Evidence Continuity in Evaluations of Artificial Intelligence and Algorithmic Decision Support in Nursing: A Scoping Review

Background/Objectives: Artificial intelligence (AI) and algorithmic decision support are increasingly used to extract, predict, classify, or generate information relevant to nursing care. However, technical or task performance alone does not establish preservation of meaning, appropriate human use, effective care delivery, or patient benefit. This scoping review mapped nursing-specific AI evaluations and examined continuity from source information through AI output, nurse response, delivered action, and patient outcomes. Methods: PubMed, CINAHL Complete, Embase, Scopus, and Web of Science Core Collection were searched from inception on 1 July 2026 for peer-reviewed reports available by 30 June 2026, supplemented by citation and targeted searching. Two reviewers independently assessed each record at title/abstract and full-text stages; charting and verification were conducted in reviewer pairs across four authors. Related reports were linked at the investigation level. An author-developed heuristic mapped evidence location and within-case continuity from AI input (S0) to patient outcome (S5), with reference standards coded separately. Global quality, risk-of-bias, and certainty ratings were not assigned. No protocol was prospectively registered or published. Results: Fifty-nine publications spanned educational and simulation studies, technical evaluations, and clinical implementation studies across four domains: learning and benchmarking; generative care planning; wound systems; and information, workflow, and outcomes. Under the review-defined heuristic, no identified investigation documented continuity across all task-relevant transitions. Generative systems produced plausible outputs but also omissions, unsupported additions, weak individualisation, and task-dependent performance. Workflow findings varied: some documentation and handover studies reported shorter completion times, whereas an AI-supported wound-care pilot required longer completion. Patient-outcome findings were mixed and design-dependent. No included investigation directly compared a safeguard with an alternative using an endpoint matched to clinically important error detection, correction, inappropriate reliance, residual error, or harm. Conclusions: The identified corpus supports function-specific findings but provides insufficient evidence for general claims of improved nursing judgement, safety, whole-workflow efficiency or patient outcomes; this does not establish absence of benefit. Future evaluations should distinguish model performance from human–AI benefit, assess final fidelity and residual error, and link decisions, delivered care and outcomes when explaining mechanisms.

Authors

Institutions

Publication Details

Journal
Nursing Reports
Published
2026-11-17
DOI
https://doi.org/10.3390/nursrep16110417
Primary Topic
Artificial Intelligence in Healthcare and Education
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Evidence Continuity in Evaluations of Artificial Intelligence and Algorithmic Decision Support in Nursing: A Scoping Review

Rabie Adel El Arab, Victor Savi, Sameer Abdulmalik Alkubati, Wesam Taher Almagharbeh et al.
4 links
Nursing Reports
Artificial Intelligence in Healthcare and Education
article

Evidence Continuity in Evaluations of Artificial Intelligence and Algorithmic Decision Support in Nursing: A Scoping Review

Rabie Adel El Arab, Victor Savi, Sameer Abdulmalik Alkubati, Wesam Taher Almagharbeh, Mohammad Hussein Mustafa
article en

Abstract

Background/Objectives: Artificial intelligence (AI) and algorithmic decision support are increasingly used to extract, predict, classify, or generate information relevant to nursing care. However, technical or task performance alone does not establish preservation of meaning, appropriate human use, effective care delivery, or patient benefit. This scoping review mapped nursing-specific AI evaluations and examined continuity from source information through AI output, nurse response, delivered action, and patient outcomes. Methods: PubMed, CINAHL Complete, Embase, Scopus, and Web of Science Core Collection were searched from inception on 1 July 2026 for peer-reviewed reports available by 30 June 2026, supplemented by citation and targeted searching. Two reviewers independently assessed each record at title/abstract and full-text stages; charting and verification were conducted in reviewer pairs across four authors. Related reports were linked at the investigation level. An author-developed heuristic mapped evidence location and within-case continuity from AI input (S0) to patient outcome (S5), with reference standards coded separately. Global quality, risk-of-bias, and certainty ratings were not assigned. No protocol was prospectively registered or published. Results: Fifty-nine publications spanned educational and simulation studies, technical evaluations, and clinical implementation studies across four domains: learning and benchmarking; generative care planning; wound systems; and information, workflow, and outcomes. Under the review-defined heuristic, no identified investigation documented continuity across all task-relevant transitions. Generative systems produced plausible outputs but also omissions, unsupported additions, weak individualisation, and task-dependent performance. Workflow findings varied: some documentation and handover studies reported shorter completion times, whereas an AI-supported wound-care pilot required longer completion. Patient-outcome findings were mixed and design-dependent. No included investigation directly compared a safeguard with an alternative using an endpoint matched to clinically important error detection, correction, inappropriate reliance, residual error, or harm. Conclusions: The identified corpus supports function-specific findings but provides insufficient evidence for general claims of improved nursing judgement, safety, whole-workflow efficiency or patient outcomes; this does not establish absence of benefit. Future evaluations should distinguish model performance from human–AI benefit, assess final fidelity and residual error, and link decisions, delivered care and outcomes when explaining mechanisms.

Nursing ReportsVol. 16(11)
University of Turku (FI), Satakunta University of Applied Sciences (FI), King Fahd Medical City (SA), University of Ha'il (SA), Mohammed Almana College for Medical Sciences (SA), University of Tabuk (SA)
Good health and well-being
Artificial Intelligence in Healthcare and Education
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon