CoChat

Detecting clinical medication errors with AI enabled wearable cameras

Open access

The DOI matches its Crossref record; 50 of the 50 references it cites were found in a registry or an index.

Abstract

Drug-related errors are a leading cause of preventable patient harm in the clinical setting. We present the first wearable camera system to automatically detect potential errors, prior to medication delivery. We demonstrate that using deep learning algorithms, our system can detect and classify drug labels on syringes and vials in drug preparation events recorded in real-world operating rooms. We created a first-of-its-kind large-scale video dataset from head-mounted cameras comprising 4K footage across 13 anesthesiology providers, 2 hospitals and 17 operating rooms over 55 days. The system was evaluated on 418 drug draw events in routine patient care and a controlled environment and achieved 99.6% sensitivity and 98.8% specificity at detecting vial swap errors. These results suggest that our wearable camera system has the potential to provide a secondary check when a medication is selected for a patient, and a chance to intervene before a potential medical error.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41746-024-01295-2

Cites

50 of 50 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Overview of wearable camera system for detecting medication errors. a Head-mounted camera (GoPro Hero8) attached to a head strap and powered by an external battery pack. b Recording drug preparation events using wearable camera system. c Vial swap error from midazolam vial into syringe with ondansetron label. d Real-world operating room use showing that syringe and vial labels are not visible in every frame and can be obscured by the provider’s hand or tilted away. e Real-world environments containing other syringes and vials in the background unrelated to the drug drawup event. f Workflow demonstrating how drug drawup events captured through our system are detected and classified. Medication errors can be communicated to the provider through real-time auditory or visual feedback and automatically recorded.
Fig. 1 Overview of wearable camera system for detecting medication errors. a Head-mounted camera (GoPro Hero8) attached to a head strap and powered by an external battery pack. b Recording drug preparation events using wearable camera system. c Vial swap error from midazolam vial into syringe with ondansetron label. d Real-world operating room use showing that syringe and vial labels are not visible in every frame and can be obscured by the provider’s hand or tilted away. e Real-world environments containing other syringes and vials in the background unrelated to the drug drawup event. f Workflow demonstrating how drug drawup events captured through our system are detected and classified. Medication errors can be communicated to the provider through real-time auditory or visual feedback and automatically recorded.CC BY-NC-ND · PubMed Central
Fig. 2 Survey results across anesthesiology providers (n = 109). Cumulative distribution function (CDF) of providers' self-assessment: a percentage of time they scan the barcode of a drug or manually record it prior to administration and b percentage of accurately recorded drug administration events, including drug name, dose, and administration time. c Histogram displaying providers' willingness (1 lowest, 7 highest) to wear a camera capable of detecting potential medication errors, such as when a syringe label doesn’t match the drug vial during syringe preparation or when a medication is selected to which the patient is allergic. d Histogram displaying providers' willingness (1 lowest, 7 highest) to wear a camera that analyzes drug delivery actions and automatically records them in the patient’s chart.
Fig. 2 Survey results across anesthesiology providers (n = 109). Cumulative distribution function (CDF) of providers' self-assessment: a percentage of time they scan the barcode of a drug or manually record it prior to administration and b percentage of accurately recorded drug administration events, including drug name, dose, and administration time. c Histogram displaying providers' willingness (1 lowest, 7 highest) to wear a camera capable of detecting potential medication errors, such as when a syringe label doesn’t match the drug vial during syringe preparation or when a medication is selected to which the patient is allergic. d Histogram displaying providers' willingness (1 lowest, 7 highest) to wear a camera that analyzes drug delivery actions and automatically records them in the patient’s chart.CC BY-NC-ND · PubMed Central
Fig. 3 Drug label detection performance in real-world operating room environment. Precision-recall curves for a, b syringe and syringe label detector with and without image augmentation and c, d vial and vial label detector with and without image augmentation.
Fig. 3 Drug label detection performance in real-world operating room environment. Precision-recall curves for a, b syringe and syringe label detector with and without image augmentation and c, d vial and vial label detector with and without image augmentation.CC BY-NC-ND · PubMed Central
Fig. 4 Drug label classification performance in real-world operating room environments. Confusion matrices for a, b syringe labels with and without image augmentation and a background image class and c, d vial labels with and without image augmentation and a background image class.
Fig. 4 Drug label classification performance in real-world operating room environments. Confusion matrices for a, b syringe labels with and without image augmentation and a background image class and c, d vial labels with and without image augmentation and a background image class.CC BY-NC-ND · PubMed Central
Fig. 5 Subgroup analysis of drug label classification accuracy in real-world operating room environments. Classification performance of syringe and vial label classifiers across a different anesthesiology providers b different objects in the background and c whether the providers wore gloves. Numbers above each bar indicate the number of videos that were annotated in that scenario.
Fig. 5 Subgroup analysis of drug label classification accuracy in real-world operating room environments. Classification performance of syringe and vial label classifiers across a different anesthesiology providers b different objects in the background and c whether the providers wore gloves. Numbers above each bar indicate the number of videos that were annotated in that scenario.CC BY-NC-ND · PubMed Central
Fig. 6 Benchmark testing. Classification performance of syringe drug labels under a bright, dim, and dark ambient lighting conditions in the operating room using the head-mounted camera’s b default camera light settings and c low-light settings. Fraction of syringe label video frames correctly classified at different distances when recorded from d top view and e side view.
Fig. 6 Benchmark testing. Classification performance of syringe drug labels under a bright, dim, and dark ambient lighting conditions in the operating room using the head-mounted camera’s b default camera light settings and c low-light settings. Fraction of syringe label video frames correctly classified at different distances when recorded from d top view and e side view.CC BY-NC-ND · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

