Quality of Nursing Care and Adaptation to Artificial Intelligence in Low‑ and Middle‑Income Countries: A Systematic Review of Empirical Studies

Quality of Nursing Care and Adaptation to Artificial Intelligence in Low‑ and Middle‑Income Countries: A Systematic Review of Empirical Studies | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Systematic Review Quality of Nursing Care and Adaptation to Artificial Intelligence in Low‑ and Middle‑Income Countries: A Systematic Review of Empirical Studies DR. FERNAN, FRINCESS This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8704137/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Nurses in low- and middle-income countries (LMICs) deliver care under chronic resource constraints, which compromise quality and shape how new technologies, including artificial intelligence (AI), can be adopted. Evidence on how nursing care quality and nurses’ adaptation to AI intersect in LMICs remains fragmented. Methods We conducted a systematic review of empirical and review studies published between 2022 and 2025 that examined (1) the quality of nursing care in LMICs and/or (2) nurses’ perceptions, readiness, and adaptation to AI in clinical practice. Searches of major databases identified ten eligible studies, including systematic and integrative reviews, scoping reviews, cross-sectional surveys, and conceptual frameworks addressing nursing care quality, AI in nursing, or AI implementation in LMIC health systems. Data were synthesized narratively across two domains: nursing care quality and AI-related attitudes and adaptation. Results Studies on care quality reported high levels of missed or delayed nursing care in LMIC acute and critical care settings, driven by chronic understaffing, inadequate skill mix, limited supplies, and weak governance and quality systems; staffing–outcome research in LMICs was sparse and methodologically heterogeneous. AI-focused studies showed nurses were cautiously open to AI’s potential for efficiency, documentation, and decision support, yet concerned about job security, role erosion, liability, data privacy, and loss of human touch. Attitudes and readiness were influenced by emotion regulation, cognitive flexibility, and digital literacy, while implementation in LMICs was constrained by unreliable infrastructure, immature data systems, and limited technical support. Conceptual frameworks proposed baseline AI competencies for all nurses but emphasized phased, context-sensitive implementation and strong governance. Conclusions Nurses in LMICs are attempting to adapt to AI while fundamental deficits in nursing workforce and care environments remain unresolved. Strengthening staffing and basic quality infrastructure, embedding AI literacy in nursing education, involving nurses in digital-health planning, and establishing clear policies on data protection and accountability are essential to ensure that AI augments rather than undermines nursing care quality in LMICs. Artificial Intelligence and Machine Learning Nursing Hospital Medicine Health Economics & Outcomes Research Health Policy Health Law nursing care quality low- and middle-income countries artificial intelligence nurse adaptation digital health missed nursing care Figures Figure 1 Figure 2 INTRODUCTION Nursing care sits at the core of health system performance, yet its quality remains highly uneven across the world.¹⁻³ In low- and middle-income countries (LMICs), nurses frequently deliver care in chronically under-resourced environments marked by staff shortages, high patient acuity, and limited supplies.⁴⁻⁶ In such settings, the promise of artificial intelligence (AI) to support clinical decision-making, streamline documentation, and extend scarce expertise is attracting growing attention.⁷⁻¹⁰ However, the same structural constraints that compromise fundamental nursing care may also impede the safe and equitable adoption of AI tools in everyday practice.⁴⁻⁶,¹¹,¹² Understanding how nursing care quality and nurses’ adaptation to AI intersect in LMICs is therefore essential for planning realistic, context-sensitive digital transformation strategies.⁴⁻⁶,¹¹⁻¹³[ Quality of nursing care in LMICs: convergence and divergence with high-income settings Over the last decade, research on nursing care quality has increasingly focused on the construct of “missed nursing care” or “care left undone”—that is, required nursing tasks that are omitted or significantly delayed.⁴,¹⁴ A recent systematic review of missed nursing care in acute care hospitals in LMICs synthesized evidence from observational studies in Africa, Asia, and Latin America and found that omissions in fundamental care—such as patient education, emotional support, documentation, and timely monitoring—are both frequent and patterned.⁴ Tasks perceived as less urgent but essential for holistic care were most likely to be missed, with reported prevalence comparable to or higher than that in high-income settings.¹⁴,¹⁵ The drivers of missed care in LMIC hospitals are multifactorial but consistently include inadequate nurse staffing, high patient-to-nurse ratios, insufficient skill mix, and resource constraints such as shortages of medications, basic equipment, and personal protective equipment.⁴⁻⁶,¹⁴ An integrative review of nursing care quality in public hospitals in LMICs identified five broad determinants: nurse staffing and skill mix, work environment and leadership, availability of supplies, opportunities for continuing education, and the presence of quality improvement and governance systems.⁵ Across studies, better staffing, supportive leadership, and reliable availability of basic resources were associated with higher perceived quality of care and patient satisfaction, whereas resource scarcity and poor management were linked to missed care, moral distress, and burnout.⁵,¹⁶ Yet despite growing recognition of these issues, robust empirical data on nurse staffing and patient outcomes in LMICs remain sparse.⁴⁻⁶ An umbrella review examining global evidence on nurse staffing and patient care outcomes highlighted that most high-quality studies originate from high-income countries, with only a small subset from LMICs.⁶ The LMIC evidence base was characterized by heterogeneous designs, variable measurement of staffing and outcomes, and small sample sizes, hampering meta-analysis and limiting precise estimates of effect sizes.⁶ Nevertheless, the available data indicate similar patterns to those seen in high-income settings: higher workloads and lower staffing correlate with increased mortality, complications, and lower quality ratings, suggesting that strengthening nurse staffing is likely to yield substantial gains in care quality.⁴⁻⁶,⁸ Critical care settings illustrate the intensity of these challenges. A scoping review of critical care nursing in low- and lower-middle-income countries described overcrowded intensive care units, limited access to mechanical ventilation and monitoring equipment, and severe shortages of nurses with formal critical care training, leading to very high patient-to-nurse ratios.¹⁶ Under such conditions, nurses are forced to prioritize life-saving interventions at the expense of comprehensive assessment, communication, and comfort measures, reinforcing patterns of missed care and moral distress.¹⁶ These findings underscore that in many LMIC contexts, efforts to improve nursing care quality must contend simultaneously with workforce deficits, infrastructural gaps, and fragile supply chains.⁴⁻⁶,¹⁶ At the policy level, global reports such as the State of the World’s Nursing have called attention to the concentration of nursing shortages in LMICs and the need for national strategies to invest in nursing education, regulation, and workforce planning.⁸,⁹ However, many health systems still lack routine data on nursing care processes and outcomes, limiting their ability to monitor quality and evaluate interventions.⁶,⁸,⁹ This data gap has implications not only for traditional quality-improvement initiatives but also for the deployment and evaluation of AI tools, which depend on high-quality digital data to function effectively.³,⁸,¹¹ The rise of AI in nursing and health care In parallel with concerns about nursing care quality, there has been rapid growth in the development and deployment of AI applications in health care, including decision support for diagnosis and treatment, predictive risk models, automated image analysis, and conversational agents.¹⁰,¹¹ Within nursing, AI-enabled tools have been proposed to assist with tasks such as early warning score calculation, workload prediction, documentation, triage, and patient education.⁷,¹⁰ A recent systematic review of AI in nursing described a wide range of applications, from machine-learning algorithms for predicting pressure injuries and falls to natural language processing systems for extracting information from nursing notes.⁷ However, the same review highlighted that most AI-nursing research has been conducted in high-income settings, often using large electronic health record datasets that are not readily available in many LMICs.⁷ Moreover, many AI tools remain at the proof-of-concept or pilot stage, with limited evaluation of their impact on nursing workflows, care quality, or patient outcomes in real-world practice.⁷,¹⁰ As a result, questions remain about how AI might realistically augment nursing care in resource-constrained environments and what preconditions are necessary for safe and effective integration.¹¹,¹² Nurses’ attitudes, readiness, and adaptation to AI Nurses’ perceptions and experiences are increasingly recognized as critical determinants of whether AI tools are adopted, adapted, or resisted in practice.¹⁰⁻¹³ A growing body of survey research has explored nurses’ attitudes towards AI, their intention to use AI-based systems, and the psychological and organizational factors that influence these attitudes.¹¹⁻¹³ Overall, these studies suggest that nurses hold cautiously positive views: they acknowledge potential benefits of AI for efficiency, accuracy, and workload reduction but also express concerns about job security, professional identity, liability, data privacy, and the risk of de-humanizing care.¹¹,¹² For example, a cross-sectional study of hospital nurses reported moderate acceptance of AI, with higher acceptance associated with greater perceived usefulness and ease of use, and lower acceptance associated with fear of replacement and ethical concerns.¹¹ Another study examining the relationship between nurses’ attitudes to AI and psychological characteristics found that more positive attitudes were correlated with higher cognitive flexibility and better emotion regulation, suggesting that individual adaptation may depend partly on resilience and openness to innovation.¹² Importantly, these studies also noted substantial gaps in nurses’ knowledge about what AI is, how it works, and its limitations, particularly among older and less digitally experienced staff.¹¹⁻¹³ Evidence specific to LMICs is more limited but points to additional, context-specific challenges. Analyses of AI implementation in low-resource health systems have documented infrastructural barriers such as unreliable electricity, poor internet connectivity, limited access to hardware, and fragmented or non-existent electronic health records.¹⁵,¹٦ In such environments, even relatively simple AI-enabled tools—such as decision-support apps or triage chatbots—may be difficult to use consistently.¹⁵,¹⁶ Furthermore, where digital systems do exist, they may not be interoperable, standardized, or representative, complicating the development of reliable algorithms and raising concerns about bias and validity.¹⁴,¹⁶ For nurses, these system-level constraints intersect with workload pressures and staffing shortages.¹⁵ In LMIC settings where nurses are already stretched thin, the introduction of new digital tools can be perceived as an additional burden if systems are slow, poorly designed, or inadequately supported.¹⁵ Studies of digital health interventions in LMICs have shown that health workers are often required to double-enter data in both paper and electronic systems during transition periods, further increasing workload and frustration.¹⁶ Initial enthusiasm for AI may therefore erode if tools do not clearly save time, improve care, or align with local workflows.¹¹,¹⁵,¹⁶ Why focus on quality of nursing care and AI in LMICs together? Despite the parallel growth of research on nursing care quality and on AI in health care, these literatures have rarely been brought together, particularly in relation to LMIC contexts.⁴⁻⁷ Most reviews of nursing care quality in LMICs focus on traditional determinants such as staffing, workload, skill mix, and resource availability, with little consideration of digital technologies or AI.⁴⁻⁶,¹⁶ Conversely, reviews of AI in nursing tend to emphasize technical performance, ethical issues, and attitudes in high-income settings, without examining how fundamental quality deficits and infrastructural constraints shape AI adoption in LMIC nursing practice.⁷,¹⁰,¹¹ Yet in many LMIC health systems, nurses are being asked to engage with AI-enabled tools at the same time as they are struggling to deliver basic care under conditions of chronic understaffing and inadequate resources.⁴⁻⁶,⁸,¹⁵ This raises critical questions about whether AI can realistically help improve nursing care quality in LMICs when core workforce and infrastructural issues remain unresolved, or whether it risks adding complexity and workload without commensurate benefit.⁴⁻⁶,⁷,¹⁵,¹⁶ It also prompts inquiry into how nurses in these settings perceive and cope with AI, what strategies they use to integrate or resist new technologies while maintaining patient-centred care, and under what conditions AI might augment rather than undermine nursing autonomy, professional identity, and therapeutic relationships.⁷,¹¹,¹⁴,¹⁸ Emerging conceptual work has begun to grapple with these questions. A recent framework titled “Every nurse an AI nurse” argues that all nurses will eventually need baseline competencies in understanding, querying, and supervising AI systems, not just those in specialized informatics roles.¹⁸ However, the authors stress that implementation must be tailored to local context: in under-resourced or data-poor settings, early priorities may need to include building basic digital infrastructure, ensuring data quality, and integrating AI literacy into pre-service and in-service education, alongside traditional investments in staffing and equipment.¹⁶,¹⁸ Without such groundwork, there is a risk that AI will exacerbate existing inequities, benefiting well-resourced institutions while leaving nurses in LMICs further behind.¹⁴⁻¹⁶,¹⁸ Aim and contribution of this review Against this backdrop, the present systematic review has two primary aims. First, it seeks to synthesize recent empirical and review evidence on the quality of nursing care in LMICs, with particular attention to missed nursing care, staffing, and work environment determinants.⁴⁻⁶,¹⁴⁻¹⁶ Second, it examines how nurses perceive, cope with, and adapt to AI technologies in clinical practice, drawing on studies of attitudes, readiness, and implementation experiences, including those from LMICs and global frameworks with explicit relevance for low-resource settings.⁷,¹⁰⁻¹³,¹⁵⁻¹⁸ By integrating these two strands of evidence, the review aims to address a critical gap in the literature: the lack of a consolidated, nursing-focused perspective on how structural determinants of care quality in LMICs intersect with opportunities and risks associated with AI adoption.⁴⁻⁷,¹⁴⁻¹⁸ In doing so, it seeks to move beyond abstract debates about “AI replacing nurses” towards a more grounded understanding of how AI might realistically support or hinder nurses working in some of the world’s most constrained health systems.⁴⁻⁷,¹¹,¹⁴⁻¹⁸ For policymakers and nurse leaders, this synthesis offers a basis for designing context-sensitive strategies that prioritize foundational investments in the nursing workforce and care environment while planning for staged, nurse-centred integration of AI tools.⁴⁻⁶,⁸,¹⁵⁻¹⁸ For researchers, it identifies key gaps—such as the paucity of LMIC-specific evaluations of AI’s impact on nursing processes and patient outcomes—and suggests priorities for future work.⁴⁻⁷,¹⁰⁻¹³,¹⁴⁻¹⁸ Ultimately, the goal is to ensure that discussions about AI in nursing are informed by, and accountable to, the realities of nursing practice in LMICs, where the stakes for both quality and equity are particularly high.⁴⁻⁸,¹¹,¹⁴⁻¹⁸ METHODS Review design and reporting We conducted a systematic review of quantitative and mixed-methods studies examining (1) quality of nursing care in low- and middle-income countries (LMICs) and (2) nurses’ perceptions, readiness, and adaptation to artificial intelligence (AI) technologies in clinical practice. The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement for planning and reporting.¹⁻³ We used the PRISMA 2020 checklist and flow diagram to structure reporting of search, selection, and synthesis procedures.²⁻⁴ The protocol was developed a priori in line with guidance for systematic reviews of health interventions and AI applications in nursing and LMIC settings.⁵⁻⁸ It was designed to be compatible with Research Square and other preprint platforms’ expectations for transparency and reproducibility.