Abstract
Abstract
Introduction
Artificial intelligence (AI) technologies are increasingly being developed and deployed to support clinical decision-making, care delivery and patient monitoring in healthcare. However, the adoption of AI-driven solutions by nurses, who comprise the largest segment of the healthcare workforce and are central to patient care, has been limited to date. Understanding nurses’ perceptions of barriers and facilitators to AI adoption is critical for successful integration of AI in nursing practice. This systematic review aims to identify, appraise and synthesise qualitative evidence on nurses’ perceived barriers and facilitators to adopting AI-driven solutions in their clinical practice.
Methods and analysis
We will conduct systematic searches across eight electronic databases (PubMed, Web of Science, Embase, CINAHL, MEDLINE, The Cochrane Library, PsycINFO and Scopus) from inception to January 2025, supplemented by hand-searching reference lists and grey literature. Primary qualitative studies and qualitative components of mixed-methods studies exploring licensed/registered nurses’ perceptions of AI adoption in clinical settings will be included. Two independent reviewers will screen studies, extract data using standardised forms and assess methodological quality using the Critical Appraisal Skills Programme checklist. We will employ meta-ethnography to synthesise the qualitative evidence, involving systematic comparison and translation of concepts across studies to develop overarching themes and a theoretical framework. The Grading of Recommendations Assessment, Development and Evaluation Confidence in the Evidence from Reviews of Qualitative Research (GRADE-CERQual) approach will be used to assess confidence in review findings. The protocol follows the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) guidelines and the Enhancing Transparency in Reporting the Synthesis of Qualitative Research (ENTREQ) statement.
Ethics and dissemination
No ethical approval is required as this systematic review will synthesise data from published studies only. The findings will provide valuable insights to inform the development, implementation and evaluation of nurse-oriented strategies for AI integration in healthcare delivery. Results will be disseminated through peer-reviewed publication, conference presentations and stakeholder engagement activities.
PROSPERO registration number
CRD42024602808.
Keywords: Artificial Intelligence, Machine Learning, Nursing Care, Implementation Science, Digital Technology
STRENGTHS AND LIMITATIONS OF THIS STUDY.
This systematic review employs rigorous meta-ethnography methodology with independent duplicate screening, data extraction and quality assessment using validated tools (Critical Appraisal Skills Programme checklist and Grading of Recommendations Assessment, Development and Evaluation Confidence in the Evidence from Reviews of Qualitative Research (GRADE-CERQual)), ensuring high methodological quality and transparency in evidence synthesis.
The meta-ethnography approach will enable deep interpretative analysis to generate new theoretical insights beyond simple aggregation of findings.
The rigorous methodology includes independent duplicate screening, data extraction and quality assessment, with systematic use of GRADE-CERQual to assess confidence in findings.
The review is limited to English language publications, potentially missing relevant evidence from non-English speaking countries.
The synthesis will be dependent on the quality and quantity of available qualitative studies, which may be limited given the emerging nature of AI implementation in healthcare.
Introduction
Background
Artificial intelligence (AI) encompasses a range of advanced computational methods and systems that can perform tasks requiring human-like intelligence, such as learning, reasoning, problem-solving and decision-making.1 In healthcare, AI technologies—including machine learning, natural language processing, robotics and computer vision—are increasingly being developed and deployed to support or augment various aspects of clinical practice.2 Examples of AI applications in healthcare include diagnostic support tools, clinical decision support systems (CDSS), patient risk prediction models, care coordination platforms and remote patient monitoring devices.3,5
The integration of AI-driven solutions in healthcare delivery has the potential to improve the efficiency, quality, safety and value of care.6 7 By processing and analysing vast amounts of structured and unstructured clinical data, AI systems can provide clinicians with timely, evidence-based recommendations and insights to support their decision-making at the point of care.8 AI tools can also automate routine tasks, streamline workflows and enable more personalised and proactive care management.9 10 Additionally, AI technologies can extend the reach and impact of the healthcare workforce by facilitating virtual care delivery, remote monitoring and patient self-management support.11
Despite the growing availability and potential benefits of AI in healthcare, the adoption of AI-driven solutions in routine clinical practice has been limited to date.12 13 Previous studies have identified various technical, professional, organisational, financial, legal and ethical challenges to the successful implementation and use of AI systems by healthcare providers.14,16 These challenges include issues related to data quality and interoperability, algorithmic bias and transparency, model validation and updating, user trust and acceptance, workflow integration, liability and accountability, privacy and security, health equity, among others.17,19
Nurses, as the largest group of healthcare professionals and the primary providers of direct patient care, play a critical role in the adoption and effective use of AI technologies at the point of care.20 21 Nurses’ perceptions, attitudes, skills and behaviours related to AI can significantly influence the success of AI implementation efforts in healthcare organisations.22 The successful integration of AI-driven solutions into nursing practice relies heavily on nurses’ adoption and acceptance of these technologies.23 Nurses’ willingness to embrace AI-driven solutions is influenced by a complex interplay of individual, organisational and technological factors.24 Understanding these factors is crucial for the effective implementation and utilisation of AI technologies in nursing care.
The Technology Acceptance Model (TAM) and its extension, the Unified Theory of Acceptance and Use of Technology (UTAUT), provide established theoretical frameworks for understanding technology adoption in healthcare settings. TAM emphasises perceived usefulness and perceived ease of use as key determinants of technology acceptance, while UTAUT incorporates additional factors including social influence, facilitating conditions and effort expectancy. These theoretical models will guide our synthesis by providing a conceptual lens through which to interpret nurses’ perceptions and experiences with AI adoption. The multilevel factors identified in our secondary research questions (individual, professional, organisational and technological) align with the broader socio-technical systems approach to technology implementation, recognising that successful AI adoption depends on the complex interplay between human, organisational and technological elements.
