Conclusion: There was a discrepancy between prospective observers’ documentation of meeting quality standards and that of the retrospective review of “real-time” EHR documentation. A further study is required to understand the cause of discrepancy and its consequences.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Discrepancies between Retrospective Review of “Real-Time” Electronic Health Record Documentation and Prospective Observer Documentation of In-Hospital Cardiac Arrest Quality Metrics in an Academic Cardiac Intensive Care Unit

Morris,

Couperus,

Jasani

et al. 2023

JCM

View full text Add to dashboard Cite

show abstract

“…If the AI generates unexpected results and unreasonable explanations inconsistent with clinical experience or show potential risk of bias/discrimination, clinicians can bypass the AI and review results for errors or bias [ 29 ]. For example, if a patient is unexpectedly evaluated as having low likelihood of a favorable outcome, mainly driven by ethnicity or socioeconomic status of the patient, clinicians may suspect hidden discrimination or bias in the AI training data [ 30 , 31 ].…”

Section: Why Is Explainable Ai Needed In Emergency Medicine?mentioning

confidence: 99%

Explainable artificial intelligence in emergency medicine: an overview

Okada,

Ning,

Ong

2023

Clin Exp Emerg Med

Self Cite

View full text Add to dashboard Cite

Artificial intelligence (AI) and machine learning (ML) have potential to revolutionize emergency medical care by enhancing triage systems, improving diagnostic accuracy, refining prognostication, and optimizing various aspects of clinical care. However, as clinicians often lack AI expertise, they might perceive AI as a “black box,” leading to trust issues. To address this, “explainable AI,” which teaches AI functionalities to end-users, is important. This review presents the definitions, importance, and role of explainable AI, as well as potential challenges in emergency medicine. First, we introduce the terms explainability, interpretability, and transparency of AI models. These terms sound similar but have different roles in discussion of AI. Second, we indicate that explainable AI is required in clinical settings for reasons of justification, control, improvement, and discovery and provide examples. Third, we describe three major categories of explainability: pre-modeling explainability, interpretable models, and post-modeling explainability and present examples (especially for post-modeling explainability), such as visualization, simplification, text justification, and feature relevance. Last, we show the challenges of implementing AI and ML models in clinical settings and highlight the importance of collaboration between clinicians, developers, and researchers. This paper summarizes the concept of “explainable AI” for emergency medicine clinicians. This review may help clinicians understand explainable AI in emergency contexts.

show abstract