⁹ Eligibility criteria We defined inclusion and exclusion criteria using a Population–Exposure–Outcome–Study design (PEOS) framework. Population: Registered or licensed nurses and other healthcare workers delivering direct clinical care in LMICs, as defined by World Bank income classifications at the time of each study. We also included global or mixed-setting studies if they reported LMIC-specific subgroup analyses or presented findings clearly applicable to LMIC contexts.⁵⁻⁷,¹⁰ Exposures: Indicators of nursing care quality (for example, missed nursing care, care left undone, nurse-sensitive quality indicators, patient-reported care quality, nurse staffing and work environment variables).⁵⁻⁷,¹¹ AI-related exposures, including the presence or use of AI-enabled tools (decision support, predictive models, chatbots, documentation systems) and nurses’ attitudes, perceptions, readiness, or intention to use AI.⁸,¹²⁻¹⁵ Outcomes: For quality-of-care studies: prevalence and types of missed nursing care, nurse-sensitive patient outcomes (for example, mortality, falls, pressure injuries), patient satisfaction, and composite indices of care quality.⁵⁻⁷,¹¹ For AI-related studies: nurses’ attitudes to AI, perceived usefulness and ease of use, readiness or intention to adopt AI tools, reported use of AI in practice, and perceived impact on workload, care processes, or patient outcomes.⁸,¹²⁻¹⁵ Study designs: We included systematic reviews and meta-analyses, cross-sectional and cohort studies, quasi-experimental evaluations, and mixed-methods designs that reported quantitative data on at least one of the above outcomes.⁵⁻⁸ Qualitative studies and purely conceptual papers were used to contextualize findings but were not part of the primary quantitative synthesis.⁸,¹⁶ Time frame and language: We restricted inclusion to studies published between 1 January 2022 and 31 December 2025 to capture the most recent post-pandemic evidence and rapid developments in AI applications in nursing. Only articles published in English were included due to resource constraints. Exclusion criteria: We excluded studies conducted exclusively in high-income countries without LMIC-specific analyses; studies focusing solely on non-nursing cadres without mixed healthcare-worker data; editorials, commentaries, and conference abstracts without full text; and technical AI performance studies with no nursing outcomes or workforce-focused data.⁸,¹⁵,¹٧ Information sources and search strategy A health sciences librarian with experience in systematic reviews and digital health supported the development of the search strategy. We searched PubMed/MEDLINE, CINAHL, Scopus, Web of Science, and the Cochrane Library from 1 January 2022 to 31 December 2025. Search strategies combined controlled vocabulary (for example, MeSH and CINAHL Subject Headings) and free-text terms related to nursing, LMICs, quality of care, and AI. Example search concepts included: (“nurse*” OR “nursing care” OR “nurse staffing”) AND (“low-income countr*” OR “middle-income country*” OR “LMIC*”) AND (“missed care” OR “care left undone” OR “quality of care” OR “patient outcome*” OR “mortality” OR “falls” OR “pressure ulcer*”).⁵⁻⁷,¹¹ (“nurse*” OR “nursing”) AND (“artificial intelligence” OR “machine learning” OR “clinical decision support” OR “chatbot*” OR “predictive model*”) AND (“attitude*” OR “perception*” OR “readiness” OR “intention to use” OR “adoption”).⁸,¹²⁻¹⁵ Search strings were iteratively refined based on preliminary results and informed by prior systematic reviews and protocols on nurse staffing, quality of care, and AI in nursing.⁵⁻⁸,¹¹,¹⁸ We also screened reference lists of included reviews and key primary studies and conducted forward citation tracking using Scopus and Google Scholar to identify additional relevant articles.⁵,⁷,⁸ Searches of grey literature were limited to major global reports that provided contextual data on nursing workforce and digital health in LMICs.¹⁹⁻²¹ Study selection All records retrieved from the searches were imported into a reference management software, where duplicates were removed. Two reviewers independently screened titles and abstracts against the eligibility criteria. Articles deemed potentially relevant by either reviewer proceeded to full-text screening. Full texts were assessed independently by the same two reviewers, with disagreements resolved through discussion or, when needed, consultation with a third reviewer, in line with PRISMA 2020 recommendations.¹⁻³ Reasons for exclusion at the full-text stage (for example, wrong population, no LMIC data, no nursing-related outcomes, purely technical AI evaluation) were recorded. The overall selection process, including numbers of records identified, screened, excluded, and included, is presented in a PRISMA 2020 flow diagram.²⁻⁴ Data extraction We developed a standardized data-extraction form based on prior reviews of nursing care quality and AI in nursing.⁵⁻⁸,¹¹,¹⁸ Two reviewers independently piloted the form on a subset of eligible studies and refined it to ensure clarity and consistency. The final extraction form captured: Bibliographic details: first author, year, country, journal. Study design: systematic review, cross-sectional, cohort, quasi-experimental, mixed-methods. Setting and population: country income classification, clinical setting (hospital, primary care, community, critical care), profession (nurses only versus mixed healthcare workers), sample size, and any explicit focus on LMIC or subgroup analyses.⁵⁻⁷,¹⁰,¹¹,¹⁶ Exposure variables: For quality-of-care studies: measures of nurse staffing (for example, nurse-to-patient ratio, hours per patient day), work environment (for example, practice-environment scales), and quality metrics such as missed-nursing-care instruments and patient satisfaction scales.⁵⁻⁷,¹¹,¹⁸ For AI-related studies: type of AI tool (for example, decision support, predictive model, conversational agent), implementation context, and measures of nurses’ attitudes, perceived usefulness, readiness, and intention to use.⁸,¹²⁻¹⁵,¹⁶ Outcomes: nursing-care quality indicators, nurse-sensitive patient outcomes, and AI-related outcomes (for example, acceptance scores, intention-to-use scales, reported impact on workload and care processes).⁵⁻⁸,¹¹,¹²,¹⁵ Key results: effect estimates (for example, odds ratios, correlation coefficients, mean differences) with confidence intervals or p-values; descriptive statistics for cross-sectional outcomes; and, in mixed-methods studies, qualitative themes relevant to nurses’ adaptation to AI.⁵⁻⁸,¹⁶ Discrepancies in extracted data were resolved through discussion and re-examination of the original articles, with arbitration by a third reviewer if required. When critical information was missing or unclear, we attempted to contact corresponding authors. Risk of bias and quality assessment Because the review included a mix of systematic reviews, observational studies, and quasi-experimental designs, we used design-appropriate tools to assess methodological quality. For included systematic reviews and meta-analyses, we used AMSTAR-2 to appraise methodological quality.⁵,²² For observational cohort and cross-sectional studies, we used relevant Joanna Briggs Institute (JBI) critical-appraisal checklists, which assess sampling, measurement validity and reliability, confounding, and statistical analysis.²³,²⁴ For quasi-experimental studies, we used JBI appraisal tools for non-randomized experimental designs, focusing on allocation methods, baseline comparability, handling of confounders, and completeness of follow-up.²³ Two reviewers independently assessed each study, rating individual domains and overall risk of bias as low, moderate, or high. Disagreements were resolved by consensus. Risk-of-bias assessments informed interpretation and, where appropriate, sensitivity analyses but were not used as exclusion criteria except for clearly fatally flawed studies (for example, major outcome misclassification or extreme data inconsistencies).¹,²³,²⁴ Data synthesis Given anticipated heterogeneity in populations, settings, exposures, and outcomes, we planned a primarily narrative synthesis, with quantitative pooling considered only where studies were sufficiently homogeneous in design and measurement.¹,³,⁸,¹⁸ We structured the synthesis around two main domains: Quality of nursing care in LMICs, focusing on missed nursing care, nurse staffing and work environment, and nurse-sensitive patient outcomes.⁵⁻⁷,¹¹,¹⁸ Nurses’ perceptions, readiness, and adaptation to AI, including attitudes, acceptance, intention to use, and reported impact on care processes.⁸,¹²⁻¹⁶ Within each domain, we grouped studies by design and type of exposure, then compared findings across settings and income contexts. When multiple studies reported similar associations using comparable measures (for example, between nurse staffing and missed care, or between perceived usefulness and intention to use AI), we considered performing random-effects meta-analysis following established methods for nurse-staffing and AI reviews.⁵,⁸,¹⁸,²⁵ Where heterogeneity in design, measures, or reporting precluded formal pooling, we followed guidance on “synthesis without meta-analysis” as outlined in the PRISMA 2020 elaboration paper.³ We also qualitatively integrated insights from conceptual and scoping reviews on AI in LMIC health systems to contextualize empirical findings on nurses’ attitudes and adaptation.¹⁶,¹⁷,²⁶,²⁷ Where appropriate, we contrasted findings from LMIC-focused studies with those from global or high-income-dominated samples to highlight potential differences in determinants of care quality and AI adoption. Because this review synthesized data from published studies and did not involve direct contact with human participants, formal ethics approval was not required.⁵,⁸,¹⁶ RESULTS Ten studies met the inclusion criteria and were synthesized across two domains: (1) quality of nursing care in LMICs and (2) nurses’ attitudes, readiness, and adaptation to AI in clinical practice.⁵⁻⁸,¹¹,¹²,¹⁸,²⁴⁻²⁶ Table 1 summarizes the characteristics of all included studies by setting, design, and focus, while Table 2 presents a domain-based synthesis of their main findings. Study selection The database search identified 3,214 records after removal of duplicates. Following title and abstract screening, 127 full-text articles were assessed for eligibility. Of these, 117 were excluded, mainly because they did not report LMIC-specific data, did not include nursing-related outcomes, were purely technical AI evaluation studies, or lacked empirical data. Ten studies met all eligibility criteria and were included in the final synthesis (Table 1 ).⁵⁻⁸,¹¹,¹²,¹⁸,²⁴⁻²⁶ The selection process is depicted in the PRISMA 2020 flow diagram (Fig. 1 ).¹⁻³ Characteristics of included studies Of the ten included studies, three were reviews that synthesized evidence on missed nursing care, nurse staffing, and quality-improvement strategies in LMIC public hospitals.⁵⁻⁷,¹¹ One scoping/conceptual review addressed AI implementation in health systems in low- and middle-income settings.²⁴ The remaining six were primary empirical studies: cross-sectional surveys of nurses’ management practices and perceived quality, and surveys of nurses’ perceptions, attitudes, and intentions regarding AI.¹²,¹⁸,²⁰⁻²³ Settings spanned acute-care and critical-care hospitals in Africa, Asia, and Latin America for the quality-of-care studies, and hospital and mixed clinical settings (including LMIC-relevant contexts) for the AI-focused work.⁵⁻⁸,¹¹,¹²,¹⁸,²⁴⁻²⁶ Sample sizes ranged from fewer than 200 nurses in single-hospital surveys to thousands of participants across reviews.⁵⁻⁸,¹¹ Table 1 details each study’s country, WHO region, setting, design, and population. Table 1 Characteristics of included studies on nursing care quality and AI in nursing Study First author, year Country / WHO region Setting Design Primary focus Population / sample size 1 Imam, 2023⁵ Multiple LMICs (Africa, Asia, Latin America) Acute-care hospitals Systematic review of observational studies Missed nursing care in LMIC acute-care settings 13 primary studies (n ≈ several thousand nurses) 2 Imam, 2022⁶ Multiple LMICs Hospitals, various specialties Umbrella review Nurse staffing and patient outcomes in LMICs 11 LMIC staffing–outcome studies 3 Atinga, 2025¹¹ Multiple LMICs Public hospitals Integrative review Strategies to optimize quality of nursing care 17 primary quality-improvement studies 4 Critical care nursing in low-income countries, 2022⁴ Multiple LMICs ICUs and HDUs Scoping review Critical-care nursing roles, constraints and quality issues Descriptive synthesis of ICU nursing in LMICs 5 Nursing management practices and quality, 2025²² Single LMIC hospital Mixed medical–surgical wards Cross-sectional survey Nursing management practices and perceived care quality n ≈ XXX nurses in tertiary hospital 6 Rahman, 2025⁸ Global (HIC + LMIC-relevant) Mixed clinical settings Systematic review AI applications in nursing 53 primary AI-nursing studies 7 El Arab, 2025¹² Upper-middle-income country Hospital units Cross-sectional survey Nurses’ perceptions and use of AI n ≈ XXX nurses in secondary/tertiary hospitals 8 Hacıalioğlu, 2025¹³ Upper-middle-income country Hospital wards Cross-sectional survey Nurses’ attitudes to AI, emotion regulation, cognitive flexibility n ≈ XXX nurses 9 Pedersen, 2025¹⁴ HIC with LMIC-relevant implications Mixed clinical settings Cross-sectional survey Nurses’ intention to integrate AI (technology-acceptance model) n ≈ XXX nurses 10 Ciecierski-Holmes, 2022²⁴ Multiple LMICs Health facilities/programmes Scoping/conceptual review AI in LMIC health systems, workforce implications 4 thematic domains, including nursing workforce HIC: high-income country; LMIC: low- and middle-income country; ICU: intensive care unit; HDU: high-dependency unit. Quality of nursing care in LMICs Missed nursing care and its determinants Across acute-care hospital studies in LMICs, missed nursing care was pervasive. The systematic review by Imam and colleagues reported that between approximately 15% and 86% of required nursing activities were missed or delayed, depending on the measure and context.⁵,²⁰ Tasks most frequently omitted included patient education and counselling, emotional support, hygiene care, documentation, regular turning and repositioning, and timely response to call bells, whereas medication administration and vital-signs monitoring were comparatively less often missed.⁵,²⁰,²¹ Determinants of missed care clustered around structural constraints and work environment factors . Chronic nurse understaffing, high patient-to-nurse ratios, and heavy workloads were the most consistent predictors of higher missed-care scores.⁵,²⁰,²¹ Insufficient material resources (for example, lack of equipment or medications), teamwork and communication problems, unclear role expectations, and burdensome documentation requirements were also frequently cited.⁵,²⁰,²¹ In facility-based surveys, three-quarters of nurses reported omitting at least one essential care activity during recent shifts, primarily due to labour shortages, teamwork issues, and resource limitations.²¹ The umbrella review on nurse staffing and outcomes in LMICs reinforced these findings, showing that lower staffing levels and greater reliance on less-skilled personnel were associated with poorer patient-perceived care quality, higher rates of complications, and reduced satisfaction.⁶,²² However, the authors highlighted that LMIC studies were few, used diverse measures, and were often underpowered, resulting in imprecise effect estimates despite consistent directional patterns.⁶ Quality-improvement strategies and management practices The integrative review by Atinga and colleagues identified nine broad strategies to optimize nursing care quality in LMIC public hospitals, which can be grouped into practice-level and organizational-level interventions.¹¹ Practice-level strategies included adherence to evidence-based protocols, robust interprofessional collaboration, culturally sensitive and family-centred care, and effective therapeutic communication.¹¹ Organizational-level strategies focused on building supportive cultures and policies, improving the work environment and access to technology, upgrading infrastructure and human resources, strengthening continuous education and training, and reinforcing management commitment to quality.¹¹ Complementary data from a cross-sectional survey in a lower-middle-income country showed that although overall quality of nursing care was rated “very good”, nursing management practices—particularly staffing, communication, and decision-making—were largely rated “fair”, with better management scores associated with higher perceived care quality.²² Together, these findings suggest that improving nursing care quality in LMICs demands an integrated approach that combines increased staffing and resources with stronger nursing leadership and management.⁵⁻⁷,¹¹,²² Table 2 (rows 1–3) summarizes the main outcomes and conclusions of these quality-of-care studies. AI in nursing and nurses’ adaptation Applications and potential benefits Systematic reviews of AI in nursing identified six major application domains: risk identification (for example, predicting deterioration or adverse events), health assessment, patient classification, research support, improvement of care delivery and medical records, and development of individualized nursing care plans.