Previous studies have explored the barriers and facilitators to nurses’ adoption of healthcare technologies, such as electronic health records (EHRs) and CDSS.25 26 These studies have identified various factors influencing nurses’ acceptance and utilisation of technologies, such as ease of use, perceived usefulness, training and support, organisational culture and compatibility with existing workflows.27 However, the unique characteristics of AI-driven solutions, such as their autonomous decision-making capabilities and the need for extensive training data, may present distinct challenges and opportunities for nurses’ adoption.19
The adoption of AI-driven solutions in nursing practice has the potential to improve patient care, enhance clinical decision-making and optimise workflows.28 AI technologies can assist nurses in various tasks, such as risk assessment, early detection of deterioration, medication management and patient education.3 For example, AI-powered predictive models can identify patients at high risk of adverse events, enabling nurses to provide timely interventions and prevent complications.29 AI-driven virtual nursing assistants can provide personalised patient education and support, improving patient engagement and self-management.30
Despite the potential benefits, the adoption of AI-driven solutions in nursing practice faces several challenges. Nurses may have concerns about the reliability and accuracy of AI-generated recommendations, particularly in complex clinical situations.31 The lack of transparency in AI algorithms, often referred to as the ‘black box’ problem, can hinder nurses’ trust and confidence in these technologies.32 Additionally, the integration of AI-driven solutions into existing workflows may require significant changes in nursing practice, which can be met with resistance and apprehension.33
Understanding the barriers and facilitators influencing nurses’ adoption of AI-driven solutions is crucial for the effective implementation and utilisation of these technologies in clinical practice.34 Identifying the factors that hinder or facilitate nurses’ acceptance of AI can inform strategies to support the successful integration of these innovative tools into nursing care.28 Qualitative research methods, such as interviews and focus groups, provide valuable insights into nurses’ perspectives, experiences and attitudes towards AI adoption.35 Synthesising the qualitative evidence on this topic can provide a comprehensive understanding of the complex factors influencing nurses’ adoption of AI-driven solutions.
Previous reviews have not explored the unique perspectives of nurses.36 37 A few recent studies have qualitatively investigated nurses’ experiences with and perceptions of AI in different practice settings,38,42 but their findings have not been systematically synthesised. Therefore, this systematic review aims to address this gap by identifying, appraising and synthesising available qualitative evidence on nurses’ perceived barriers and facilitators to adopting AI-driven solutions in their clinical practice.
Review aim and questions
The aim of this systematic review is to identify, appraise and synthesise available qualitative evidence on nurses’ perceived barriers and facilitators to adopting AI-driven solutions in their clinical practice. The primary review question is: What are nurses’ perceptions of the barriers and facilitators to adopting AI-driven solutions in their clinical practice?
The secondary review questions are organised around four distinct but interconnected levels of analysis:
Individual-level factors: What demographic characteristics, technology attitudes and AI knowledge/skills do nurses perceive as barriers or facilitators to adopting AI solutions in practice? This level will examine personal attributes, self-efficacy and individual readiness for change.
Professional-level factors: What education/training needs, scope of practice considerations, professional standards/ethics concerns and interprofessional collaboration issues do nurses perceive as barriers or facilitators to adopting AI solutions in practice? This level will focus on professional identity and role-related factors.
Organisational-level factors: What leadership support mechanisms, resource availability, policies/procedures and implementation processes do nurses perceive as barriers or facilitators to adopting AI solutions in practice? This level will examine institutional and systemic influences.
Technological factors: What design/usability features, interoperability requirements, workflow integration capabilities and technical support needs do nurses perceive as barriers or facilitators to adopting AI solutions in practice? This level will focus on the AI technology characteristics themselves.
Contextual variations: How do nurses’ perceptions of barriers and facilitators differ by practice setting, specialty area, experience level, geographic location and other contextual factors?
The synthesis will explicitly examine how these levels interact and influence each other, avoiding conceptual overlap by clearly delineating the scope of each analytical category during data extraction and analysis.
Methods
Study design
This systematic review will follow best practice guidelines for the synthesis of qualitative research, including the Enhancing Transparency in Reporting the Synthesis of Qualitative Research (ENTREQ) statement43 and the PRISMA 2020 statement.44 The review protocol was developed using the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) checklist45 and the ENTREQ statement.43 The protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) prior to commencing the review. This protocol was registered in PROSPERO in December 2024 (CRD42024602808). The systematic searches will commence in January 2025, with study selection and data extraction planned for completion by July 2025. Data synthesis and manuscript preparation are anticipated to be completed by September 2025.
