Implementing Artificial Intelligence in Daily Clinical Practice, from Voice Dictation to Operating Room Organization: A Targeted Narrative Review with Hysteroscopy as a Clinical Exemplar
- 1Department of Obstetrics, Gynecology and Reproduction, Dexeus University Hospital, Gran Via Carles III 71–75, 08028 Barcelona, Spain
- 2Clinique Bizet, 23 Rue Georges Bizet, 75116 Paris, France
- 3Department of Obstetrics and Gynecology, Hôpital Lariboisière, 2 Rue Ambroise Paré, 75010 Paris, France
Abstract
Artificial intelligence (AI) is entering routine care through documentation tools, decision support, and hospital operations. The central question is whether these systems improve clinical work and patient care once embedded in routine workflows, not only whether they perform well on test data. This targeted narrative review follows AI from voice dictation and ambient documentation to operating room (OR) organization, using hysteroscopy as a clinical exemplar of the transition from technical performance to clinical utility. Evidence identified through targeted searches completed on 6 September 2026 was appraised with a framework that distinguishes image-level from patient-level results, internal from independent validation, and discrimination from clinical utility. Randomized trials and multisite observational data show modest, product-dependent benefits of clinician-reviewed ambient documentation; inaccuracies persist. Evidence for patient-message drafting derives from ratings of responses to public-forum questions and supports only bounded, supervised feasibility testing. In hysteroscopy, selected models discriminate well in narrowly defined tasks, but most studies are retrospective, image-level, or internally validated, and patient-level errors in high-risk categories are seldom reported. Reported gains in OR scheduling come largely from simulation. Safe implementation requires a defined clinical need, baseline measurement, subgroup-aware local validation, accountable review, and predefined criteria for expansion or withdrawal.
Related
- The Use of Ambient Dictation Artificial Intelligence in Clinical Spaces in Surgery: A Scoping Review
- Studying the effect of implementing the World Health Organization's labor and delivery management program on improving mothers' experience and satisfaction of childbirth.
- A Systems Approach toward AI-Enabled Radiology Dictation for Workflow Efficiency in India.
- The Role of Hysteroscopy in Reproductive Medicine and Subfertility: A Critical Narrative Review
- Chronic Retained Products of Conception Masked by Müllerian Duct Anomaly: The Diagnostic Value of Hysteroscopy
BibTeX
@article{haimovich2026implementing,
title = {{Implementing Artificial Intelligence in Daily Clinical Practice, from Voice Dictation to Operating Room Organization: A Targeted Narrative Review with Hysteroscopy as a Clinical Exemplar}},
author = {Sergio Haimovich and Afchine Fazel},
journal = {Journal of Clinical Medicine},
year = {2026},
month = nov,
doi = {10.3390/jcm15218688},
url = {https://www.mdpi.com/2077-0383/15/21/8688},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Implementing Artificial Intelligence in Daily Clinical Practice, from Voice Dictation to Operating Room Organization: A Targeted Narrative Review with Hysteroscopy as a Clinical Exemplar
AU - Sergio Haimovich
AU - Afchine Fazel
PY - 2026
DA - 2026/11/07
JO - Journal of Clinical Medicine
DO - 10.3390/jcm15218688
UR - https://www.mdpi.com/2077-0383/15/21/8688
AB - Artificial intelligence (AI) is entering routine care through documentation tools, decision support, and hospital operations. The central question is whether these systems improve clinical work and patient care once embedded in routine workflows, not only whether they perform well on test data. This targeted narrative review follows AI from voice dictation and ambient documentation to operating room (OR) organization, using hysteroscopy as a clinical exemplar of the transition from technical performance to clinical utility. Evidence identified through targeted searches completed on 6 September 2026 was appraised with a framework that distinguishes image-level from patient-level results, internal from independent validation, and discrimination from clinical utility. Randomized trials and multisite observational data show modest, product-dependent benefits of clinician-reviewed ambient documentation; inaccuracies persist. Evidence for patient-message drafting derives from ratings of responses to public-forum questions and supports only bounded, supervised feasibility testing. In hysteroscopy, selected models discriminate well in narrowly defined tasks, but most studies are retrospective, image-level, or internally validated, and patient-level errors in high-risk categories are seldom reported. Reported gains in OR scheduling come largely from simulation. Safe implementation requires a defined clinical need, baseline measurement, subgroup-aware local validation, accountable review, and predefined criteria for expansion or withdrawal.
SN - 2077-0383
ER -