Clinical Implementation of Artificial Intelligence Scribes in Health Care: A Systematic Review
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Abstract Artificial intelligence (AI) scribes use advanced speech recognition and natural language processing to automate clinical documentation and ease administrative burden. However, little is known about the effect of AI scribes on clinicians, patients, and organizations. This study aimed to (1) propose an evaluation framework to guide future AI scribe implementations, (2) describe the effect of AI scribes along the domains proposed in the developed evaluation framework, and (3) identify gaps in the AI scribe implementation literature to be evaluated in future studies. Databases including Embase, Embase Classic, and Ovid Medline were searched, and a manual review was conducted of the New England Journal of Medicine AI. Studies published after 2021 that reported on the implementation of AI scribes in health care were included. Descriptive analysis was undertaken. Quality assessment was undertaken using the Newcastle–Ottawa Scale. The nominal group technique was used to develop an evaluation framework. Eleven studies met the inclusion criteria, with 10 published in 2024. The most frequently used AI scribe was Dragon Ambient eXperience (n = 7, 64%). While clinicians often reported improved documentation quality, AI scribe accuracy varied, frequently requiring manual edits and raising occasional concerns about errors. Nine of 10 studies reported improvements in at least one efficiency metric, and seven of ten studies highlighted positive effects on clinician wellness and burnout. Patient experience was assessed in three studies, all reporting favorable outcomes. AI scribes represent a promising tool for improving clinical efficiency and alleviating documentation burden. This systematic review highlights the potential benefits of AI scribes, including reduced documentation time and enhanced clinician satisfaction, while also identifying critical challenges such as variable adoption, performance limitations, and gaps in evaluation.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.011 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.008 | 0.001 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it