This review will employ the meta-ethnography approach originally described by Noblit and Hare46 and further developed by France and colleagues47 48 to synthesise qualitative evidence. Meta-ethnography is an interpretative approach that involves the reciprocal translation and synthesis of concepts across studies to generate new insights and theoretical understandings of a phenomenon.46 The seven phases of meta-ethnography are: (1) getting started, (2) deciding what is relevant, (3) reading included studies, (4) determining how studies are related, (5) translating studies into one another, (6) synthesising translations and (7) expressing the synthesis.46
Eligibility criteria
Types of studies
This review will include primary qualitative studies that explore nurses’ perceptions of factors influencing their adoption of AI technologies in clinical practice. Qualitative studies of any design will be eligible, including, but not limited to phenomenology, grounded theory, ethnography, case study and qualitative description. Mixed-methods studies that report qualitative findings will also be included. For mixed-methods studies, we will extract only the qualitative components and ensure these can be clearly distinguished from quantitative findings. We will assess whether the qualitative data collection and analysis methods are adequately described and whether the qualitative findings are reported separately from quantitative results. Studies where qualitative and quantitative findings are inseparably integrated or where qualitative methods are inadequately described will be excluded. We anticipate challenges in isolating qualitative data from mixed-methods studies, particularly when findings are presented in integrated formats. To address this, we will contact study authors when necessary to obtain separate qualitative findings or additional methodological details. If qualitative components cannot be adequately isolated, these studies will be excluded from the synthesis. Studies must use qualitative methods for both data collection (eg, interviews, focus groups, observations and documents) and data analysis (eg, thematic analysis, grounded theory and framework analysis). Studies that collect qualitative data but do not analyse it using a qualitative approach will be excluded. Non-research publications such as editorials, commentaries and literature reviews will also be excluded.
Types of participants
Studies must include licensed/registered nurses of any level (eg, registered nurses, nurse practitioners and licensed practical/vocational nurses) who provide direct patient care in any healthcare setting (eg, hospitals, primary care practices, long-term care facilities, community health centres and schools). Studies may include nurses from a single setting or from multiple settings. Studies with mixed samples of healthcare professionals will be included if nurses’ perspectives are reported separately and can be clearly distinguished from those of other professionals.
Phenomenon of interest
The phenomenon of interest is nurses' perceptions of factors influencing their adoption of AI technologies in clinical practice. AI technologies refer to any applications or systems that use AI techniques (eg, machine learning, natural language processing, computer vision and robotics) to perform tasks that typically require human-like intelligence, with the goal of supporting or augmenting nurses’ clinical decision-making, care delivery and patient monitoring. Examples of AI technologies relevant to nursing practice include CDSS, predictive analytics tools, intelligent documentation assistants, care coordination platforms, patient risk stratification models, remote monitoring devices, among others. Studies may focus on a specific AI application or discuss AI technologies more broadly.
Context
Studies must explore nurses’ perceptions related to the adoption or use of AI technologies in real-world clinical practice settings. Studies conducted in educational or simulation settings without a practice component will be excluded. Studies from any geographic location and healthcare system will be eligible.
Outcome
The outcome of interest is nurses’ perceived barriers and facilitators to adopting AI technologies in their clinical practice. Barriers refer to any factors that nurses perceive as hindering or challenging their ability or willingness to adopt AI technologies. Facilitators refer to any factors that nurses perceive as enabling or motivating their adoption of AI technologies. Barriers and facilitators may be identified at multiple levels, including the individual nurse, the healthcare team/profession, the healthcare organisation and the technology itself. Perceived barriers and facilitators may be related to nurses’ knowledge, skills, attitudes/beliefs, role/scope of practice, workflow/workload, interprofessional collaboration, leadership/organisational support, education/training, resources/infrastructure, policies/protocols, technology design/performance, among other factors. Table 1 outlines the inclusion and exclusion criteria for selecting studies included in this systematic review.
Table 1. Inclusion and Exclusion Criteria.
| Domain | Inclusion criteria | Exclusion criteria |
|---|---|---|
| Population | Licensed nurses (RNs, NPs, LPNs/LVNs) who provide direct patient care in any healthcare setting | Non-nurses, nursing students, nurses in non-clinical roles |
| Phenomena of Interest | Nurses’ perceptions of factors influencing their adoption of AI technologies in clinical practice | Nurses’ perceptions of non-AI technologies, quantitative measures of AI adoption |
| Design | Primary qualitative studies of any design, qualitative components of mixed methods studies | Quantitative studies, reviews, commentaries, editorials, conference abstracts |
| Evaluation | Studies that identify, describe, or interpret barriers and/or facilitators to AI adoption from nurses’ perspectives | Studies that do not report on barriers or facilitators, studies that only report researchers’ or other stakeholders’ perspectives |
| Research Type | Original research published in peer-reviewed journals or grey literature | Non-research publications such as opinion pieces, magazine articles, blog posts |
| Time Frame | No time restrictions; studies will be included from database inception to the search date | N/A |
| Language | English | Non-English |
AI, Artificial Intelligence ; LPNs, Licensed Practical Nurses; LVNs, Licensed Vocational Nurses; NPs, Nurse Practitioners; RNs, Registered Nurses.
Information sources and search strategy
A comprehensive search strategy will be developed in consultation with a health sciences librarian to identify potentially eligible studies. The following electronic databases will be searched from inception to January 2025: PubMed, Web of Science, Embase, CINAHL, MEDLINE, The Cochrane Library, PsycINFO and Scopus.