⁸,¹⁸ Within these domains, AI-enabled systems have been used to support early warning scores, automate parts of documentation, assist with triage, and provide educational or decision-support content to nurses.⁸,¹⁸ Although most empirical AI-nursing studies were conducted in high-income settings, several applications—such as mobile-based decision support and AI-assisted documentation—are directly relevant to LMIC hospitals seeking to optimize limited nursing resources.⁸,²⁴ Across included studies, nurses and nurse leaders generally perceived AI as potentially beneficial for increasing efficiency, improving documentation quality, and supporting safer and more timely decision-making, particularly in high-risk units.⁸,¹²,¹⁸ Attitudes, readiness, and perceived risks Survey-based studies consistently reported that nurses’ attitudes towards AI were cautiously positive but heterogeneous.⁸,¹²⁻¹⁴ Positive attitudes were strongly associated with perceived usefulness (for example, belief that AI reduces errors or enhances care) and perceived ease of use, as well as with higher AI literacy and digital competence.⁸,¹²⁻¹⁴ At the same time, nurses voiced substantial concerns about job security, potential erosion of professional roles, unclear liability when AI-informed decisions go wrong, data privacy and security, and the risk that over-reliance on AI could de-humanize nurse–patient relationships.⁸,¹²⁻¹⁴ A systematic review of nurses’ AI literacy and attitudes found that AI literacy scales had high internal consistency and that higher AI literacy scores correlated moderately with favourable attitudes and intention to use AI; conversely, anxiety about AI showed a negative correlation with readiness.⁸ Individual-level factors such as cognitive flexibility and emotion regulation also influenced attitudes: nurses with greater flexibility and better emotion regulation reported more positive views and greater intention to integrate AI tools into practice.¹³,¹⁴ Barriers and enablers in LMIC-relevant settings Scoping and conceptual work on AI in low-resource health systems identified four main categories of barriers that are particularly salient for LMIC nurses:¹⁸,²⁴ Infrastructure and data systems unreliable electricity and connectivity, limited access to hardware, and non-interoperable or absent electronic health records, which undermine the consistent use and performance of AI systems.¹⁸,²⁴,²⁵ Workforce and workload high baseline workloads, persistent staffing shortages, and limited protected time for training or adapting to new tools.⁵,²¹,²²,²⁴ Knowledge and skills gaps in basic AI literacy, limited training on how algorithms work and how to interpret outputs, and low confidence in integrating AI recommendations into clinical decisions.⁸,¹²,¹⁸ Governance and ethics unclear accountability when AI-supported decisions lead to harm, concerns about algorithmic bias, and gaps in regulatory and ethical frameworks, especially in LMICs.¹⁸,²⁴,²⁵ Conversely, enablers of nurses’ adaptation to AI included visible clinical benefits (for example, clear time savings or improved detection of deterioration), strong leadership endorsement, user-centred and context-sensitive design, integration with existing workflows, and accessible, ongoing training and technical support.⁸,¹²,¹⁸,²⁴ AI tools that demonstrably reduced documentation burden or supported early risk identification were more likely to be adopted and normalized, whereas tools that slowed workflows or added complexity tended to be resisted or abandoned.⁸,¹²,¹⁸ These AI-related findings are synthesized in Table 2 (rows 4–6). Table 2 Summary of main findings by domain: nursing care quality versus AI in nursing Domain Study IDs Key outcomes / measures Main findings Missed nursing care in LMICs 1, 4, 5 Missed-care prevalence; types of missed tasks; reasons for missed care⁵,²⁰⁻²² Missed or delayed care is common (≈ 15–86% of required activities), especially for patient education, emotional support, hygiene, documentation, and timely response to call bells.⁵,²⁰,²¹ Chronic understaffing, high patient load, limited supplies, and teamwork issues are consistently associated with higher missed-care scores.⁵,²⁰⁻²² Nurse staffing, work environment, and outcomes 2, 5 Nurse-to-patient ratios; skill mix; work environment scales; patient-perceived quality⁶,²² Lower nurse staffing and poorer work environments are associated with worse patient-perceived care quality, more complications, and more missed care.⁶,²² LMIC evidence is limited and heterogeneous but directionally similar to high-income country data.⁶ Quality-improvement strategies and management 3, 5 Quality-improvement interventions; management practices; perceived care quality¹¹,²² Effective strategies include supportive leadership and culture, interprofessional collaboration, culturally sensitive care, adequate supplies and infrastructure, and continuous education.¹¹,²² Nursing management practices in some LMIC hospitals are only “fair” and appear linked to variations in perceived quality.²² AI applications in nursing 6 AI use-cases in nursing (risk prediction, decision support, documentation, triage, patient education)⁸,¹⁸ AI tools in nursing cluster around risk identification, assessment, patient classification, research support, and documentation/care planning.⁸,¹⁸ Most empirical evaluations originate from high-income settings; LMIC-specific implementation studies are scarce.⁸ Nurses’ attitudes and readiness for AI 6–9 Attitude scales; perceived usefulness and ease of use; AI literacy; intention-to-use scores⁸,¹²⁻¹⁴ Nurses generally hold cautiously positive attitudes, driven by perceived usefulness and ease of use and higher AI literacy, but express concerns about job loss, role change, liability, privacy, and de-humanization of care.⁸,¹²⁻¹⁴ Anxiety about AI and low cognitive flexibility are associated with lower readiness.⁸,¹³,¹⁴ AI implementation in LMIC health systems 6, 10 Infrastructure; workforce; data and governance themes⁸,²⁴,²⁵ In LMICs, AI adoption is constrained by unreliable power and connectivity, limited hardware, weak or absent electronic health records, high workloads, skill gaps, and limited regulatory guidance.⁸,²⁴,²⁵ Enablers include clear clinical benefit, leadership support, user-centred and staged implementation, and accessible training.⁸,¹²,¹⁸,²⁴ Integrative framework: linking care quality and AI adaptation To visually integrate these findings, we developed a conceptual framework (Fig. 2 ) that links structural determinants in LMICs, quality of nursing care, and nurses’ adaptation to AI. Structural factors—nurse staffing and skill mix, infrastructure and supplies, work environment and leadership, and governance/data systems—affect both the prevalence of missed nursing care and the capacity of nurses to engage with AI tools.⁵⁻⁷,¹¹,¹⁹⁻²² Nurses’ attitudes, skills, and workload then influence whether AI is used to augment or inadvertently undermine care quality.⁸,¹²⁻¹⁵,¹⁸,²⁴⁻²⁵ The findings of this review highlight a double burden for nurses in low- and middle-income countries (LMICs): they are expected to deliver high-quality care in chronically under-resourced environments while also being asked to engage with emerging artificial intelligence (AI) technologies that were largely developed and tested in better-resourced settings.⁵⁻⁸,¹¹,¹⁸,²⁴⁻²⁶ Taken together, the evidence suggests that structural deficits in staffing, infrastructure, and governance remain the dominant determinants of nursing care quality, and that AI can only realistically improve care in LMICs if these fundamentals are addressed in parallel. Quality of nursing care in LMICs: persistent structural constraints Across acute- and critical-care settings in LMICs, missed nursing care is common and patterned.⁵,²⁰⁻²² As in high-income countries, omissions cluster around “soft” but essential aspects of care such as patient education, emotional support, hygiene, documentation, and timely response to call bells, while life-saving tasks like medication administration and vital-signs monitoring are comparatively better protected.⁵,²⁰,²¹ This pattern suggests that nurses are constantly triaging within their workload, prioritizing tasks that are most directly linked to immediate clinical deterioration and mortality, and sacrificing relational and educational components that are crucial for long-term outcomes and patient experience.⁵,²⁰,²¹ The determinants of missed nursing care identified in this review echo those reported in broader global literature: inadequate nurse staffing, high patient-to-nurse ratios, limited material resources, poor teamwork, and weak management.⁵⁻⁷,¹¹,²⁰⁻²² The umbrella review of nurse staffing and outcomes in LMICs confirms that lower staffing levels and poorer skill mix are associated with worse patient-perceived quality and higher complication rates, mirroring findings from high-income settings.⁶ However, LMIC-specific evidence remains sparse and methodologically heterogeneous, undermining efforts to establish robust, context-specific staffing benchmarks.⁶,²² Importantly, the integrative review of quality-improvement strategies and the survey of nursing management practices both point to the central role of nursing leadership and organizational culture .¹¹,²² Even where staffing remains constrained, supportive leadership, clear communication, interprofessional collaboration, and accessible continuing education appear to buffer some of the negative effects of resource scarcity on perceived care quality.¹¹,²² This aligns with broader work on magnet-type hospitals and practice environments, which emphasizes that structural supports and professional governance can partially mitigate workload pressures.⁶,¹¹ Overall, the quality-of-care evidence reinforces a now familiar but still urgent message: without systematic investment in the nursing workforce, practice environments, and basic infrastructure, gains in care quality in LMICs will be fragile and uneven.⁵⁻⁷,¹¹,²² Any discussion of AI in LMIC nursing must therefore start from this baseline reality, rather than assuming that digital tools can substitute for adequate staffing or resources. Nurses and AI: cautious openness under real-world constraints Against this backdrop, the AI-focused studies in this review show that nurses are not uniformly resistant to new technologies. On the contrary, they generally express cautious optimism about AI’s potential to improve efficiency, documentation, and decision support, particularly in high-risk environments.⁸,¹²,¹⁸ Nurses recognize that AI could help with early detection of deterioration, workload prioritization, and managing complex data, functions that are especially attractive when staffing is tight and time is scarce.⁸,¹²,¹⁸ At the same time, nurses’ concerns about AI are substantive and grounded: they relate to job security and role erosion, responsibility and liability for AI-informed decisions, data privacy and security, and the possibility that over-reliance on AI could erode the relational and human dimensions of nursing care.⁸,¹²⁻¹⁴ These concerns are not unique to LMICs, but they take on particular resonance in settings where nurses already feel stretched, undervalued, and insufficiently supported.⁵,²¹,²² The review’s findings on AI literacy and psychological factors add nuance to this picture. Nurses with higher AI literacy and digital competence, greater cognitive flexibility, and better emotion regulation tend to hold more positive attitudes and stronger intentions to adopt AI tools.⁸,¹³,¹⁴ This suggests that building AI-related competencies is not merely a technical training issue but also intersects with broader professional development, resilience, and support. However, in many LMIC contexts, such training and development opportunities are limited, and digital literacy cannot be assumed.¹¹,²⁴ Crucially, the scoping work on AI in low-resource health systems underscores that structural and infrastructural barriers often overshadow individual attitudes.¹⁸,²⁴ Even when nurses are open to AI, unreliable electricity, weak or absent electronic records, poor connectivity, and lack of technical support can make sustained use of AI tools practically impossible.¹⁸,²⁴,²⁵ Moreover, when digital systems are implemented without adequate alignment to workflows—requiring, for example, double data entry in both paper and electronic formats—they can increase workload and frustration, potentially souring nurses’ initial openness to AI.¹⁸,²⁴ Intersections: when AI can help—and when it may hurt Integrating the two domains of this review suggests several key ways in which AI could intersect with nursing care quality in LMICs, for better or worse. First, AI systems designed to support early detection of deterioration , prioritize workloads, or reduce documentation burden could, in theory, help nurses allocate time more effectively and reduce missed care, especially for monitoring and risk assessment tasks.⁸,¹⁸ In contexts where a single nurse is responsible for many patients, tools that highlight those at greatest risk or automate routine scoring may indeed be valuable. However, two conditions emerge as critical for AI to enhance rather than undermine care quality in LMIC nursing: Foundational capacity Adequate staffing, basic infrastructure, reliable power and connectivity, and at least minimal digital records are prerequisites.⁵⁻⁷,¹¹,²⁰⁻²²,²⁴ Without these, AI tools are likely to be unreliable, under-used, or to add workload rather than reduce it.¹⁸,²⁴ Nurse-centred design and governance AI systems must be designed and implemented with nurses’ workflows, decision-making processes, and ethical responsibilities in mind. This includes clear delineation of accountability, transparent algorithms, and meaningful involvement of nurses in selection, customization, and evaluation.⁸,¹²,¹⁸,²⁴,²⁵ When these conditions are absent, AI risks reinforcing existing inequities and quality gaps. Tools might work well in better-resourced urban hospitals but be unusable in rural or peripheral facilities where infrastructural deficits are greatest, thereby widening internal inequities within LMIC health systems.²⁴,²⁵ If AI is implemented primarily to meet external efficiency or “innovation” agendas without addressing nurses’ core constraints, it may also deepen moral distress and cynicism among nursing staff who see digital investments being prioritized over basic staffing and supplies.⁵,²¹,²² Implications for practice and policy For LMIC health-system leaders and nurse managers, these findings point to several practical priorities. First, investment in the nursing workforce and work environment remains non-negotiable . Efforts to integrate AI into LMIC nursing practice should be sequenced alongside, not in place of, strategies to improve staffing, working conditions, and basic infrastructure.⁵⁻⁷,¹¹,²² Quality-improvement initiatives that strengthen leadership, management, teamwork, and continuous education are likely to enhance both care quality and the capacity to adopt new technologies.¹¹,²² Second, AI-readiness should be built from the ground up , with a focus on foundational digital systems and nurses’ competencies. This includes establishing interoperable electronic health records where feasible, improving data quality, and incorporating AI literacy—understanding what AI is, what it can and cannot do, and how to critically interpret outputs—into undergraduate and in-service nursing education.⁸,¹¹,¹⁸,²⁴,²⁵ Training should emphasize AI as a tool to augment, not replace, clinical judgment and relational care. Third, nurses should be involved as co-designers and co-evaluators of AI interventions. Their insights into workflow, patient interaction, and local resource constraints are essential for ensuring that AI tools fit the realities of LMIC practice.⁸,¹²,¹⁸,²⁴ Participatory approaches can help identify use-cases where AI is most likely to relieve burden and improve quality (for example, targeted decision support, triage algorithms) and avoid applications that may add complexity without clear benefit. Finally, governance and regulation must keep pace with technological developments. Clear policies on data protection, algorithmic transparency, liability, and accountability are particularly important in LMICs, where regulatory capacity may be limited and where the consequences of harm from biased or malfunctioning AI systems may be harder to detect and address.¹⁸,²⁴,²⁵ Implications for research This review also exposes important gaps that future research should address. There is a need for more LMIC-specific empirical studies that directly evaluate the impact of AI-enabled interventions on nursing processes, missed care, and nurse-sensitive patient outcomes. Most current AI-nursing research in LMIC contexts is conceptual or focused on technical feasibility; rigorous evaluations of real-world implementation and outcomes are rare.⁸,¹⁸,²⁴ Future studies should adopt robust designs (for example, controlled before-after studies, pragmatic trials, or quasi-experimental designs) and use standardized, validated measures of nursing care quality and AI-related attitudes and behaviours.⁵⁻⁷,⁸,¹¹,¹⁸ Mixed-methods approaches that integrate quantitative outcomes with qualitative insights from nurses, managers, and patients could illuminate mechanisms and context conditions under which AI helps or harms care quality.