CoChat

Integrating a Machine Learning System Into Clinical Workflows: Qualitative Study

Open access
Sandhu, Sahil, Lin, Anthony L, Brajer, Nathan, Sperling, Jessica, Ratliff, William, Bedoya, Armando D, Balu, Suresh, O'Brien, Cara, Sendak, Mark P · 2020 · Journal of Medical Internet Research

Published version of a preprint

The DOI matches its Crossref record; 50 of the 50 references it cites were found in a registry or an index; 25 of the 34 sources its claims cite were checked for support.

Abstract

Background: Machine learning models have the potential to improve diagnostic accuracy and management of acute conditions. Despite growing efforts to evaluate and validate such models, little is known about how to best translate and implement these products as part of routine clinical care. Objective: This study aims to explore the factors influencing the integration of a machine learning sepsis early warning system (Sepsis Watch) into clinical workflows. Methods: We conducted semistructured interviews with 15 frontline emergency department physicians and rapid response team nurses who participated in the Sepsis Watch quality improvement initiative. Interviews were audio recorded and transcribed. We used a modified grounded theory approach to identify key themes and analyze qualitative data. Results: A total of 3 dominant themes emerged: perceived utility and trust, implementation of Sepsis Watch processes, and workforce considerations. Participants described their unfamiliarity with machine learning models. As a result, clinician trust was influenced by the perceived accuracy and utility of the model from personal program experience. Implementation of Sepsis Watch was facilitated by the easy-to-use tablet application and communication strategies that were developed by nurses to share model outputs with physicians. Barriers included the flow of information among clinicians and gaps in knowledge about the model itself and broader workflow processes. Conclusions: This study generated insights into how frontline clinicians perceived machine learning models and the barriers to integrating them into clinical workflows. These findings can inform future efforts to implement machine learning interventions in real-world settings and maximize the adoption of these interventions.

Overview

CoChat Summary

TL;DR

This qualitative study interviewed 15 frontline clinicians to identify factors affecting integration of a machine learning sepsis early warning system, Sepsis Watch, into emergency department workflows. It found three dominant themes—perceived utility and trust, implementation processes, and workforce considerations—with clinician trust shaped by personal experience, model transparency limits, and human communication.

Key findings

  • Three dominant themes emerged: perceived utility and trust, implementation of Sepsis Watch processes, and workforce considerations, with 10 corresponding subthemes.
  • RRT nurses, the primary users, spoke positively about Sepsis Watch accuracy and the easy-to-use color-coded application; ED physicians reported less trust, citing algorithm components such as blood cultures being too heavily weighted, missed sepsis cases, and false positives.
  • Implementation was facilitated by the tablet/web application and nurse-developed communication strategies, but barriers included information flow challenges, workflow interruptions, and gaps in knowledge about the black box model.
  • The program created a new Sepsis Watch Nurse role; nurses valued it and identified good clinical judgment, sepsis knowledge, critical care experience, and communication skills as important, while strong computer skills were not considered necessary.

Methods

  • Semistructured interviews were conducted with 15 participants—7 ED attending physicians and 8 RRT nurses—from January 2019 to April 2019, starting 4 months into Sepsis Watch implementation at Duke University Hospital; interviews averaged 35 minutes and continued until thematic saturation.
  • A modified grounded theory approach was used with three coding phases, and data were analyzed in NVivo 12; interview guides were informed by the situational awareness model and piloted among 3 clinicians.
  • Sepsis Watch used a deep learning model trained on 42,000 inpatient encounters and 32 million data points to predict likelihood of meeting a sepsis phenotype within the subsequent 4 hours, with outputs displayed as black, red, orange, or yellow cards.

Limitations

  • The paper states that findings should be generalized with caution to settings differing in organizational structure, capacity, and professional norms; the study focuses on workflow integration rather than the multi-year planning and stakeholder engagement process; only frontline clinicians were interviewed, excluding leadership, administrators, data scientists, and patients; and including clinicians involved in program development may have biased findings positively.
  • The paper also states that further studies with larger samples and survey approaches are needed to quantify adoption factors and examine variation by clinician characteristics.
  • Evident from the text: this was a single-site academic hospital study using convenience sampling and qualitative self-report, and it did not measure patient outcomes.

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

Integrity signals

Registry

Crossref · 10.2196/22421

Cites

50 of 50 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Sepsis Watch workflow. ED: emergency department; RRT: rapid response team.
Figure 1 Sepsis Watch workflow. ED: emergency department; RRT: rapid response team.CC BY · PubMed Central

Verification

Claims the paper makes

15 Valid · 1 Not valid · 3 Not checked · 87 Own results and interpretations

47 more claims · 3 with their cited sources checked

Create a free account to see all 106 claims and every source verdict.

See all claims — free account
  • It may state two separate claims as one. 27
  • Probably not a substantial claim. 30
  • The quote only partly states this claim. 1

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

4 more claims about this paper

Create a free account to see every claim other papers make about this one.

See all claims — free account

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

3 of 3

Citations

Cites

Show all 50 references

Research works printed but not linked: 2

  1. TorioCMMooreBJNational Inpatient Hospital Costs: The Most Expensive Conditions by Payer, 2013: Statistical Brief #204Healthcare Cost and Utilization Project (HCUP)2020-11-13https://europepmc.org/abstract/med/27359025
  2. Futoma, J, arXivstat

Cited by

Metadata

Authors
Sandhu, S., Lin, A. L., Brajer, N., Sperling, J., Ratliff, W., Bedoya, A. D., Balu, S., O'Brien, C., & Sendak, M. P.
Affiliations
Duke University (US)
Corresponding author
Sandhu, Sahil
Venue
Journal of Medical Internet Research
Published
2020-11-19
Volume
22
Issue
11
Pages
e22421
ISSN
1438-8871 (ISSN-L)
OpenAlex
W3094937636
Cited by
168
References
50
Citations by year
2026: 37 · 2025: 35 · 2024: 38 · 2023: 25 · 2022: 23 · 2021: 10
FWCI
4.29
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • Journal of Medical Internet Research (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
  • Health Sciences › Health Professions › Health Information Management › Electronic Health Records Systems
  • Health Sciences › Medicine › Critical Care and Intensive Care Medicine › Sepsis Diagnosis and Treatment
Keywords
machine learning integration, clinical workflow, implementation barriers, clinician trust, perceived usefulness, emergency physicians, rapid response teams, nurse-physician communication, clinical decision support, technology adoption
MeSH terms
Machine Learning, Humans, Qualitative Research, Workflow

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 2026-10-28, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-28 (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.