We will use a combination of controlled vocabulary terms (eg, MeSH and Emtree) and free-text keywords related to three main concepts: (1) AI, (2) nurses/nursing, and (3) barriers/facilitators to technology adoption. The search strategy will be adapted for each database and use appropriate truncation, wildcards and proximity operators to maximise sensitivity. A sample search strategy for PubMed is:
((artificial intelligence(MeSH Terms) OR artificial intelligence[Title/Abstract] OR machine learning[Title/Abstract] OR deep learning[Title/Abstract] OR natural language processing[Title/Abstract] OR computer vision[Title/Abstract] OR automated reasoning[Title/Abstract] OR intelligent agent[Title/Abstract] OR intelligent system[Title/Abstract] OR smart technology[Title/Abstract] OR virtual reality[Title/Abstract] OR augmented reality[Title/Abstract] OR robotics[Title/Abstract] OR automated decision support[Title/Abstract]) AND (nurses(MeSH Terms) OR nursing(MeSH Terms) OR nurses[Title/Abstract] OR nursing[Title/Abstract] OR nurse practitioner[Title/Abstract] OR nurse specialist[Title/Abstract] OR nurse leader[Title/Abstract] OR nurse manager[Title/Abstract] OR nurse educator[Title/Abstract] OR nurse informaticist[Title/Abstract] OR registered nurse[Title/Abstract] OR advanced practice nurse[Title/Abstract] OR licensed practical nurse[Title/Abstract]) AND (barrier*[Title/Abstract] OR challenge*[Title/Abstract] OR obstacle*[Title/Abstract] OR impediment*[Title/Abstract] OR facilitator*[Title/Abstract] OR enabler*[Title/Abstract] OR driver*[Title/Abstract] OR acceptance[Title/Abstract] OR acceptability[Title/Abstract] OR adoption[Title/Abstract] OR attitude*[Title/Abstract] OR perception*[Title/Abstract] OR perspective*[Title/Abstract] OR belief*[Title/Abstract] OR knowledge[Title/Abstract] OR skill*[Title/Abstract] OR training[Title/Abstract] OR education[Title/Abstract] OR self-efficacy[Title/Abstract] OR motivation[Title/Abstract] OR intention*[Title/Abstract] OR readiness[Title/Abstract] OR willingness[Title/Abstract] OR ability[Title/Abstract] OR capacity[Title/Abstract] OR capability[Title/Abstract] OR workflow[Title/Abstract] OR workload[Title/Abstract] OR professional role[Title/Abstract] OR scope of practice[Title/Abstract] OR organizational culture[Title/Abstract] OR leadership[Title/Abstract] OR strategy[Title/Abstract] OR resources[Title/Abstract] OR infrastructure[Title/Abstract] OR policy[Title/Abstract] OR ethics[Title/Abstract])). The complete search strategies for all databases are provided in online supplemental file 1. The search results will be limited to English language qualitative studies. We will also hand-search the reference lists of included studies and relevant reviews, as well as conduct targeted searches of key nursing informatics journals and conference proceedings, to identify additional eligible studies.
We acknowledge that limiting our search to English-language publications represents a significant methodological limitation that may introduce cultural and linguistic bias to our findings. This restriction potentially excludes valuable insights from AI implementation experiences in non-English speaking healthcare systems, which may differ substantially in their technological infrastructure, regulatory environments and cultural approaches to AI adoption. To partially mitigate this limitation, we will conduct targeted searches of key international nursing and health informatics journals that publish English-language abstracts of non-English articles, and we will identify and acknowledge in our discussion any major AI initiatives or research programmes in non-English speaking countries that our search strategy may have missed. Future research should prioritise multilingual systematic reviews to capture the full global landscape of nurses’ experiences with AI adoption.
Selection of studies
The results of the search will be imported into Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia) for screening. After removing duplicates, two reviewers will independently screen the titles and abstracts of all records against the eligibility criteria. Studies that appear to meet the criteria or where there is uncertainty will be retrieved in full text. Two reviewers will then independently assess the full-text articles for inclusion. Any disagreements will be resolved through discussion or arbitration by a third reviewer. Reasons for excluding studies at the full-text stage will be recorded. The study selection process will be documented in a PRISMA flow diagram (figure 1).44
Figure 1. Preferred Reporting Items for Systematic Review and Meta-Analysis Flow Diagram.
Data extraction
A standardised data extraction form will be developed in Covidence and piloted on a sample of included studies. Two reviewers will independently extract data from each study, with discrepancies resolved through discussion. The following data will be extracted:
Study details: authors, year of publication, country, aim/research question(s) and theoretical/conceptual framework
Methods: study design, setting, sampling strategy, participant characteristics, data collection method(s) and data analysis approach
AI technology: type of technology (eg, machine learning, natural language processing), intended use/application and stage of development/implementation)
Barriers/facilitators: description of perceived barriers and facilitators, level (eg, individual, team, organisation and technology), supporting quotes from participants, authors’ interpretations
Recommendations: implications of findings for research, practice, education, and policy
Where study data are unclear or missing, we will contact the authors for clarification or additional information (detailed data extraction form is provided in online supplemental Appendix 1).
Quality appraisal
The methodological quality of included studies will be critically appraised using the Critical Appraisal Skills Programme (CASP) Qualitative Studies Checklist.49 The CASP checklist includes 10 questions that assess different domains of quality in qualitative research, including appropriateness of qualitative methodology, data collection and analysis methods, consideration of researcher reflexivity and clarity of reporting. Two reviewers will independently apply the CASP checklist to each study, with disagreements resolved through discussion. The appraisal results will be used to inform data synthesis and interpretation, but studies will not be excluded based on quality.
The appraisal results will be used to inform data synthesis and interpretation in several specific ways. Studies with significant methodological limitations will be clearly identified, and their contributions to the final synthesised themes will be explicitly noted and contextualised. While studies will not be excluded based on quality alone, we will employ a differential weighting approach where higher-quality studies with robust methodological rigour will be given greater emphasis in developing overarching themes and theoretical insights. Studies with moderate to high risk of bias will contribute to the synthesis but their limitations will be explicitly acknowledged, and we will examine whether patterns differ between higher and lower quality studies. This approach ensures transparency about how study quality influences our confidence in specific findings while maximising the use of available evidence.