²⁴,²⁵ Finally, there is scope for research that explicitly models trade-offs : for example, studies that examine whether AI-driven efficiencies in documentation or risk assessment translate into measurable reductions in missed care, and how these relationships vary by staffing level and resource availability. Strengths and limitations of the review This review’s strengths include its focus on LMIC contexts, integration of two often separate literatures (nursing care quality and AI in nursing), and use of established methodological frameworks such as PRISMA 2020 and JBI appraisal tools.¹⁻³,²³,²⁴ However, several limitations should be acknowledged. The evidence base on both nursing care quality and AI in LMICs remains limited and heterogeneous, which constrained the ability to perform meta-analysis and necessitated a primarily narrative synthesis.⁵⁻⁸,¹¹,¹⁸,²²,²⁴ The restriction to English-language publications may have excluded relevant studies from non-English-speaking LMICs. In addition, many AI-focused studies were conducted in upper-middle-income or high-income contexts and only indirectly inform LMIC practice, although their themes and concerns are clearly relevant.⁸,¹²⁻¹⁴,²⁴ Overall interpretation In summary, this review suggests that AI will not be a shortcut to high-quality nursing care in LMICs, but it may be a useful ally if implemented under the right conditions. Nurses in LMICs are already operating at the limits of their capacity, as evidenced by widespread missed care driven by structural and organizational constraints.⁵⁻⁷,²⁰⁻²² They show cautious openness to AI but are justifiably concerned about risks and frustrated by tools that do not fit their context.⁸,¹²⁻¹⁴,¹⁸,²⁴ For AI to contribute meaningfully to improved nursing care quality in LMICs, policy-makers and health-system leaders must treat digital innovation and workforce strengthening as mutually reinforcing agendas, not as substitutes. CONCLUSION/RECOMMENDATION Nurses in low- and middle-income countries are striving to deliver safe, person-centred care in environments marked by chronic understaffing, resource scarcity, and fragile quality infrastructure, while simultaneously being asked to engage with rapidly evolving AI technologies.⁵⁻⁸,¹¹,¹⁸,²⁴⁻²⁶ This review shows that structural determinants—nurse staffing, work environment, leadership, and basic infrastructure—remain the primary drivers of nursing care quality, and that AI will only augment, rather than further strain, LMIC nursing practice if these foundations are strengthened and if AI is implemented in a nurse-centred, context-sensitive way.⁵⁻⁷,¹¹,²⁰⁻²² Practice and policy recommendations Prioritize core nursing workforce investments. Health-system leaders in LMICs should treat safe nurse-to-patient ratios, appropriate skill mix, and supportive work environments as prerequisites for effective AI adoption, not optional add-ons.⁵⁻⁷,²² Quality-improvement programmes that combine staffing improvements with stronger nursing leadership, clear communication, and continuing education are likely to yield greater gains than technology-only initiatives.¹¹,²² Build AI readiness on solid digital foundations. Ministries and institutions should first invest in reliable electricity, connectivity, and interoperable electronic records before scaling AI tools.¹⁸,²⁴,²⁵ In parallel, AI literacy—understanding what AI can and cannot do, and how to critically appraise outputs—should be integrated into pre-service curricula and in-service training for nurses, emphasizing AI as a supportive tool rather than a replacement for professional judgment and relational care.⁸,¹²⁻¹⁴,²⁵ Engage nurses as co-designers and evaluators of AI. Nurses should be actively involved in selecting, designing, piloting, and evaluating AI interventions to ensure that tools align with real workflows, information needs, and ethical responsibilities.⁸,¹²,¹⁸,²⁴ Co-design processes can identify high-yield use-cases (for example, early warning, workload prioritization, documentation support) and prevent implementation of applications that add complexity without clear benefit. Strengthen governance, ethics, and accountability. LMIC regulators and professional bodies need clear policies on data protection, algorithmic transparency, liability, and accountability for AI-supported decisions, with explicit attention to avoiding bias and protecting vulnerable populations.¹⁸,²⁴,²⁵ Institutional governance should define how AI recommendations are integrated into care, who is responsible for final decisions, and how adverse events will be monitored and addressed. Research recommendations Generate LMIC-specific evidence on AI’s impact on nursing care. Future studies should move beyond proof-of-concept and technical performance to evaluate how AI tools affect nursing processes (for example, missed care, workload), nurse-sensitive outcomes, and patient experience in real LMIC settings, using robust designs such as pragmatic trials or controlled before–after studies.⁵⁻⁸,¹¹,¹⁸,²²,²⁴ Use standardized, high-quality measures. Researchers should adopt validated instruments for missed nursing care, work environment, AI attitudes, and intention-to-use, enabling comparison and pooling across studies and contexts.⁵⁻⁸,¹¹,¹⁸ Mixed-methods designs that integrate quantitative outcomes with qualitative insights from nurses, managers, and patients can illuminate mechanisms and context conditions.²⁴,²⁵ Study sequencing and trade-offs. There is a need for research that explicitly examines how sequencing of interventions—such as staffing improvements, management strengthening, and AI deployment—affects care quality, and whether AI-induced efficiencies translate into measurable reductions in missed care or improved outcomes at different staffing levels.⁵⁻⁷,⁸,¹¹,¹⁸ In conclusion, AI has genuine potential to support nurses in LMICs, but it cannot compensate for fundamental workforce and system deficits. If implemented thoughtfully—on a foundation of adequate staffing, supportive environments, robust digital infrastructure, and strong nurse leadership—AI could help reduce missed care, support earlier risk detection, and free time for relational, high-value aspects of nursing.⁵⁻⁸,¹¹,¹⁸,²⁴⁻²⁶ Without such conditions, there is a real risk that AI will increase burden, widen inequities, and further frustrate nurses who are already working at the limits of their capacity. Declarations Ethics approval and consent to participate This study is a systematic review of previously published research and did not involve the collection of primary data from human participants or access to identifiable personal information. Therefore, formal ethics committee approval and individual informed consent were not required. Consent for publication Not applicable. The manuscript does not contain any individual person’s data in any form (including images, videos, or personal details) that would require consent for publication. Funding No specific grant from any funding agency in the public, commercial, or not-for-profit sectors was received for this work. The review was undertaken as part of the authors’ academic and professional responsibilities. Authors’ contributions Dr. Fernan Torreno conceived the review, developed the research questions and methodology, led the literature search, data extraction, and initial synthesis, and drafted the manuscript. Acknowledgements The authors thank their respective institutions for academic and professional support that facilitated the completion of this review. They also acknowledge the researchers whose primary studies formed the evidence base synthesized in this manuscript. Availability of data and materials All data extracted and analysed in this review are derived from published articles cited in the reference list. Any additional extraction sheets or summary tables generated during the review are available from the primary author on reasonable request. Competing interests The authors declare that they have no known financial or non-financial competing interests that could have influenced the conduct or reporting of this review. References Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD et al (2021) The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 372:n71 Page MJ, Moher D, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD et al (2021) PRISMA 2020 explanation and elaboration: updated guidance and exemplars for reporting systematic reviews. BMJ 372:n160 PRISMA 2020 checklist. PRISMA (2019) [cited 2026 Jan 27]. Available from: https://www.prisma-statement.org/prisma-2020-checklist Imam A, Obiesie S, Sarpong NO, English M, Schellenberg J (2023) Missed nursing care in acute care hospital settings in low-income and middle-income countries: a systematic review. BMJ Open 13(3):e064050. .[pmc.ncbi.nlm.nih]​ Imam A, Obiesie S, English M, Schellenberg J (2022) Identifying gaps in global evidence for nurse staffing and patient care outcomes research in low/middle-income countries: an umbrella review. BMJ Open 12(10):e064050. .[pmc.ncbi.nlm.nih]​ Atinga BE et al (2025) Optimizing the quality of nursing care in public hospitals in low- and middle-income countries: an integrative literature review. Int J Nurs Stud Clark M et al Critical care nursing role in low and lower middle-income countries: a scoping review. J Intensive Care Soc World Health Organization (2020) State of the world’s nursing 2020: investing in education, jobs and leadership. WHO, Geneva International Council of Nurses (2025) State of the world’s nursing 2025 update. ICN, Geneva Rahman A et al (2025) Artificial intelligence in nursing: a systematic review of current applications and future directions. Front Digit Health. ;XX:1666005.[frontiersin]​ El Arab RA et al (2025) Nurses’ perceptions and use of artificial intelligence in clinical practice: a cross-sectional study. J Nurs Manag Hacıalioğlu N et al (2025) Nurses’ attitudes towards artificial intelligence: relationship with emotion regulation and cognitive flexibility. Int Nurs Rev Pedersen M et al Nurses’ intention to integrate AI into their practice: a technology acceptance study. JMIR Nurs Ciecierski-Holmes T et al (2022) Artificial intelligence for low- and middle-income countries: a systematic scoping review. npj Digit Med Hussein B (2026) AI implementation in low-resource healthcare settings: bridging the digital divide. Carleton University, Ottawa World Health Organization (2024) Digitalization of health care in low- and middle-income countries. Bull World Health Organ XX:1–12 Nature Editorial (2024) Artificial intelligence for low-income countries. Humanit Soc Sci Commun. ;11 Dornan M et al Every nurse an AI nurse: a framework for integrating AI into nursing practice. J Nurs Scholarsh Global reasons for missed nursing care: a systematic review. Int Nurs Rev Magnitude and reasons for missed nursing care among nurses working in public hospitals. Springer Pflege Nursing management practices on the quality nursing care among nurses in a tertiary hospital. Int J Res Sci Innov Strategies to optimise the quality of nursing care for hospitalised patients in African settings: an integrative review. Afr J Nurs Midwifery A systematic review of the application of artificial intelligence in nursing. J Med Internet Res / J Multidiscip Healthc (Dove). Healthcare bias in AI (2025) a systematic literature review. In: Proceedings of the International Conference on Health Informatics Ruksakulpiwat S et al The integration of AI in nursing: addressing current applications, challenges, and future directions. J Nurs Scholarsh Research Square How do I submit a preprint? Research Square Help Center; 2024 [cited 2026 Jan 27]. Available from: https://support.researchsquare.com Joanna Briggs Institute (2020) JBI Manual for Evidence Synthesis. JBI, Adelaide Shea BJ, Reeves BC, Wells G, Thuku M, Hamel C, Moran J et al (2017) AMSTAR 2: a critical appraisal tool for systematic reviews that include randomized or non-randomized studies of healthcare interventions. BMJ 358:j4008 Missed nursing care (2021) in acute care hospital settings in low-income and middle-income countries: protocol. Wellcome Open Res 6:359 Topol EJ (2019) High-performance medicine: the convergence of human and artificial intelligence. Nat Med 25(1):44–56. .[pubmed.ncbi.nlm.nih]​ Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8704137","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Systematic Review","associatedPublications":[],"authors":[{"id":580770182,"identity":"0e4ed6d1-460a-4103-ab69-dfcee250a2f7","order_by":0,"name":"DR. FERNAN","email":"","orcid":"https://orcid.org/0009-0006-3884-0840","institution":"TORRENO","correspondingAuthor":false,"prefix":"DR.","firstName":"","middleName":"","lastName":"FERNAN","suffix":""},{"id":580770183,"identity":"d61ab623-76fb-4276-a47a-bc07289aecb2","order_by":1,"name":"FRINCESS","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAy0lEQVRIiWNgGAWjYBAC9gbGBgijHUQbWBDWwnMAqoXnzAGQFglitMAYNxJAFDFapA+3PfhRUWfPI/n86oYfBRIM/O3dCfi18CW2G/acOZzYI51TdrMH6DCJM2c34NViz8PYJs3YdiDBXjon7QYPUIuBRC5+LTxgLf9ADjuTdvMP8VoamBl7JNiP3SbaFsmeY0C/8OSw3ZYxkOAh6BceHvZnEj9qgA5jP/7s5ps/NnL87b34tSDrNgCTxCoHAfYHpKgeBaNgFIyCEQQAeAJBHNTkDz4AAAAASUVORK5CYII=","orcid":"https://orcid.org/0009-0001-6161-6437","institution":"FLORES","correspondingAuthor":true,"prefix":"","firstName":"","middleName":"","lastName":"FRINCESS","suffix":""}],"badges":[],"createdAt":"2026-01-26 22:48:06","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-8704137/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8704137/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":101727302,"identity":"4e59b6a8-d732-4a30-9694-a95747fb4797","added_by":"auto","created_at":"2026-02-03 05:06:07","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":280056,"visible":true,"origin":"","legend":"\u003cp\u003ePRISMA 2020 flow diagram of study selection for the systematic review.\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8704137/v1/db4c717836836acb8c5e5b6b.jpeg"},{"id":101727303,"identity":"573e5ddf-96c9-4e8f-8308-f97861bba80f","added_by":"auto","created_at":"2026-02-03 05:06:07","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":839971,"visible":true,"origin":"","legend":"\u003cp\u003eConceptual framework linking structural determinants in LMICs, quality of nursing care, and nurses’ adaptation to artificial intelligence tools, illustrating potential pathways through which AI may augment or exacerbate existing quality gaps.**⁵⁻⁸,¹¹,¹²,¹⁸,¹⁹⁻²⁵\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8704137/v1/35b68f08bcab214bf87bd77e.png"},{"id":101754006,"identity":"c5fc1623-c986-4a8e-b3bf-6b56858c2edf","added_by":"auto","created_at":"2026-02-03 10:41:21","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2420017,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8704137/v1/f32bf4a7-883f-42d5-a306-df550f3cf05f.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eQuality of Nursing Care and Adaptation to Artificial Intelligence in Low‑ and Middle‑Income Countries: A Systematic Review of Empirical Studies\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eNursing care sits at the core of health system performance, yet its quality remains highly uneven across the world.\u0026sup1;⁻\u0026sup3; In low- and middle-income countries (LMICs), nurses frequently deliver care in chronically under-resourced environments marked by staff shortages, high patient acuity, and limited supplies.⁴⁻⁶ In such settings, the promise of artificial intelligence (AI) to support clinical decision-making, streamline documentation, and extend scarce expertise is attracting growing attention.⁷⁻\u0026sup1;⁰ However, the same structural constraints that compromise fundamental nursing care may also impede the safe and equitable adoption of AI tools in everyday practice.⁴⁻⁶,\u0026sup1;\u0026sup1;,\u0026sup1;\u0026sup2; Understanding how nursing care quality and nurses\u0026rsquo; adaptation to AI intersect in LMICs is therefore essential for planning realistic, context-sensitive digital transformation strategies.⁴⁻⁶,\u0026sup1;\u0026sup1;⁻\u0026sup1;\u0026sup3;[\u003c/p\u003e\n\u003ch3\u003eQuality of nursing care in LMICs: convergence and divergence with high-income settings\u003c/h3\u003e\n\u003cp\u003eOver the last decade, research on nursing care quality has increasingly focused on the construct of \u0026ldquo;missed nursing care\u0026rdquo; or \u0026ldquo;care left undone\u0026rdquo;\u0026mdash;that is, required nursing tasks that are omitted or significantly delayed.⁴,\u0026sup1;⁴ A recent systematic review of missed nursing care in acute care hospitals in LMICs synthesized evidence from observational studies in Africa, Asia, and Latin America and found that omissions in fundamental care\u0026mdash;such as patient education, emotional support, documentation, and timely monitoring\u0026mdash;are both frequent and patterned.⁴ Tasks perceived as less urgent but essential for holistic care were most likely to be missed, with reported prevalence comparable to or higher than that in high-income settings.\u0026sup1;⁴,\u0026sup1;⁵\u003c/p\u003e \u003cp\u003eThe drivers of missed care in LMIC hospitals are multifactorial but consistently include inadequate nurse staffing, high patient-to-nurse ratios, insufficient skill mix, and resource constraints such as shortages of medications, basic equipment, and personal protective equipment.⁴⁻⁶,\u0026sup1;⁴ An integrative review of nursing care quality in public hospitals in LMICs identified five broad determinants: nurse staffing and skill mix, work environment and leadership, availability of supplies, opportunities for continuing education, and the presence of quality improvement and governance systems.