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?

Self-check

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

The self-check needs the paper’s full text.

Citations

Cites

Show all 50 references

Research works printed but not linked: 2

  1. Grissinger, M. The Five Rights: a destination without a map. Pharm. Ther 35, 542 (2010).
  2. Grissinger, M. Reducing errors with injectable medications. Pharm. Ther. 35, 428–451 (2010).

Cited by

See all 18

Metadata

Authors
Chan, J., Nsumba, S., Wortsman, M., Dave, A., Schmidt, L., Gollakota, S., & Michaelsen, K.
Affiliations
  • University of Washington (US)
  • Makerere University (UG)
  • Toyota Research Institute (US)
Corresponding author
  • Gollakota, Shyamnath
  • Michaelsen, Kelly
Venue
npj Digital Medicine
Published
2024-10-22
Volume
7
Issue
1
Pages
287-287
ISSN
2398-6352 (ISSN-L)
OpenAlex
W4403619512
Cited by
18
References
49
Citations by year
2026: 8 · 2025: 10
FWCI
9.94
Type
journal article
Language
en
Open access
Licence
Locations
  • npj Digital Medicine (publishedVersion, cc-by-nc-nd)
  • PubMed Central (publishedVersion, cc-by-nc-nd)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • Washington Research Foundation
  • Foundation for the National Institutes of Health (K08GM153069)
  • Foundation for Anesthesia Education and Research
Topics
  • Health Sciences › Medicine › Public Health, Environmental and Occupational Health › Patient Safety and Medication Errors
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
  • Health Sciences › Medicine › Surgery › Healthcare Technology and Patient Monitoring
Keywords
medication errors, wearable cameras, head-mounted camera, drug preparation, operating room, anesthesiology, deep learning, patient safety, error detection

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.

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

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Detecting clinical medication errors with AI enabled wearable cameras. https://app.cochat.ai/papers/10.1038/s41746-024-01295-2
The paper
Chan, J., Nsumba, S., Wortsman, M., Dave, A., Schmidt, L., Gollakota, S., & Michaelsen, K. (2024). Detecting clinical medication errors with AI enabled wearable cameras. npj Digital Medicine. https://doi.org/10.1038/s41746-024-01295-2

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.