Risk of bias assessment
Risk of bias will be assessed using the CASP checklist,49 which includes criteria related to validity, reliability and objectivity of qualitative research. Specifically, the checklist assesses:
Appropriateness of qualitative methodology to address the research aim
Appropriateness of recruitment strategy to select participants
Rigour of data collection method(s) to address the research question(s)
Adequate consideration of the relationship between researcher and participants
Completeness and accuracy of data analysis to support the findings
Clarity and sufficiency of data presented to support the findings
Coherence between data, analysis and interpretation
Consideration of ethical issues and approval by an appropriate body
Two reviewers will independently assess the risk of bias of each included study, with disagreements resolved through discussion. Studies will be categorised as ‘low risk’, ‘moderate risk’ or ‘high risk’ based on the CASP criteria. The risk of bias assessments will be presented in a table and narrative summary in the full review.
Data synthesis
We will synthesise the data using the meta-ethnography approach.46,48 The synthesis will involve the following steps:
Read and re-read the included studies to become familiar with their content and contexts.
Determine how the studies are related by comparing their concepts, themes and interpretations.
Translate the studies into one another by identifying common and divergent concepts across studies and developing overarching concepts that encompass the meanings of individual studies.
Synthesise the translations by constructing a line-of-argument that integrates the concepts into a coherent theoretical framework that addresses the review questions.
Specifically, we will:
Extract key concepts, themes and participant quotes from each study and enter them into NVivo qualitative data analysis software (QSR International, Melbourne, Australia).
Compare and contrast the concepts across studies and group them into descriptive categories based on their similarities and differences.
Translate the concepts in each category into each other, maintaining their original meaning and context while developing overarching concepts that encompass the individual studies.
Synthesise the translations within and across categories into a line-of-argument that addresses the review questions and provide a new interpretative understanding of the phenomenon of nurses' perceptions of barriers and facilitators to AI adoption.
The synthesis will result in a theoretical framework that describes the relationships between the key concepts and illuminates the multilevel factors that influence nurses’ adoption of AI technologies in different clinical contexts. The framework will be illustrated with a concept map and narrative description. The quality and quantity of available qualitative evidence may limit the depth and breadth of synthesis. Meta-ethnography can accommodate studies of varying quality through conceptual synthesis rather than aggregation; however, if included studies lack theoretical depth, the explanatory power may be limited.
Two reviewers will independently conduct the initial synthesis, and the entire team will collaborate to refine the overarching concepts and line-of-argument through discussion and consensus. We will maintain a reflexive stance throughout the synthesis process by considering how our backgrounds, perspectives and assumptions may influence our interpretations.
Managing divergent and contradictory findings across studies will require systematic attention to context and methodology. When conflicting findings emerge, we will first examine whether these differences can be explained by variations in study context (eg, practice setting, healthcare system and AI technology type), participant characteristics (eg, experience level and specialty) or methodological approaches. We will use the constant comparative method to identify patterns in these divergences and will explicitly report both convergent and divergent themes in our findings. Where contradictions cannot be resolved through contextual analysis, we will present these as areas of uncertainty or debate within the field, highlighting the need for additional research. To address potential publication bias, we will examine whether our included studies represent a diverse range of study contexts, methodologies and findings, and will discuss any apparent gaps in the literature that might indicate selective reporting of positive or negative experiences with AI adoption.
Certainty of evidence
We will assess the certainty of the evidence for each review finding using the Grading of Recommendations Assessment, Development and Evaluation Confidence in the Evidence from Reviews of Qualitative Research (GRADE-CERQual) approach.50 51 CERQual provides a systematic and transparent framework for assessing how much confidence to place in individual review findings based on four components:
Methodological limitations of the studies contributing to the finding
Coherence of the finding across studies
Adequacy of data supporting the finding
Relevance of the studies to the context of the review question
For each review finding, we will assess the four CERQual components and make an overall judgement about the level of confidence in the finding (high, moderate, low or very low). The CERQual assessments will be conducted by two reviewers independently and then discussed with the full team to reach consensus. The CERQual evidence profiles and summary of findings will be presented in tables and narratives in the full review.
Ethics and dissemination
No ethical approval is required as this systematic review will synthesise data from published studies only. The findings will provide valuable insights to inform the development, implementation and evaluation of nurse-oriented strategies for AI integration in healthcare delivery. Results will be disseminated through peer-reviewed publication in a high-impact nursing or health informatics journal, presentations at relevant international conferences including the International Medical Informatics Association (IMIA) and American Medical Informatics Association (AMIA) meetings and stakeholder engagement activities with nursing professional organisations. We will also develop practical implementation toolkits based on our theoretical framework for different stakeholder groups, including technology developers, healthcare administrators and nursing educators.
Discussion
Significance
This systematic review will provide a comprehensive and rigorous synthesis of qualitative evidence on nurses’ perceived barriers and facilitators to adopting AI technologies in clinical practice. By consolidating evidence from multiple qualitative studies and interpreting their findings through a higher-level theoretical lens, this review can generate novel insights into the complex, context-dependent and multilevel factors that influence AI adoption in nursing.