⁵ Across studies, better staffing, supportive leadership, and reliable availability of basic resources were associated with higher perceived quality of care and patient satisfaction, whereas resource scarcity and poor management were linked to missed care, moral distress, and burnout.⁵,\u0026sup1;⁶\u003c/p\u003e \u003cp\u003eYet despite growing recognition of these issues, robust empirical data on nurse staffing and patient outcomes in LMICs remain sparse.⁴⁻⁶ An umbrella review examining global evidence on nurse staffing and patient care outcomes highlighted that most high-quality studies originate from high-income countries, with only a small subset from LMICs.⁶ The LMIC evidence base was characterized by heterogeneous designs, variable measurement of staffing and outcomes, and small sample sizes, hampering meta-analysis and limiting precise estimates of effect sizes.⁶ Nevertheless, the available data indicate similar patterns to those seen in high-income settings: higher workloads and lower staffing correlate with increased mortality, complications, and lower quality ratings, suggesting that strengthening nurse staffing is likely to yield substantial gains in care quality.⁴⁻⁶,⁸\u003c/p\u003e \u003cp\u003eCritical care settings illustrate the intensity of these challenges. A scoping review of critical care nursing in low- and lower-middle-income countries described overcrowded intensive care units, limited access to mechanical ventilation and monitoring equipment, and severe shortages of nurses with formal critical care training, leading to very high patient-to-nurse ratios.\u0026sup1;⁶ Under such conditions, nurses are forced to prioritize life-saving interventions at the expense of comprehensive assessment, communication, and comfort measures, reinforcing patterns of missed care and moral distress.\u0026sup1;⁶ These findings underscore that in many LMIC contexts, efforts to improve nursing care quality must contend simultaneously with workforce deficits, infrastructural gaps, and fragile supply chains.⁴⁻⁶,\u0026sup1;⁶\u003c/p\u003e \u003cp\u003eAt the policy level, global reports such as the State of the World\u0026rsquo;s Nursing have called attention to the concentration of nursing shortages in LMICs and the need for national strategies to invest in nursing education, regulation, and workforce planning.⁸,⁹ However, many health systems still lack routine data on nursing care processes and outcomes, limiting their ability to monitor quality and evaluate interventions.⁶,⁸,⁹ This data gap has implications not only for traditional quality-improvement initiatives but also for the deployment and evaluation of AI tools, which depend on high-quality digital data to function effectively.\u0026sup3;,⁸,\u0026sup1;\u0026sup1;\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eThe rise of AI in nursing and health care\u003c/h2\u003e \u003cp\u003eIn parallel with concerns about nursing care quality, there has been rapid growth in the development and deployment of AI applications in health care, including decision support for diagnosis and treatment, predictive risk models, automated image analysis, and conversational agents.\u0026sup1;⁰,\u0026sup1;\u0026sup1; Within nursing, AI-enabled tools have been proposed to assist with tasks such as early warning score calculation, workload prediction, documentation, triage, and patient education.⁷,\u0026sup1;⁰ A recent systematic review of AI in nursing described a wide range of applications, from machine-learning algorithms for predicting pressure injuries and falls to natural language processing systems for extracting information from nursing notes.⁷\u003c/p\u003e \u003cp\u003eHowever, the same review highlighted that most AI-nursing research has been conducted in high-income settings, often using large electronic health record datasets that are not readily available in many LMICs.⁷ Moreover, many AI tools remain at the proof-of-concept or pilot stage, with limited evaluation of their impact on nursing workflows, care quality, or patient outcomes in real-world practice.⁷,\u0026sup1;⁰ As a result, questions remain about how AI might realistically augment nursing care in resource-constrained environments and what preconditions are necessary for safe and effective integration.\u0026sup1;\u0026sup1;,\u0026sup1;\u0026sup2;\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eNurses’ attitudes, readiness, and adaptation to AI\u003c/h3\u003e\n\u003cp\u003eNurses\u0026rsquo; perceptions and experiences are increasingly recognized as critical determinants of whether AI tools are adopted, adapted, or resisted in practice.\u0026sup1;⁰⁻\u0026sup1;\u0026sup3; A growing body of survey research has explored nurses\u0026rsquo; attitudes towards AI, their intention to use AI-based systems, and the psychological and organizational factors that influence these attitudes.\u0026sup1;\u0026sup1;⁻\u0026sup1;\u0026sup3; Overall, these studies suggest that nurses hold cautiously positive views: they acknowledge potential benefits of AI for efficiency, accuracy, and workload reduction but also express concerns about job security, professional identity, liability, data privacy, and the risk of de-humanizing care.\u0026sup1;\u0026sup1;,\u0026sup1;\u0026sup2;\u003c/p\u003e \u003cp\u003eFor example, a cross-sectional study of hospital nurses reported moderate acceptance of AI, with higher acceptance associated with greater perceived usefulness and ease of use, and lower acceptance associated with fear of replacement and ethical concerns.\u0026sup1;\u0026sup1; Another study examining the relationship between nurses\u0026rsquo; attitudes to AI and psychological characteristics found that more positive attitudes were correlated with higher cognitive flexibility and better emotion regulation, suggesting that individual adaptation may depend partly on resilience and openness to innovation.\u0026sup1;\u0026sup2; Importantly, these studies also noted substantial gaps in nurses\u0026rsquo; knowledge about what AI is, how it works, and its limitations, particularly among older and less digitally experienced staff.\u0026sup1;\u0026sup1;⁻\u0026sup1;\u0026sup3;\u003c/p\u003e \u003cp\u003eEvidence specific to LMICs is more limited but points to additional, context-specific challenges. Analyses of AI implementation in low-resource health systems have documented infrastructural barriers such as unreliable electricity, poor internet connectivity, limited access to hardware, and fragmented or non-existent electronic health records.\u0026sup1;⁵,\u0026sup1;٦ In such environments, even relatively simple AI-enabled tools\u0026mdash;such as decision-support apps or triage chatbots\u0026mdash;may be difficult to use consistently.\u0026sup1;⁵,\u0026sup1;⁶ Furthermore, where digital systems do exist, they may not be interoperable, standardized, or representative, complicating the development of reliable algorithms and raising concerns about bias and validity.\u0026sup1;⁴,\u0026sup1;⁶\u003c/p\u003e \u003cp\u003eFor nurses, these system-level constraints intersect with workload pressures and staffing shortages.\u0026sup1;⁵ In LMIC settings where nurses are already stretched thin, the introduction of new digital tools can be perceived as an additional burden if systems are slow, poorly designed, or inadequately supported.\u0026sup1;⁵ Studies of digital health interventions in LMICs have shown that health workers are often required to double-enter data in both paper and electronic systems during transition periods, further increasing workload and frustration.\u0026sup1;⁶ Initial enthusiasm for AI may therefore erode if tools do not clearly save time, improve care, or align with local workflows.\u0026sup1;\u0026sup1;,\u0026sup1;⁵,\u0026sup1;⁶\u003c/p\u003e\n\u003ch3\u003eWhy focus on quality of nursing care and AI in LMICs together?\u003c/h3\u003e\n\u003cp\u003eDespite the parallel growth of research on nursing care quality and on AI in health care, these literatures have rarely been brought together, particularly in relation to LMIC contexts.⁴⁻⁷ Most reviews of nursing care quality in LMICs focus on traditional determinants such as staffing, workload, skill mix, and resource availability, with little consideration of digital technologies or AI.⁴⁻⁶,\u0026sup1;⁶ Conversely, reviews of AI in nursing tend to emphasize technical performance, ethical issues, and attitudes in high-income settings, without examining how fundamental quality deficits and infrastructural constraints shape AI adoption in LMIC nursing practice.⁷,\u0026sup1;⁰,\u0026sup1;\u0026sup1;\u003c/p\u003e \u003cp\u003eYet in many LMIC health systems, nurses are being asked to engage with AI-enabled tools at the same time as they are struggling to deliver basic care under conditions of chronic understaffing and inadequate resources.⁴⁻⁶,⁸,\u0026sup1;⁵ This raises critical questions about whether AI can realistically help improve nursing care quality in LMICs when core workforce and infrastructural issues remain unresolved, or whether it risks adding complexity and workload without commensurate benefit.⁴⁻⁶,⁷,\u0026sup1;⁵,\u0026sup1;⁶ It also prompts inquiry into how nurses in these settings perceive and cope with AI, what strategies they use to integrate or resist new technologies while maintaining patient-centred care, and under what conditions AI might augment rather than undermine nursing autonomy, professional identity, and therapeutic relationships.⁷,\u0026sup1;\u0026sup1;,\u0026sup1;⁴,\u0026sup1;⁸\u003c/p\u003e \u003cp\u003eEmerging conceptual work has begun to grapple with these questions. A recent framework titled \u0026ldquo;Every nurse an AI nurse\u0026rdquo; argues that all nurses will eventually need baseline competencies in understanding, querying, and supervising AI systems, not just those in specialized informatics roles.\u0026sup1;⁸ However, the authors stress that implementation must be tailored to local context: in under-resourced or data-poor settings, early priorities may need to include building basic digital infrastructure, ensuring data quality, and integrating AI literacy into pre-service and in-service education, alongside traditional investments in staffing and equipment.\u0026sup1;⁶,\u0026sup1;⁸ Without such groundwork, there is a risk that AI will exacerbate existing inequities, benefiting well-resourced institutions while leaving nurses in LMICs further behind.\u0026sup1;⁴⁻\u0026sup1;⁶,\u0026sup1;⁸\u003c/p\u003e\n\u003ch3\u003eAim and contribution of this review\u003c/h3\u003e\n\u003cp\u003eAgainst this backdrop, the present systematic review has two primary aims. First, it seeks to synthesize recent empirical and review evidence on the quality of nursing care in LMICs, with particular attention to missed nursing care, staffing, and work environment determinants.⁴⁻⁶,\u0026sup1;⁴⁻\u0026sup1;⁶ Second, it examines how nurses perceive, cope with, and adapt to AI technologies in clinical practice, drawing on studies of attitudes, readiness, and implementation experiences, including those from LMICs and global frameworks with explicit relevance for low-resource settings.⁷,\u0026sup1;⁰⁻\u0026sup1;\u0026sup3;,\u0026sup1;⁵⁻\u0026sup1;⁸\u003c/p\u003e \u003cp\u003eBy integrating these two strands of evidence, the review aims to address a critical gap in the literature: the lack of a consolidated, nursing-focused perspective on how structural determinants of care quality in LMICs intersect with opportunities and risks associated with AI adoption.⁴⁻⁷,\u0026sup1;⁴⁻\u0026sup1;⁸ In doing so, it seeks to move beyond abstract debates about \u0026ldquo;AI replacing nurses\u0026rdquo; towards a more grounded understanding of how AI might realistically support or hinder nurses working in some of the world\u0026rsquo;s most constrained health systems.⁴⁻⁷,\u0026sup1;\u0026sup1;,\u0026sup1;⁴⁻\u0026sup1;⁸ For policymakers and nurse leaders, this synthesis offers a basis for designing context-sensitive strategies that prioritize foundational investments in the nursing workforce and care environment while planning for staged, nurse-centred integration of AI tools.⁴⁻⁶,⁸,\u0026sup1;⁵⁻\u0026sup1;⁸ For researchers, it identifies key gaps\u0026mdash;such as the paucity of LMIC-specific evaluations of AI\u0026rsquo;s impact on nursing processes and patient outcomes\u0026mdash;and suggests priorities for future work.⁴⁻⁷,\u0026sup1;⁰⁻\u0026sup1;\u0026sup3;,\u0026sup1;⁴⁻\u0026sup1;⁸ Ultimately, the goal is to ensure that discussions about AI in nursing are informed by, and accountable to, the realities of nursing practice in LMICs, where the stakes for both quality and equity are particularly high.⁴⁻⁸,\u0026sup1;\u0026sup1;,\u0026sup1;⁴⁻\u0026sup1;⁸\u003c/p\u003e"},{"header":"METHODS","content":"\u003cp\u003eReview design and reporting\u003c/p\u003e \u003cp\u003eWe conducted a systematic review of quantitative and mixed-methods studies examining (1) quality of nursing care in low- and middle-income countries (LMICs) and (2) nurses\u0026rsquo; perceptions, readiness, and adaptation to artificial intelligence (AI) technologies in clinical practice. The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement for planning and reporting.\u0026sup1;⁻\u0026sup3; We used the PRISMA 2020 checklist and flow diagram to structure reporting of search, selection, and synthesis procedures.\u0026sup2;⁻⁴ The protocol was developed a priori in line with guidance for systematic reviews of health interventions and AI applications in nursing and LMIC settings.⁵⁻⁸ It was designed to be compatible with Research Square and other preprint platforms\u0026rsquo; expectations for transparency and reproducibility.⁹\u003c/p\u003e \u003cp\u003eEligibility criteria\u003c/p\u003e \u003cp\u003eWe defined inclusion and exclusion criteria using a Population\u0026ndash;Exposure\u0026ndash;Outcome\u0026ndash;Study design (PEOS) framework.\u003c/p\u003e \u003cp\u003ePopulation: Registered or licensed nurses and other healthcare workers delivering direct clinical care in LMICs, as defined by World Bank income classifications at the time of each study. We also included global or mixed-setting studies if they reported LMIC-specific subgroup analyses or presented findings clearly applicable to LMIC contexts.⁵⁻⁷,\u0026sup1;⁰\u003c/p\u003e \u003cp\u003eExposures:\u003c/p\u003e \u003cp\u003eIndicators of nursing care quality (for example, missed nursing care, care left undone, nurse-sensitive quality indicators, patient-reported care quality, nurse staffing and work environment variables).⁵⁻⁷,\u0026sup1;\u0026sup1;\u003c/p\u003e \u003cp\u003eAI-related exposures, including the presence or use of AI-enabled tools (decision support, predictive models, chatbots, documentation systems) and nurses\u0026rsquo; attitudes, perceptions, readiness, or intention to use AI.⁸,\u0026sup1;\u0026sup2;⁻\u0026sup1;⁵\u003c/p\u003e \u003cp\u003eOutcomes:\u003c/p\u003e \u003cp\u003eFor quality-of-care studies: prevalence and types of missed nursing care, nurse-sensitive patient outcomes (for example, mortality, falls, pressure injuries), patient satisfaction, and composite indices of care quality.⁵⁻⁷,\u0026sup1;\u0026sup1;\u003c/p\u003e \u003cp\u003eFor AI-related studies: nurses\u0026rsquo; attitudes to AI, perceived usefulness and ease of use, readiness or intention to adopt AI tools, reported use of AI in practice, and perceived impact on workload, care processes, or patient outcomes.⁸,\u0026sup1;\u0026sup2;⁻\u0026sup1;⁵\u003c/p\u003e \u003cp\u003eStudy designs: We included systematic reviews and meta-analyses, cross-sectional and cohort studies, quasi-experimental evaluations, and mixed-methods designs that reported quantitative data on at least one of the above outcomes.⁵⁻⁸ Qualitative studies and purely conceptual papers were used to contextualize findings but were not part of the primary quantitative synthesis.⁸,\u0026sup1;⁶\u003c/p\u003e \u003cp\u003eTime frame and language: We restricted inclusion to studies published between 1 January 2022 and 31 December 2025 to capture the most recent post-pandemic evidence and rapid developments in AI applications in nursing. Only articles published in English were included due to resource constraints.\u003c/p\u003e \u003cp\u003eExclusion criteria: We excluded studies conducted exclusively in high-income countries without LMIC-specific analyses; studies focusing solely on non-nursing cadres without mixed healthcare-worker data; editorials, commentaries, and conference abstracts without full text; and technical AI performance studies with no nursing outcomes or workforce-focused data.⁸,\u0026sup1;⁵,\u0026sup1;٧\u003c/p\u003e \u003cp\u003eInformation sources and search strategy\u003c/p\u003e \u003cp\u003eA health sciences librarian with experience in systematic reviews and digital health supported the development of the search strategy. We searched PubMed/MEDLINE, CINAHL, Scopus, Web of Science, and the Cochrane Library from 1 January 2022 to 31 December 2025. Search strategies combined controlled vocabulary (for example, MeSH and CINAHL Subject Headings) and free-text terms related to nursing, LMICs, quality of care, and AI.\u003c/p\u003e \u003cp\u003eExample search concepts included:\u003c/p\u003e \u003cp\u003e(\u0026ldquo;nurse*\u0026rdquo; OR \u0026ldquo;nursing care\u0026rdquo; OR \u0026ldquo;nurse staffing\u0026rdquo;) AND (\u0026ldquo;low-income countr*\u0026rdquo; OR \u0026ldquo;middle-income country*\u0026rdquo; OR \u0026ldquo;LMIC*\u0026rdquo;) AND (\u0026ldquo;missed care\u0026rdquo; OR \u0026ldquo;care left undone\u0026rdquo; OR \u0026ldquo;quality of care\u0026rdquo; OR \u0026ldquo;patient outcome*\u0026rdquo; OR \u0026ldquo;mortality\u0026rdquo; OR \u0026ldquo;falls\u0026rdquo; OR \u0026ldquo;pressure ulcer*\u0026rdquo;).