CoChat

Clinician Perspectives on Ambient AI Scribes in the Intensive Care Unit: Qualitative Interview Study

Open access

The DOI matches its Crossref record; 35 of the 35 references it cites were found in a registry or an index; 12 of the 22 sources its claims cite were checked for support.

Abstract

Background In intensive care unit (ICU) settings, structured team-based communication, such as multidisciplinary rounds, handoffs, and goals-of-care discussions, is foundational to high-quality care. However, accurately documenting these complex discussions in the medical record remains a challenge due to time pressures, documentation burdens, and competing clinical demands. Ambient artificial intelligence (AI) scribes, which passively transcribe and summarize spoken interactions, offer a potential solution to assist ICU clinicians with documentation. Yet, little is known about how ICU clinicians perceive the integration of these tools into their high-stakes, collaborative workflows. Objective This study explores clinicians' perceptions of integrating ambient AI scribes into structured team-based ICU discussions, including multidisciplinary rounds, handoffs and transitions of care, and goals-of-care discussions, with the broader goal of informing the implementation of these scribes into real-world ICU clinical workflows. Methods Interviews and focus groups were conducted with ICU clinicians, including nurses, attendings, trainees (residents/fellows), respiratory therapists, and advanced practice practitioners, who routinely participate in structured ICU discussions. Transcripts were analyzed using grounded theory to identify documentation needs, barriers to documentation, and considerations for the implementation of ambient AI scribes in the ICU setting. Results A total of 52 ICU clinicians participated in the study.

Overview

CoChat Summary

TL;DR

ICU clinicians in this qualitative study were optimistic that ambient AI scribes could reduce documentation burden and improve capture of structured team-based discussions, but they emphasized needs for transparency, consent, and customization. Successful implementation was seen as contingent on data governance, clinician training, specialty-specific output, and trust.

Key findings

  • Among 52 ICU clinicians—18 attending physicians, 5 advanced practice practitioners, 10 trainees, 9 nurses, and 10 respiratory therapists—participants valued accurate documentation but reported persistent barriers including time constraints, documentation burden, and competing patient care and teaching responsibilities.
  • Clinicians were enthusiastic about ambient AI scribes’ potential to reduce documentation burden and improve quality, but requested personalized outputs, robust consent protocols, and transparency around data use.
  • Participants viewed ambient AI scribes as promising for enhancing documentation fidelity and communication quality, especially for multidisciplinary rounds, handoffs/transitions of care, and goals-of-care discussions.
  • Implementation was perceived as contingent on transparent data governance, specialty-specific customization, and sustained efforts to build clinician trust.

Methods

  • Qualitative interview and focus group study conducted April–July 2025 with ICU clinicians at two UCLA Health hospitals; participants included physicians, trainees, advanced practice practitioners, nurses, and respiratory therapists routinely involved in ICU structured discussions.
  • Semistructured interviews and focus groups used a guide developed for the study, guided by the RE-AIM/PRISM framework; a live demonstration of the Nabla ambient AI scribe was provided at the start of interviews.
  • Transcripts were analyzed using grounded theory with inductive and deductive thematic analysis; two authors developed and refined a codebook, independently coded transcripts, and met regularly to reconcile differences and reach consensus using Atlas.ti.

Limitations

  • The paper states that ambient AI scribes were not yet in routine inpatient use, so participants’ views were based on anticipated rather than observed use.
  • The paper states that patient and family perspectives were absent, and that the live Nabla demonstration may have introduced priming effects.
  • The text also makes evident that saturation within individual clinician groups cannot be assumed because some roles were underrepresented; transcripts were not returned for member checking, and individual demographic data were not collected to protect anonymity in smaller subgroups.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.2196/81445

Cites

35 of 35 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

5 Valid · 2 Not valid · 5 Not checked · 62 Own results and interpretations

26 more claims

Create a free account to see all 74 claims and every source verdict.

See all claims — free account
  • It may state two separate claims as one. 19
  • Probably not a substantial claim. 29

Create a free account to see these claims.

Citations

Cites

Show all 35 references

Research works printed but not linked: 3

  1. Weir, Methods Inf Med, 42, 61
  2. Creswell JW . Research Design: Qualitative, Quantitative, and Mixed Methods Approaches. 3rd ed. Sage Publications; 2009. ISBN: 978-1-4129-6556-9
  3. Hays DG Singh AA . Qualitative Inquiry in Clinical and Educational Settings. Guilford Press; 2012. ISBN: 978-1-60918-245-8

Cited by

See all 1

Metadata

Authors
Jalilian, L., Manafi, N., Vandiver, M. S., Lukac, P., & Kadambi, A.
Affiliations
  • Ronald Reagan UCLA Medical Center (US)
  • University of California, Los Angeles (US)
Corresponding author
Jalilian, Laleh
Venue
JMIR Medical Informatics
Published
2026-07-02
Volume
14
Pages
e81445-e81445
ISSN
2291-9694 (ISSN-L)
OpenAlex
W7160917441
Cited by
0
References
31
FWCI
0.00
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • JMIR Medical Informatics (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • PubMed Central (publishedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
Topics
  • Health Sciences › Health Professions › Health Information Management › Electronic Health Records Systems
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
  • Health Sciences › Nursing › Research and Theory › Nursing Diagnosis and Documentation
Keywords
ambient AI scribes, intensive care unit, clinical documentation, documentation burden, multidisciplinary rounds, handover, care transitions, goals of care discussions, clinician perceptions, data governance, data consent, clinician trust
MeSH terms
Intelligent Systems, Artificial Intelligence, Attitude of Health Personnel, Documentation, Humans, Intensive Care Units, Interviews as Topic, Focus Groups, Qualitative Research

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.