Understanding nurses’ perceptions is critical for informing the development, implementation and evaluation of AI technologies that meet the needs of nurses and support their effective use at the point of care. As the largest healthcare profession, nurses play a central role in determining the success of AI initiatives in healthcare organisations.20 21 However, nurses’ perspectives have often been overlooked in the current discourse on AI in healthcare, which has focused more on the technological capabilities and potential impacts of AI rather than the human and organisational factors that shape its adoption.52 53
This review will address this gap by foregrounding nurses’ firsthand experiences and perceptions related to AI technologies in real-world clinical settings. By synthesising evidence on both the barriers and facilitators to AI adoption, the review can inform strategies to mitigate challenges and leverage enablers at multiple levels—from the individual nurse to the wider healthcare system. The review findings can guide efforts to design AI technologies that fit with nurses’ workflows and decision-making processes, prepare nurses with the necessary knowledge and skills to effectively use AI, create organisational environments that support AI implementation and develop policies and standards that ensure the ethical and equitable use of AI in nursing practice.
The meta-ethnography approach used in this review is particularly suited for synthesising qualitative evidence on complex social phenomena like technology adoption.47 By going beyond aggregation of findings to translation and interpretation of concepts across studies, meta-ethnography can yield new theoretical understandings that are greater than the sum of the individual studies. The line-of-argument synthesis can provide an explanatory framework that accounts for the variability and contextual nuances in nurses’ experiences with AI, while still identifying common themes and mechanisms that cut across different settings.
The review findings can also highlight areas where further research is needed to advance the knowledge on AI adoption in nursing. For example, the review may identify gaps in understanding the perspectives of certain subgroups of nurses (eg, by specialty, setting and demographics) or the influence of specific contextual factors (eg, organisational culture and implementation strategies). The review may also reveal methodological limitations or biases in the existing qualitative evidence that can inform future study designs and research questions.
Implications
Despite the limitations, this systematic review has significant implications for research, practice, education and policy related to AI in nursing and healthcare.
For researchers, the review can identify priorities and directions for future qualitative and quantitative studies on AI adoption in nursing. The review findings can inform the development of research questions and hypotheses, selection of theoretical frameworks and study designs, and identification of key variables and outcomes to measure. Researchers can also use the review findings to guide the development and testing of interventions to support AI adoption, such as educational programmes, organisational change strategies and user-centred design approaches.
For practitioners, the review can provide evidence-based insights into the factors that may hinder or enable their adoption of AI technologies in different clinical settings. The review can help nurses and other healthcare professionals to anticipate and address potential challenges, as well as to recognise and leverage facilitators to AI adoption. The review can also inform the selection, implementation and evaluation of AI technologies that are more likely to be accepted and effectively used by nurses in practice.
For educators, the review can guide the development of curricula and training programmes to prepare nurses for the use of AI in practice. The review can identify the knowledge, skills and attitudes that nurses need to effectively adopt and integrate AI into their clinical decision-making and care delivery. Educators can use the review findings to design educational interventions that target the individual and organisational barriers to AI adoption, as well as to foster the facilitators such as self-efficacy, motivation and readiness for change.
For policymakers, the review can inform the development of guidelines, standards and regulations related to the use of AI in nursing and healthcare. The review can highlight the ethical, legal and social implications of AI adoption that need to be addressed through policy and governance mechanisms. Policymakers can use the review findings to create policies and programmes that support the safe, effective and equitable use of AI in healthcare, while also protecting the rights and interests of patients, providers and communities.
Translation of findings into practice will be facilitated through the development of evidence-based implementation toolkits and educational resources. We will create practical guidance documents that translate our theoretical framework into actionable strategies for different stakeholder groups, such as (1) AI design recommendations for technology developers based on nurses’ usability and workflow integration needs; (2) organisational readiness assessment tools for healthcare institutions planning AI implementation; (3) competency-based training curricula for nursing education programmes and (4) policy briefs for healthcare administrators and policymakers. Each practical output will clearly link back to the evidence synthesis and specify the level of confidence in recommendations based on our GRADE-CERQual assessments. We will also establish partnerships with nursing professional organisations and healthcare technology companies to ensure our findings reach relevant implementation audiences.
Patient and public involvement
None
Data availability statement
This systematic review protocol does not use or reference any publicly available datasets. The protocol describes the methodology for a future systematic review that will synthesise qualitative evidence from published studies identified through systematic searches of electronic databases. No primary data collection, analysis of existing datasets or dataset citations are involved in this protocol paper. All references cited in this protocol refer to published literature (journal articles, books and guidelines) rather than datasets. The systematic review itself, when conducted, will extract and synthesise data from published qualitative studies, with all included studies to be properly cited according to standard academic referencing practices in the final review manuscript.
Supplementary material
Footnotes
Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2025-099875).
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent for publication: Not applicable.
Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
References
- 1.Ahmed SK. Artificial intelligence in nursing: Current trends, possibilities and pitfalls. J Med Surg Public Health . 2024;3:100072. doi: 10.1016/j.glmedi.2024.100072. [DOI] [Google Scholar]
- 2.Buchanan C, Howitt ML, Wilson R, et al. Predicted Influences of Artificial Intelligence on Nursing Education: Scoping Review. JMIR Nursing . 2021;4:e23933. doi: 10.2196/23933. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Wahl B, Cossy-Gantner A, Germann S, et al. Artificial intelligence (AI) and global health: how can AI contribute to health in resource-poor settings? BMJ Glob Health. 2018;3:e000798. doi: 10.1136/bmjgh-2018-000798. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Robert N. How artificial intelligence is changing nursing. Nurs Manage. 2019;50:30–9. doi: 10.1097/01.NUMA.0000578988.56622.21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Topaz M, Pruinelli L. Big Data and Nursing: Implications for the Future. Stud Health Technol Inform. 2017;232:165–71. [PubMed] [Google Scholar]
- 6.Thilo FJ, Bilger S, Halfens RJ, et al. Involvement of the end user: exploration of older people’s needs and preferences for a wearable fall detection device - a qualitative descriptive study. Patient Prefer Adherence. 2017;11:11–22. doi: 10.2147/PPA.S119177. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Clancy TR. Artificial Intelligence and Nursing: The Future Is Now. J Nurs Adm. 2020;50:125–7. doi: 10.1097/NNA.0000000000000855. [DOI] [PubMed] [Google Scholar]
- 8.Elhaddad M, Hamam S. AI-Driven Clinical Decision Support Systems: An Ongoing Pursuit of Potential. Cureus. 2024;16:e57728. doi: 10.7759/cureus.57728. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Zafer A, Graepel T. Transforming patient care with AI: leveraging electronic health records for personalized healthcare and proactive care. 2024
- 10.Machon M, Cundy D, Case H. Innovation in Nursing Leadership: A Skill That Can Be Learned. Nurs Adm Q. 2019;43:267–73. doi: 10.1097/NAQ.0000000000000361. [DOI] [PubMed] [Google Scholar]
- 11.Robinson R, Liday C, Lee S, et al. Artificial Intelligence in Health Care-Understanding Patient Information Needs and Designing Comprehensible Transparency: Qualitative Study. JMIR AI . 2023;2:e46487. doi: 10.2196/46487. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Naik N, Hameed BMZ, Shetty DK, et al. Legal and Ethical Consideration in Artificial Intelligence in Healthcare: Who Takes Responsibility? Front Surg. 2022;9:862322. doi: 10.3389/fsurg.2022.862322. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kim JP, Yang H-J, Kim B, et al. Understanding Physician’s Perspectives on AI in Health Care: Protocol for a Sequential Multiple Assignment Randomized Vignette Study. JMIR Res Protoc. 2024;13:e54787. doi: 10.2196/54787. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Davenport T, Kalakota R. The potential for artificial intelligence in healthcare. Future Healthc J . 2019;6:94–8. doi: 10.7861/futurehosp.6-2-94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hager G. Artificial intelligence for social good. 2019
- 16.Laï M-C, Brian M, Mamzer MF. Perceptions of artificial intelligence in healthcare: findings from a qualitative survey study among actors in France. J Transl Med. 2020;18:14. doi: 10.1186/s12967-019-02204-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Visco V, Izzo C, Mancusi C, et al. Artificial Intelligence in Hypertension Management: An Ace up Your Sleeve. J Cardiovasc Dev Dis. 2023;10:74. doi: 10.3390/jcdd10020074. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Singhal A, Neveditsin N, Tanveer H, et al. Toward Fairness, Accountability, Transparency, and Ethics in AI for Social Media and Health Care: Scoping Review. JMIR Med Inform. 2024;12:e50048. doi: 10.2196/50048. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lysaght T, Lim HY, Xafis V, et al. AI-Assisted Decision-making in Healthcare: The Application of an Ethics Framework for Big Data in Health and Research. Asian Bioeth Rev. 2019;11:299–314. doi: 10.1007/s41649-019-00096-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Risling T. Educating the nurses of 2025: Technology trends of the next decade. Nurse Educ Pract. 2017;22:89–92. doi: 10.1016/j.nepr.2016.12.007. [DOI] [PubMed] [Google Scholar]
- 21.McGrow K. Artificial intelligence: Essentials for nursing. Nursing (Auckl) 2019;49:46–9. doi: 10.1097/01.NURSE.0000577716.57052.8d. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Pepito JA, Locsin R. Can nurses remain relevant in a technologically advanced future? Int J Nurs Sci . 2019;6:106–10. doi: 10.1016/j.ijnss.2018.09.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Gutierrez M, Moreno R, Rebelo M. Information and communication technologies and global health challenges. 2017. pp. 50–93. [Google Scholar]
- 24.Gagnon M-P, Desmartis M, Labrecque M, et al. Systematic review of factors influencing the adoption of information and communication technologies by healthcare professionals. J Med Syst. 2012;36:241–77. doi: 10.1007/s10916-010-9473-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.McGinn CA, Grenier S, Duplantie J, et al. Comparison of user groups’ perspectives of barriers and facilitators to implementing electronic health records: a systematic review. BMC Med. 2011;9:46. doi: 10.1186/1741-7015-9-46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Khairat S, Coleman C, Ottmar P, et al. Association of Electronic Health Record Use With Physician Fatigue and Efficiency. JAMA Netw Open. 2020;3:e207385. doi: 10.1001/jamanetworkopen.2020.7385. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Strudwick G, McGillis Hall L, Nagle L, et al. Acute care nurses’ perceptions of electronic health record use: A mixed method study. Nurs Open. 2018;5:491–500. doi: 10.1002/nop2.157. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Whitelaw S, Pellegrini DM, Mamas MA, et al. Barriers and facilitators of the uptake of digital health technology in cardiovascular care: a systematic scoping review. Eur Heart J Digit Health . 2021;2:62–74. doi: 10.1093/ehjdh/ztab005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Seibert K, Domhoff D, Bruch D, et al. Application Scenarios for Artificial Intelligence in Nursing Care: Rapid Review. J Med Internet Res . 2021;23:e26522. doi: 10.2196/26522. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Laranjo L, Dunn AG, Tong HL, et al. Conversational agents in healthcare: a systematic review. J Am Med Inform Assoc. 2018;25:1248–58. doi: 10.1093/jamia/ocy072. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Nundy S, Montgomery T, Wachter RM. Promoting Trust Between Patients and Physicians in the Era of Artificial Intelligence. JAMA. 2019;322:497–8. doi: 10.1001/jama.2018.20563. [DOI] [PubMed] [Google Scholar]
- 32.Rudin C. Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead. Nat Mach Intell . 2019;1:206–15. doi: 10.1038/s42256-019-0048-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Gephart SM, Davis M, Shea K. Perspectives on Policy and the Value of Nursing Science in a Big Data Era. Nurs Sci Q. 2018;31:78–81. doi: 10.1177/0894318417741122. [DOI] [PubMed] [Google Scholar]
- 34.Kapoor A, Guha S, Kanti Das M, et al. Digital healthcare: The only solution for better healthcare during COVID-19 pandemic? Indian Heart J Teach Ser. 2020;72:61–4. doi: 10.1016/j.ihj.2020.04.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Creswell JW, Poth CN. Qualitative inquiry and research design: choosing among five approaches. Sage publications; 2016. [Google Scholar]
- 36.Lambert SI, Madi M, Sopka S, et al. An integrative review on the acceptance of artificial intelligence among healthcare professionals in hospitals. NPJ Digit Med. 2023;6:111. doi: 10.1038/s41746-023-00852-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Borges do Nascimento IJ, Abdulazeem H, Vasanthan LT, et al. Barriers and facilitators to utilizing digital health technologies by healthcare professionals. NPJ Digit Med. 2023;6:161. doi: 10.1038/s41746-023-00899-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Ramadan OME, Alruwaili MM, Alruwaili AN, et al. Facilitators and barriers to AI adoption in nursing practice: a qualitative study of registered nurses’ perspectives. BMC Nurs. 2024;23:891. doi: 10.1186/s12912-024-02571-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Ma Y, Liu T, Qi J, et al. Facilitators and Barriers of Large Language Model Adoption Among Nursing Students: A Qualitative Descriptive Study. J Adv Nurs. 2025;81:4856–70. doi: 10.1111/jan.16655. [DOI] [PubMed] [Google Scholar]
- 40.Irgang L, Sestino A, Barth H, et al. Healthcare workers’ adoption of and satisfaction with artificial intelligence: The counterintuitive role of paradoxical tensions and paradoxical mindset. Technol Forecast Soc Change. 2025;212:123967. doi: 10.1016/j.techfore.2024.123967. [DOI] [Google Scholar]
- 41.Gholampour MH, Zare-Kaseb A, Arbabi Z, et al. Implementation of artificial intelligence in the nursing profession: Barriers & facilitators. JNRCP. 2023;2:47–8. doi: 10.32598/JNRCP.23.25. [DOI] [Google Scholar]
- 42.Boo S, Oh H. Perceptions of registered nurses on facilitators and barriers of implementing the AI-IoT-based healthcare pilot project for older adults during the COVID-19 pandemic in South Korea. Front Public Health. 2023;11:11. doi: 10.3389/fpubh.2023.1234626. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Tong A, Flemming K, McInnes E, et al. Enhancing transparency in reporting the synthesis of qualitative research: ENTREQ. BMC Med Res Methodol. 2012;12:181. doi: 10.1186/1471-2288-12-181. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Page MJ, Moher D, Bossuyt PM, et al. PRISMA 2020 explanation and elaboration: updated guidance and exemplars for reporting systematic reviews. BMJ. 2021;372:n160. doi: 10.1136/bmj.n160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Shamseer L, Moher D, Clarke M, et al. Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation. BMJ. 2015;349:g7647. doi: 10.1136/bmj.g7647. [DOI] [PubMed] [Google Scholar]
- 46.Noblit GW. Meta-ethnography: synthesizing qualitative studies. Stage Publication; 1988. [Google Scholar]
- 47.France EF, Cunningham M, Ring N, et al. Improving reporting of meta-ethnography: The eMERGe reporting guidance. J Adv Nurs. 2019;75:1126–39. doi: 10.1111/jan.13809. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.France EF, Ring N, Thomas R, et al. A methodological systematic review of what’s wrong with meta-ethnography reporting. BMC Med Res Methodol. 2014;14:119. doi: 10.1186/1471-2288-14-119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Programme, C A S CASP qualitative studies checklist. 2018
- 50.Lewin S, Bohren M, Rashidian A, et al. Applying GRADE-CERQual to qualitative evidence synthesis findings—paper 2: how to make an overall CERQual assessment of confidence and create a Summary of Qualitative Findings table. Implementation Sci. 2018;13:10. doi: 10.1186/s13012-017-0689-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Lewin S, Glenton C, Munthe-Kaas H, et al. Using qualitative evidence in decision making for health and social interventions: an approach to assess confidence in findings from qualitative evidence syntheses (GRADE-CERQual) PLoS Med. 2015;12:e1001895. doi: 10.1371/journal.pmed.1001895. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Chung J, Cho I. The need for academic electronic health record systems in nurse education. Nurse Educ Today. 2017;54:83–8. doi: 10.1016/j.nedt.2017.04.018. [DOI] [PubMed] [Google Scholar]
- 53.Nagle L, Kleib M, Furlong K. Digital Health in Canadian Schools of Nursing—Part B: Academic Nurse Administrators’ Perspectives. QANE-AFI. 2020;6:6. doi: 10.17483/2368-6669.1256. [DOI] [Google Scholar]