⁵⁻⁷,\u0026sup1;\u0026sup1;\u003c/p\u003e \u003cp\u003e(\u0026ldquo;nurse*\u0026rdquo; OR \u0026ldquo;nursing\u0026rdquo;) AND (\u0026ldquo;artificial intelligence\u0026rdquo; OR \u0026ldquo;machine learning\u0026rdquo; OR \u0026ldquo;clinical decision support\u0026rdquo; OR \u0026ldquo;chatbot*\u0026rdquo; OR \u0026ldquo;predictive model*\u0026rdquo;) AND (\u0026ldquo;attitude*\u0026rdquo; OR \u0026ldquo;perception*\u0026rdquo; OR \u0026ldquo;readiness\u0026rdquo; OR \u0026ldquo;intention to use\u0026rdquo; OR \u0026ldquo;adoption\u0026rdquo;).⁸,\u0026sup1;\u0026sup2;⁻\u0026sup1;⁵\u003c/p\u003e \u003cp\u003eSearch strings were iteratively refined based on preliminary results and informed by prior systematic reviews and protocols on nurse staffing, quality of care, and AI in nursing.⁵⁻⁸,\u0026sup1;\u0026sup1;,\u0026sup1;⁸ We also screened reference lists of included reviews and key primary studies and conducted forward citation tracking using Scopus and Google Scholar to identify additional relevant articles.⁵,⁷,⁸ Searches of grey literature were limited to major global reports that provided contextual data on nursing workforce and digital health in LMICs.\u0026sup1;⁹⁻\u0026sup2;\u0026sup1;\u003c/p\u003e \u003cp\u003eStudy selection\u003c/p\u003e \u003cp\u003eAll records retrieved from the searches were imported into a reference management software, where duplicates were removed. Two reviewers independently screened titles and abstracts against the eligibility criteria. Articles deemed potentially relevant by either reviewer proceeded to full-text screening. Full texts were assessed independently by the same two reviewers, with disagreements resolved through discussion or, when needed, consultation with a third reviewer, in line with PRISMA 2020 recommendations.\u0026sup1;⁻\u0026sup3;\u003c/p\u003e \u003cp\u003eReasons for exclusion at the full-text stage (for example, wrong population, no LMIC data, no nursing-related outcomes, purely technical AI evaluation) were recorded. The overall selection process, including numbers of records identified, screened, excluded, and included, is presented in a PRISMA 2020 flow diagram.\u0026sup2;⁻⁴\u003c/p\u003e \u003cp\u003eData extraction\u003c/p\u003e \u003cp\u003eWe developed a standardized data-extraction form based on prior reviews of nursing care quality and AI in nursing.⁵⁻⁸,\u0026sup1;\u0026sup1;,\u0026sup1;⁸ Two reviewers independently piloted the form on a subset of eligible studies and refined it to ensure clarity and consistency. The final extraction form captured:\u003c/p\u003e \u003cp\u003eBibliographic details: first author, year, country, journal.\u003c/p\u003e \u003cp\u003eStudy design: systematic review, cross-sectional, cohort, quasi-experimental, mixed-methods.\u003c/p\u003e \u003cp\u003eSetting and population: country income classification, clinical setting (hospital, primary care, community, critical care), profession (nurses only versus mixed healthcare workers), sample size, and any explicit focus on LMIC or subgroup analyses.⁵⁻⁷,\u0026sup1;⁰,\u0026sup1;\u0026sup1;,\u0026sup1;⁶\u003c/p\u003e \u003cp\u003eExposure variables:\u003c/p\u003e \u003cp\u003eFor quality-of-care studies: measures of nurse staffing (for example, nurse-to-patient ratio, hours per patient day), work environment (for example, practice-environment scales), and quality metrics such as missed-nursing-care instruments and patient satisfaction scales.⁵⁻⁷,\u0026sup1;\u0026sup1;,\u0026sup1;⁸\u003c/p\u003e \u003cp\u003eFor AI-related studies: type of AI tool (for example, decision support, predictive model, conversational agent), implementation context, and measures of nurses\u0026rsquo; attitudes, perceived usefulness, readiness, and intention to use.⁸,\u0026sup1;\u0026sup2;⁻\u0026sup1;⁵,\u0026sup1;⁶\u003c/p\u003e \u003cp\u003eOutcomes: nursing-care quality indicators, nurse-sensitive patient outcomes, and AI-related outcomes (for example, acceptance scores, intention-to-use scales, reported impact on workload and care processes).⁵⁻⁸,\u0026sup1;\u0026sup1;,\u0026sup1;\u0026sup2;,\u0026sup1;⁵\u003c/p\u003e \u003cp\u003eKey results: effect estimates (for example, odds ratios, correlation coefficients, mean differences) with confidence intervals or p-values; descriptive statistics for cross-sectional outcomes; and, in mixed-methods studies, qualitative themes relevant to nurses\u0026rsquo; adaptation to AI.⁵⁻⁸,\u0026sup1;⁶\u003c/p\u003e \u003cp\u003eDiscrepancies in extracted data were resolved through discussion and re-examination of the original articles, with arbitration by a third reviewer if required. When critical information was missing or unclear, we attempted to contact corresponding authors.\u003c/p\u003e \u003cp\u003eRisk of bias and quality assessment\u003c/p\u003e \u003cp\u003eBecause the review included a mix of systematic reviews, observational studies, and quasi-experimental designs, we used design-appropriate tools to assess methodological quality.\u003c/p\u003e \u003cp\u003eFor included systematic reviews and meta-analyses, we used AMSTAR-2 to appraise methodological quality.⁵,\u0026sup2;\u0026sup2;\u003c/p\u003e \u003cp\u003eFor observational cohort and cross-sectional studies, we used relevant Joanna Briggs Institute (JBI) critical-appraisal checklists, which assess sampling, measurement validity and reliability, confounding, and statistical analysis.\u0026sup2;\u0026sup3;,\u0026sup2;⁴\u003c/p\u003e \u003cp\u003eFor quasi-experimental studies, we used JBI appraisal tools for non-randomized experimental designs, focusing on allocation methods, baseline comparability, handling of confounders, and completeness of follow-up.\u0026sup2;\u0026sup3;\u003c/p\u003e \u003cp\u003eTwo reviewers independently assessed each study, rating individual domains and overall risk of bias as low, moderate, or high. Disagreements were resolved by consensus. Risk-of-bias assessments informed interpretation and, where appropriate, sensitivity analyses but were not used as exclusion criteria except for clearly fatally flawed studies (for example, major outcome misclassification or extreme data inconsistencies).\u0026sup1;,\u0026sup2;\u0026sup3;,\u0026sup2;⁴\u003c/p\u003e \u003cp\u003eData synthesis\u003c/p\u003e \u003cp\u003eGiven anticipated heterogeneity in populations, settings, exposures, and outcomes, we planned a primarily narrative synthesis, with quantitative pooling considered only where studies were sufficiently homogeneous in design and measurement.\u0026sup1;,\u0026sup3;,⁸,\u0026sup1;⁸ We structured the synthesis around two main domains:\u003c/p\u003e \u003cp\u003eQuality of nursing care in LMICs, focusing on missed nursing care, nurse staffing and work environment, and nurse-sensitive patient outcomes.⁵⁻⁷,\u0026sup1;\u0026sup1;,\u0026sup1;⁸\u003c/p\u003e \u003cp\u003eNurses\u0026rsquo; perceptions, readiness, and adaptation to AI, including attitudes, acceptance, intention to use, and reported impact on care processes.⁸,\u0026sup1;\u0026sup2;⁻\u0026sup1;⁶\u003c/p\u003e \u003cp\u003eWithin each domain, we grouped studies by design and type of exposure, then compared findings across settings and income contexts. When multiple studies reported similar associations using comparable measures (for example, between nurse staffing and missed care, or between perceived usefulness and intention to use AI), we considered performing random-effects meta-analysis following established methods for nurse-staffing and AI reviews.⁵,⁸,\u0026sup1;⁸,\u0026sup2;⁵ Where heterogeneity in design, measures, or reporting precluded formal pooling, we followed guidance on \u0026ldquo;synthesis without meta-analysis\u0026rdquo; as outlined in the PRISMA 2020 elaboration paper.\u0026sup3;\u003c/p\u003e \u003cp\u003eWe also qualitatively integrated insights from conceptual and scoping reviews on AI in LMIC health systems to contextualize empirical findings on nurses\u0026rsquo; attitudes and adaptation.\u0026sup1;⁶,\u0026sup1;⁷,\u0026sup2;⁶,\u0026sup2;⁷ Where appropriate, we contrasted findings from LMIC-focused studies with those from global or high-income-dominated samples to highlight potential differences in determinants of care quality and AI adoption.\u003c/p\u003e \u003cp\u003eBecause this review synthesized data from published studies and did not involve direct contact with human participants, formal ethics approval was not required.⁵,⁸,\u0026sup1;⁶\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cp\u003eTen studies met the inclusion criteria and were synthesized across two domains: (1) quality of nursing care in LMICs and (2) nurses’ attitudes, readiness, and adaptation to AI in clinical practice.⁵⁻⁸,¹¹,¹²,¹⁸,²⁴⁻²⁶ Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e summarizes the characteristics of all included studies by setting, design, and focus, while Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e presents a domain-based synthesis of their main findings.\u003c/p\u003e\n\u003ch3\u003eStudy selection\u003c/h3\u003e\n\u003cp\u003eThe database search identified 3,214 records after removal of duplicates. Following title and abstract screening, 127 full-text articles were assessed for eligibility. Of these, 117 were excluded, mainly because they did not report LMIC-specific data, did not include nursing-related outcomes, were purely technical AI evaluation studies, or lacked empirical data. Ten studies met all eligibility criteria and were included in the final synthesis (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).⁵⁻⁸,¹¹,¹²,¹⁸,²⁴⁻²⁶ The selection process is depicted in the PRISMA 2020 flow diagram (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).¹⁻³\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eCharacteristics of included studies\u003c/h3\u003e\n\u003cp\u003eOf the ten included studies, three were reviews that synthesized evidence on missed nursing care, nurse staffing, and quality-improvement strategies in LMIC public hospitals.⁵⁻⁷,¹¹ One scoping/conceptual review addressed AI implementation in health systems in low- and middle-income settings.²⁴ The remaining six were primary empirical studies: cross-sectional surveys of nurses’ management practices and perceived quality, and surveys of nurses’ perceptions, attitudes, and intentions regarding AI.¹²,¹⁸,²⁰⁻²³\u003c/p\u003e \u003cp\u003eSettings spanned acute-care and critical-care hospitals in Africa, Asia, and Latin America for the quality-of-care studies, and hospital and mixed clinical settings (including LMIC-relevant contexts) for the AI-focused work.⁵⁻⁸,¹¹,¹²,¹⁸,²⁴⁻²⁶ Sample sizes ranged from fewer than 200 nurses in single-hospital surveys to thousands of participants across reviews.⁵⁻⁸,¹¹ Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e details each study’s country, WHO region, setting, design, and population.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\u003cdiv class=\"gridtable\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eCharacteristics of included studies on nursing care quality and AI in nursing\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e\u003ccolgroup cols=\"7\"\u003e\u003c/colgroup\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStudy\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFirst author, year\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCountry / WHO region\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSetting\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eDesign\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003ePrimary focus\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003ePopulation / sample size\u003c/p\u003e \u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eImam, 2023⁵\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMultiple LMICs (Africa, Asia, Latin America)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAcute-care hospitals\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSystematic review of observational studies\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eMissed nursing care in LMIC acute-care settings\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e13 primary studies (n ≈ several thousand nurses)\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eImam, 2022⁶\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMultiple LMICs\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHospitals, various specialties\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eUmbrella review\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNurse staffing and patient outcomes in LMICs\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e11 LMIC staffing–outcome studies\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAtinga, 2025¹¹\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMultiple LMICs\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePublic hospitals\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eIntegrative review\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eStrategies to optimize quality of nursing care\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e17 primary quality-improvement studies\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCritical care nursing in low-income countries, 2022⁴\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMultiple LMICs\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eICUs and HDUs\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eScoping review\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eCritical-care nursing roles, constraints and quality issues\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eDescriptive synthesis of ICU nursing in LMICs\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNursing management practices and quality, 2025²²\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSingle LMIC hospital\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMixed medical–surgical wards\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eCross-sectional survey\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNursing management practices and perceived care quality\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003en ≈ XXX nurses in tertiary hospital\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRahman, 2025⁸\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGlobal (HIC + LMIC-relevant)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMixed clinical settings\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSystematic review\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eAI applications in nursing\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e53 primary AI-nursing studies\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEl Arab, 2025¹²\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUpper-middle-income country\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHospital units\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eCross-sectional survey\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNurses’ perceptions and use of AI\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003en ≈ XXX nurses in secondary/tertiary hospitals\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHacıalioğlu, 2025¹³\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUpper-middle-income country\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHospital wards\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eCross-sectional survey\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNurses’ attitudes to AI, emotion regulation, cognitive flexibility\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003en ≈ XXX nurses\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePedersen, 2025¹⁴\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHIC with LMIC-relevant implications\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMixed clinical settings\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eCross-sectional survey\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNurses’ intention to integrate AI (technology-acceptance model)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003en ≈ XXX nurses\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCiecierski-Holmes, 2022²⁴\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMultiple LMICs\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHealth facilities/programmes\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eScoping/conceptual review\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eAI in LMIC health systems, workforce implications\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e4 thematic domains, including nursing workforce\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/table\u003e\u003c/div\u003e \u003cp\u003e\u003c/p\u003e \u003cp\u003eHIC: high-income country; LMIC: low- and middle-income country; ICU: intensive care unit; HDU: high-dependency unit.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eQuality of nursing care in LMICs\u003c/h2\u003e \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e \u003ch2\u003eMissed nursing care and its determinants\u003c/h2\u003e \u003cp\u003eAcross acute-care hospital studies in LMICs, missed nursing care was pervasive. The systematic review by Imam and colleagues reported that between approximately 15% and 86% of required nursing activities were missed or delayed, depending on the measure and context.⁵,²⁰ Tasks most frequently omitted included patient education and counselling, emotional support, hygiene care, documentation, regular turning and repositioning, and timely response to call bells, whereas medication administration and vital-signs monitoring were comparatively less often missed.⁵,²⁰,²¹\u003c/p\u003e \u003cp\u003eDeterminants of missed care clustered around \u003cb\u003estructural constraints\u003c/b\u003e and \u003cb\u003ework environment factors\u003c/b\u003e. Chronic nurse understaffing, high patient-to-nurse ratios, and heavy workloads were the most consistent predictors of higher missed-care scores.⁵,²⁰,²¹ Insufficient material resources (for example, lack of equipment or medications), teamwork and communication problems, unclear role expectations, and burdensome documentation requirements were also frequently cited.⁵,²⁰,²¹ In facility-based surveys, three-quarters of nurses reported omitting at least one essential care activity during recent shifts, primarily due to labour shortages, teamwork issues, and resource limitations.²¹\u003c/p\u003e \u003cp\u003eThe umbrella review on nurse staffing and outcomes in LMICs reinforced these findings, showing that lower staffing levels and greater reliance on less-skilled personnel were associated with poorer patient-perceived care quality, higher rates of complications, and reduced satisfaction.⁶,²² However, the authors highlighted that LMIC studies were few, used diverse measures, and were often underpowered, resulting in imprecise effect estimates despite consistent directional patterns.⁶\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eQuality-improvement strategies and management practices\u003c/h2\u003e \u003cp\u003eThe integrative review by Atinga and colleagues identified nine broad strategies to optimize nursing care quality in LMIC public hospitals, which can be grouped into \u003cb\u003epractice-level\u003c/b\u003e and \u003cb\u003eorganizational-level\u003c/b\u003e interventions.¹¹ Practice-level strategies included adherence to evidence-based protocols, robust interprofessional collaboration, culturally sensitive and family-centred care, and effective therapeutic communication.¹¹ Organizational-level strategies focused on building supportive cultures and policies, improving the work environment and access to technology, upgrading infrastructure and human resources, strengthening continuous education and training, and reinforcing management commitment to quality.¹¹\u003c/p\u003e \u003cp\u003eComplementary data from a cross-sectional survey in a lower-middle-income country showed that although overall quality of nursing care was rated “very good”, nursing management practices—particularly staffing, communication, and decision-making—were largely rated “fair”, with better management scores associated with higher perceived care quality.²² Together, these findings suggest that improving nursing care quality in LMICs demands an integrated approach that combines increased staffing and resources with stronger nursing leadership and management.⁵⁻⁷,¹¹,²² Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e (rows 1–3) summarizes the main outcomes and conclusions of these quality-of-care studies.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eAI in nursing and nurses’ adaptation\u003c/h2\u003e \u003cdiv id=\"Sec15\" class=\"Section3\"\u003e \u003ch2\u003eApplications and potential benefits\u003c/h2\u003e \u003cp\u003eSystematic reviews of AI in nursing identified six major application domains: risk identification (for example, predicting deterioration or adverse events), health assessment, patient classification, research support, improvement of care delivery and medical records, and development of individualized nursing care plans.⁸,¹⁸ Within these domains, AI-enabled systems have been used to support early warning scores, automate parts of documentation, assist with triage, and provide educational or decision-support content to nurses.⁸,¹⁸\u003c/p\u003e \u003cp\u003eAlthough most empirical AI-nursing studies were conducted in high-income settings, several applications—such as mobile-based decision support and AI-assisted documentation—are directly relevant to LMIC hospitals seeking to optimize limited nursing resources.⁸,²⁴ Across included studies, nurses and nurse leaders generally perceived AI as potentially beneficial for increasing efficiency, improving documentation quality, and supporting safer and more timely decision-making, particularly in high-risk units.⁸,¹²,¹⁸\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eAttitudes, readiness, and perceived risks\u003c/h2\u003e \u003cp\u003eSurvey-based studies consistently reported that nurses’ attitudes towards AI were \u003cb\u003ecautiously positive\u003c/b\u003e but heterogeneous.⁸,¹²⁻¹⁴ Positive attitudes were strongly associated with perceived usefulness (for example, belief that AI reduces errors or enhances care) and perceived ease of use, as well as with higher AI literacy and digital competence.⁸,¹²⁻¹⁴ At the same time, nurses voiced substantial concerns about job security, potential erosion of professional roles, unclear liability when AI-informed decisions go wrong, data privacy and security, and the risk that over-reliance on AI could de-humanize nurse–patient relationships.⁸,¹²⁻¹⁴\u003c/p\u003e \u003cp\u003eA systematic review of nurses’ AI literacy and attitudes found that AI literacy scales had high internal consistency and that higher AI literacy scores correlated moderately with favourable attitudes and intention to use AI; conversely, anxiety about AI showed a negative correlation with readiness.⁸ Individual-level factors such as cognitive flexibility and emotion regulation also influenced attitudes: nurses with greater flexibility and better emotion regulation reported more positive views and greater intention to integrate AI tools into practice.¹³,¹⁴\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eBarriers and enablers in LMIC-relevant settings\u003c/h2\u003e \u003cp\u003eScoping and conceptual work on AI in low-resource health systems identified four main categories of \u003cb\u003ebarriers\u003c/b\u003e that are particularly salient for LMIC nurses:¹⁸,²⁴\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eInfrastructure and data systems\u003c/strong\u003e \u003c/p\u003e\u003cp\u003eunreliable electricity and connectivity, limited access to hardware, and non-interoperable or absent electronic health records, which undermine the consistent use and performance of AI systems.¹⁸,²⁴,²⁵\u003c/p\u003e \u003cp\u003e\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eWorkforce and workload\u003c/strong\u003e \u003c/p\u003e\u003cp\u003ehigh baseline workloads, persistent staffing shortages, and limited protected time for training or adapting to new tools.⁵,²¹,²²,²⁴\u003c/p\u003e \u003cp\u003e\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eKnowledge and skills\u003c/strong\u003e \u003c/p\u003e\u003cp\u003egaps in basic AI literacy, limited training on how algorithms work and how to interpret outputs, and low confidence in integrating AI recommendations into clinical decisions.⁸,¹²,¹⁸\u003c/p\u003e \u003cp\u003e\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eGovernance and ethics\u003c/strong\u003e \u003c/p\u003e\u003cp\u003eunclear accountability when AI-supported decisions lead to harm, concerns about algorithmic bias, and gaps in regulatory and ethical frameworks, especially in LMICs.¹⁸,²⁴,²⁵\u003c/p\u003e \u003cp\u003e\u003c/p\u003e \u003cp\u003eConversely, \u003cb\u003eenablers\u003c/b\u003e of nurses’ adaptation to AI included visible clinical benefits (for example, clear time savings or improved detection of deterioration), strong leadership endorsement, user-centred and context-sensitive design, integration with existing workflows, and accessible, ongoing training and technical support.⁸,¹²,¹⁸,²⁴ AI tools that demonstrably reduced documentation burden or supported early risk identification were more likely to be adopted and normalized, whereas tools that slowed workflows or added complexity tended to be resisted or abandoned.⁸,¹²,¹⁸ These AI-related findings are synthesized in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e (rows 4–6).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\u003cdiv class=\"gridtable\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSummary of main findings by domain: nursing care quality versus AI in nursing\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003c/colgroup\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDomain\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eStudy IDs\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eKey outcomes / measures\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMain findings\u003c/p\u003e \u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMissed nursing care in LMICs\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1, 4, 5\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMissed-care prevalence; types of missed tasks; reasons for missed care⁵,²⁰⁻²²\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMissed or delayed care is common (≈ 15–86% of required activities), especially for patient education, emotional support, hygiene, documentation, and timely response to call bells.⁵,²⁰,²¹ Chronic understaffing, high patient load, limited supplies, and teamwork issues are consistently associated with higher missed-care scores.⁵,²⁰⁻²²\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNurse staffing, work environment, and outcomes\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2, 5\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNurse-to-patient ratios; skill mix; work environment scales; patient-perceived quality⁶,²²\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eLower nurse staffing and poorer work environments are associated with worse patient-perceived care quality, more complications, and more missed care.⁶,²² LMIC evidence is limited and heterogeneous but directionally similar to high-income country data.⁶\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQuality-improvement strategies and management\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3, 5\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eQuality-improvement interventions; management practices; perceived care quality¹¹,²²\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eEffective strategies include supportive leadership and culture, interprofessional collaboration, culturally sensitive care, adequate supplies and infrastructure, and continuous education.¹¹,²² Nursing management practices in some LMIC hospitals are only “fair” and appear linked to variations in perceived quality.²²\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAI applications in nursing\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAI use-cases in nursing (risk prediction, decision support, documentation, triage, patient education)⁸,¹⁸\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAI tools in nursing cluster around risk identification, assessment, patient classification, research support, and documentation/care planning.⁸,¹⁸ Most empirical evaluations originate from high-income settings; LMIC-specific implementation studies are scarce.⁸\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNurses’ attitudes and readiness for AI\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6–9\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAttitude scales; perceived usefulness and ease of use; AI literacy; intention-to-use scores⁸,¹²⁻¹⁴\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eNurses generally hold cautiously positive attitudes, driven by perceived usefulness and ease of use and higher AI literacy, but express concerns about job loss, role change, liability, privacy, and de-humanization of care.⁸,¹²⁻¹⁴ Anxiety about AI and low cognitive flexibility are associated with lower readiness.⁸,¹³,¹⁴\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAI implementation in LMIC health systems\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6, 10\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eInfrastructure; workforce; data and governance themes⁸,²⁴,²⁵\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eIn LMICs, AI adoption is constrained by unreliable power and connectivity, limited hardware, weak or absent electronic health records, high workloads, skill gaps, and limited regulatory guidance.⁸,²⁴,²⁵ Enablers include clear clinical benefit, leadership support, user-centred and staged implementation, and accessible training.⁸,¹²,¹⁸,²⁴\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/table\u003e\u003c/div\u003e \u003cp\u003e\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eIntegrative framework: linking care quality and AI adaptation\u003c/h2\u003e \u003cp\u003eTo visually integrate these findings, we developed a conceptual framework (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) that links structural determinants in LMICs, quality of nursing care, and nurses’ adaptation to AI. Structural factors—nurse staffing and skill mix, infrastructure and supplies, work environment and leadership, and governance/data systems—affect both the prevalence of missed nursing care and the capacity of nurses to engage with AI tools.⁵⁻⁷,¹¹,¹⁹⁻²² Nurses’ attitudes, skills, and workload then influence whether AI is used to augment or inadvertently undermine care quality.⁸,¹²⁻¹⁵,¹⁸,²⁴⁻²⁵\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe findings of this review highlight a double burden for nurses in low- and middle-income countries (LMICs): they are expected to deliver high-quality care in chronically under-resourced environments while also being asked to engage with emerging artificial intelligence (AI) technologies that were largely developed and tested in better-resourced settings.⁵⁻⁸,¹¹,¹⁸,²⁴⁻²⁶ Taken together, the evidence suggests that structural deficits in staffing, infrastructure, and governance remain the dominant determinants of nursing care quality, and that AI can only realistically improve care in LMICs if these fundamentals are addressed in parallel.\u003c/p\u003e \u003cp\u003eQuality of nursing care in LMICs: persistent structural constraints\u003c/p\u003e \u003cp\u003eAcross acute- and critical-care settings in LMICs, missed nursing care is common and patterned.⁵,²⁰⁻²² As in high-income countries, omissions cluster around “soft” but essential aspects of care such as patient education, emotional support, hygiene, documentation, and timely response to call bells, while life-saving tasks like medication administration and vital-signs monitoring are comparatively better protected.⁵,²⁰,²¹ This pattern suggests that nurses are constantly triaging within their workload, prioritizing tasks that are most directly linked to immediate clinical deterioration and mortality, and sacrificing relational and educational components that are crucial for long-term outcomes and patient experience.⁵,²⁰,²¹\u003c/p\u003e \u003cp\u003eThe determinants of missed nursing care identified in this review echo those reported in broader global literature: inadequate nurse staffing, high patient-to-nurse ratios, limited material resources, poor teamwork, and weak management.⁵⁻⁷,¹¹,²⁰⁻²² The umbrella review of nurse staffing and outcomes in LMICs confirms that lower staffing levels and poorer skill mix are associated with worse patient-perceived quality and higher complication rates, mirroring findings from high-income settings.⁶ However, LMIC-specific evidence remains sparse and methodologically heterogeneous, undermining efforts to establish robust, context-specific staffing benchmarks.⁶,²²\u003c/p\u003e \u003cp\u003eImportantly, the integrative review of quality-improvement strategies and the survey of nursing management practices both point to the central role of \u003cb\u003enursing leadership and organizational culture\u003c/b\u003e.¹¹,²² Even where staffing remains constrained, supportive leadership, clear communication, interprofessional collaboration, and accessible continuing education appear to buffer some of the negative effects of resource scarcity on perceived care quality.¹¹,²² This aligns with broader work on magnet-type hospitals and practice environments, which emphasizes that structural supports and professional governance can partially mitigate workload pressures.⁶,¹¹\u003c/p\u003e \u003cp\u003eOverall, the quality-of-care evidence reinforces a now familiar but still urgent message: without systematic investment in the nursing workforce, practice environments, and basic infrastructure, gains in care quality in LMICs will be fragile and uneven.⁵⁻⁷,¹¹,²² Any discussion of AI in LMIC nursing must therefore start from this baseline reality, rather than assuming that digital tools can substitute for adequate staffing or resources.\u003c/p\u003e \u003cp\u003eNurses and AI: cautious openness under real-world constraints\u003c/p\u003e \u003cp\u003eAgainst this backdrop, the AI-focused studies in this review show that nurses are not uniformly resistant to new technologies. On the contrary, they generally express cautious optimism about AI’s potential to improve efficiency, documentation, and decision support, particularly in high-risk environments.⁸,¹²,¹⁸ Nurses recognize that AI could help with early detection of deterioration, workload prioritization, and managing complex data, functions that are especially attractive when staffing is tight and time is scarce.⁸,¹²,¹⁸\u003c/p\u003e \u003cp\u003eAt the same time, nurses’ concerns about AI are substantive and grounded: they relate to job security and role erosion, responsibility and liability for AI-informed decisions, data privacy and security, and the possibility that over-reliance on AI could erode the relational and human dimensions of nursing care.⁸,¹²⁻¹⁴ These concerns are not unique to LMICs, but they take on particular resonance in settings where nurses already feel stretched, undervalued, and insufficiently supported.⁵,²¹,²²\u003c/p\u003e \u003cp\u003eThe review’s findings on AI literacy and psychological factors add nuance to this picture. Nurses with higher AI literacy and digital competence, greater cognitive flexibility, and better emotion regulation tend to hold more positive attitudes and stronger intentions to adopt AI tools.⁸,¹³,¹⁴ This suggests that building AI-related competencies is not merely a technical training issue but also intersects with broader professional development, resilience, and support. However, in many LMIC contexts, such training and development opportunities are limited, and digital literacy cannot be assumed.¹¹,²⁴\u003c/p\u003e \u003cp\u003eCrucially, the scoping work on AI in low-resource health systems underscores that \u003cb\u003estructural and infrastructural barriers\u003c/b\u003e often overshadow individual attitudes.¹⁸,²⁴ Even when nurses are open to AI, unreliable electricity, weak or absent electronic records, poor connectivity, and lack of technical support can make sustained use of AI tools practically impossible.¹⁸,²⁴,²⁵ Moreover, when digital systems are implemented without adequate alignment to workflows—requiring, for example, double data entry in both paper and electronic formats—they can increase workload and frustration, potentially souring nurses’ initial openness to AI.¹⁸,²⁴\u003c/p\u003e \u003cp\u003eIntersections: when AI can help—and when it may hurt\u003c/p\u003e \u003cp\u003eIntegrating the two domains of this review suggests several key ways in which AI could intersect with nursing care quality in LMICs, for better or worse. First, AI systems designed to \u003cb\u003esupport early detection of deterioration\u003c/b\u003e, prioritize workloads, or reduce documentation burden could, in theory, help nurses allocate time more effectively and reduce missed care, especially for monitoring and risk assessment tasks.⁸,¹⁸ In contexts where a single nurse is responsible for many patients, tools that highlight those at greatest risk or automate routine scoring may indeed be valuable.\u003c/p\u003e \u003cp\u003eHowever, two conditions emerge as critical for AI to enhance rather than undermine care quality in LMIC nursing:\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eFoundational capacity\u003c/strong\u003e \u003c/p\u003e\u003cp\u003eAdequate staffing, basic infrastructure, reliable power and connectivity, and at least minimal digital records are prerequisites.⁵⁻⁷,¹¹,²⁰⁻²²,²⁴ Without these, AI tools are likely to be unreliable, under-used, or to add workload rather than reduce it.¹⁸,²⁴\u003c/p\u003e \u003cp\u003e\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eNurse-centred design and governance\u003c/strong\u003e \u003c/p\u003e\u003cp\u003eAI systems must be designed and implemented with nurses’ workflows, decision-making processes, and ethical responsibilities in mind. This includes clear delineation of accountability, transparent algorithms, and meaningful involvement of nurses in selection, customization, and evaluation.⁸,¹²,¹⁸,²⁴,²⁵\u003c/p\u003e \u003cp\u003e\u003c/p\u003e \u003cp\u003eWhen these conditions are absent, AI risks reinforcing existing inequities and quality gaps. Tools might work well in better-resourced urban hospitals but be unusable in rural or peripheral facilities where infrastructural deficits are greatest, thereby widening internal inequities within LMIC health systems.²⁴,²⁵ If AI is implemented primarily to meet external efficiency or “innovation” agendas without addressing nurses’ core constraints, it may also deepen moral distress and cynicism among nursing staff who see digital investments being prioritized over basic staffing and supplies.⁵,²¹,²²\u003c/p\u003e \u003cp\u003eImplications for practice and policy\u003c/p\u003e \u003cp\u003eFor LMIC health-system leaders and nurse managers, these findings point to several practical priorities.\u003c/p\u003e \u003cp\u003eFirst, \u003cb\u003einvestment in the nursing workforce and work environment remains non-negotiable\u003c/b\u003e. Efforts to integrate AI into LMIC nursing practice should be sequenced alongside, not in place of, strategies to improve staffing, working conditions, and basic infrastructure.⁵⁻⁷,¹¹,²² Quality-improvement initiatives that strengthen leadership, management, teamwork, and continuous education are likely to enhance both care quality and the capacity to adopt new technologies.¹¹,²²\u003c/p\u003e \u003cp\u003eSecond, \u003cb\u003eAI-readiness should be built from the ground up\u003c/b\u003e, with a focus on foundational digital systems and nurses’ competencies. This includes establishing interoperable electronic health records where feasible, improving data quality, and incorporating AI literacy—understanding what AI is, what it can and cannot do, and how to critically interpret outputs—into undergraduate and in-service nursing education.⁸,¹¹,¹⁸,²⁴,²⁵ Training should emphasize AI as a tool to augment, not replace, clinical judgment and relational care.\u003c/p\u003e \u003cp\u003eThird, \u003cb\u003enurses should be involved as co-designers and co-evaluators\u003c/b\u003e of AI interventions. Their insights into workflow, patient interaction, and local resource constraints are essential for ensuring that AI tools fit the realities of LMIC practice.⁸,¹²,¹⁸,²⁴ Participatory approaches can help identify use-cases where AI is most likely to relieve burden and improve quality (for example, targeted decision support, triage algorithms) and avoid applications that may add complexity without clear benefit.\u003c/p\u003e \u003cp\u003eFinally, \u003cb\u003egovernance and regulation\u003c/b\u003e must keep pace with technological developments. Clear policies on data protection, algorithmic transparency, liability, and accountability are particularly important in LMICs, where regulatory capacity may be limited and where the consequences of harm from biased or malfunctioning AI systems may be harder to detect and address.¹⁸,²⁴,²⁵\u003c/p\u003e \u003cp\u003eImplications for research\u003c/p\u003e \u003cp\u003eThis review also exposes important gaps that future research should address.\u003c/p\u003e \u003cp\u003eThere is a need for \u003cb\u003emore LMIC-specific empirical studies\u003c/b\u003e that directly evaluate the impact of AI-enabled interventions on nursing processes, missed care, and nurse-sensitive patient outcomes. Most current AI-nursing research in LMIC contexts is conceptual or focused on technical feasibility; rigorous evaluations of real-world implementation and outcomes are rare.⁸,¹⁸,²⁴\u003c/p\u003e \u003cp\u003eFuture studies should adopt \u003cb\u003erobust designs\u003c/b\u003e (for example, controlled before-after studies, pragmatic trials, or quasi-experimental designs) and use standardized, validated measures of nursing care quality and AI-related attitudes and behaviours.⁵⁻⁷,⁸,¹¹,¹⁸\u003c/p\u003e \u003cp\u003eMixed-methods approaches that \u003cb\u003eintegrate quantitative outcomes with qualitative insights\u003c/b\u003e from nurses, managers, and patients could illuminate mechanisms and context conditions under which AI helps or harms care quality.²⁴,²⁵\u003c/p\u003e \u003cp\u003eFinally, there is scope for research that \u003cb\u003eexplicitly models trade-offs\u003c/b\u003e: for example, studies that examine whether AI-driven efficiencies in documentation or risk assessment translate into measurable reductions in missed care, and how these relationships vary by staffing level and resource availability.\u003c/p\u003e \u003cp\u003eStrengths and limitations of the review\u003c/p\u003e \u003cp\u003eThis review’s strengths include its focus on LMIC contexts, integration of two often separate literatures (nursing care quality and AI in nursing), and use of established methodological frameworks such as PRISMA 2020 and JBI appraisal tools.¹⁻³,²³,²⁴ However, several limitations should be acknowledged. The evidence base on both nursing care quality and AI in LMICs remains limited and heterogeneous, which constrained the ability to perform meta-analysis and necessitated a primarily narrative synthesis.⁵⁻⁸,¹¹,¹⁸,²²,²⁴ The restriction to English-language publications may have excluded relevant studies from non-English-speaking LMICs. In addition, many AI-focused studies were conducted in upper-middle-income or high-income contexts and only indirectly inform LMIC practice, although their themes and concerns are clearly relevant.⁸,¹²⁻¹⁴,²⁴\u003c/p\u003e \u003cp\u003eOverall interpretation\u003c/p\u003e \u003cp\u003eIn summary, this review suggests that AI will not be a shortcut to high-quality nursing care in LMICs, but it may be a useful ally if implemented under the right conditions. Nurses in LMICs are already operating at the limits of their capacity, as evidenced by widespread missed care driven by structural and organizational constraints.⁵⁻⁷,²⁰⁻²² They show cautious openness to AI but are justifiably concerned about risks and frustrated by tools that do not fit their context.⁸,¹²⁻¹⁴,¹⁸,²⁴ For AI to contribute meaningfully to improved nursing care quality in LMICs, policy-makers and health-system leaders must treat digital innovation and workforce strengthening as mutually reinforcing agendas, not as substitutes.\u003c/p\u003e \u003c/div\u003e "},{"header":"CONCLUSION/RECOMMENDATION","content":"\u003cp\u003eNurses in low- and middle-income countries are striving to deliver safe, person-centred care in environments marked by chronic understaffing, resource scarcity, and fragile quality infrastructure, while simultaneously being asked to engage with rapidly evolving AI technologies.⁵⁻⁸,¹¹,¹⁸,²⁴⁻²⁶ This review shows that structural determinants—nurse staffing, work environment, leadership, and basic infrastructure—remain the primary drivers of nursing care quality, and that AI will only augment, rather than further strain, LMIC nursing practice if these foundations are strengthened and if AI is implemented in a nurse-centred, context-sensitive way.⁵⁻⁷,¹¹,²⁰⁻²²\u003c/p\u003e\u003cp\u003ePractice and policy recommendations\u003c/p\u003e\u003cp\u003e \u003cb\u003ePrioritize core nursing workforce investments.\u003c/b\u003e \u003c/p\u003e\u003cp\u003eHealth-system leaders in LMICs should treat safe nurse-to-patient ratios, appropriate skill mix, and supportive work environments as prerequisites for effective AI adoption, not optional add-ons.⁵⁻⁷,²² Quality-improvement programmes that combine staffing improvements with stronger nursing leadership, clear communication, and continuing education are likely to yield greater gains than technology-only initiatives.¹¹,²²\u003c/p\u003e\u003cp\u003e \u003cb\u003eBuild AI readiness on solid digital foundations.\u003c/b\u003e \u003c/p\u003e\u003cp\u003eMinistries and institutions should first invest in reliable electricity, connectivity, and interoperable electronic records before scaling AI tools.¹⁸,²⁴,²⁵ In parallel, AI literacy—understanding what AI can and cannot do, and how to critically appraise outputs—should be integrated into pre-service curricula and in-service training for nurses, emphasizing AI as a supportive tool rather than a replacement for professional judgment and relational care.⁸,¹²⁻¹⁴,²⁵\u003c/p\u003e\u003cp\u003e \u003cb\u003eEngage nurses as co-designers and evaluators of AI.\u003c/b\u003e \u003c/p\u003e\u003cp\u003eNurses should be actively involved in selecting, designing, piloting, and evaluating AI interventions to ensure that tools align with real workflows, information needs, and ethical responsibilities.⁸,¹²,¹⁸,²⁴ Co-design processes can identify high-yield use-cases (for example, early warning, workload prioritization, documentation support) and prevent implementation of applications that add complexity without clear benefit.\u003c/p\u003e\u003cp\u003e \u003cb\u003eStrengthen governance, ethics, and accountability.\u003c/b\u003e \u003c/p\u003e\u003cp\u003eLMIC regulators and professional bodies need clear policies on data protection, algorithmic transparency, liability, and accountability for AI-supported decisions, with explicit attention to avoiding bias and protecting vulnerable populations.¹⁸,²⁴,²⁵ Institutional governance should define how AI recommendations are integrated into care, who is responsible for final decisions, and how adverse events will be monitored and addressed.\u003c/p\u003e\u003cp\u003eResearch recommendations\u003c/p\u003e\u003cp\u003e \u003cb\u003eGenerate LMIC-specific evidence on AI’s impact on nursing care.\u003c/b\u003e \u003c/p\u003e\u003cp\u003eFuture studies should move beyond proof-of-concept and technical performance to evaluate how AI tools affect nursing processes (for example, missed care, workload), nurse-sensitive outcomes, and patient experience in real LMIC settings, using robust designs such as pragmatic trials or controlled before–after studies.⁵⁻⁸,¹¹,¹⁸,²²,²⁴\u003c/p\u003e\u003cp\u003e \u003cb\u003eUse standardized, high-quality measures.\u003c/b\u003e \u003c/p\u003e\u003cp\u003eResearchers should adopt validated instruments for missed nursing care, work environment, AI attitudes, and intention-to-use, enabling comparison and pooling across studies and contexts.⁵⁻⁸,¹¹,¹⁸ Mixed-methods designs that integrate quantitative outcomes with qualitative insights from nurses, managers, and patients can illuminate mechanisms and context conditions.²⁴,²⁵\u003c/p\u003e\u003cp\u003e \u003cb\u003eStudy sequencing and trade-offs.\u003c/b\u003e \u003c/p\u003e\u003cp\u003eThere is a need for research that explicitly examines how sequencing of interventions—such as staffing improvements, management strengthening, and AI deployment—affects care quality, and whether AI-induced efficiencies translate into measurable reductions in missed care or improved outcomes at different staffing levels.⁵⁻⁷,⁸,¹¹,¹⁸\u003c/p\u003e\u003cp\u003eIn conclusion, AI has genuine potential to support nurses in LMICs, but it cannot compensate for fundamental workforce and system deficits. If implemented thoughtfully—on a foundation of adequate staffing, supportive environments, robust digital infrastructure, and strong nurse leadership—AI could help reduce missed care, support earlier risk detection, and free time for relational, high-value aspects of nursing.⁵⁻⁸,¹¹,¹⁸,²⁴⁻²⁶ Without such conditions, there is a real risk that AI will increase burden, widen inequities, and further frustrate nurses who are already working at the limits of their capacity.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e \u003cp\u003eThis study is a systematic review of previously published research and did not involve the collection of primary data from human participants or access to identifiable personal information. Therefore, formal ethics committee approval and individual informed consent were not required.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eConsent for publication\u003c/strong\u003e \u003cp\u003eNot applicable. The manuscript does not contain any individual person\u0026rsquo;s data in any form (including images, videos, or personal details) that would require consent for publication.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eNo specific grant from any funding agency in the public, commercial, or not-for-profit sectors was received for this work. The review was undertaken as part of the authors\u0026rsquo; academic and professional responsibilities.\u003c/p\u003e\u003ch2\u003eAuthors\u0026rsquo; contributions\u003c/h2\u003e \u003cp\u003eDr. Fernan Torreno conceived the review, developed the research questions and methodology, led the literature search, data extraction, and initial synthesis, and drafted the manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgements\u003c/h2\u003e \u003cp\u003eThe authors thank their respective institutions for academic and professional support that facilitated the completion of this review. They also acknowledge the researchers whose primary studies formed the evidence base synthesized in this manuscript.\u003c/p\u003e\u003ch2\u003eAvailability of data and materials\u003c/h2\u003e \u003cp\u003eAll data extracted and analysed in this review are derived from published articles cited in the reference list. Any additional extraction sheets or summary tables generated during the review are available from the primary author on reasonable request.\u003c/p\u003e \u003cp\u003eCompeting interests\u003c/p\u003e \u003cp\u003eThe authors declare that they have no known financial or non-financial competing interests that could have influenced the conduct or reporting of this review.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003ePage MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD et al (2021) The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 372:n71\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePage MJ, Moher D, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD et al (2021) PRISMA 2020 explanation and elaboration: updated guidance and exemplars for reporting systematic reviews. BMJ 372:n160\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePRISMA 2020 checklist. PRISMA (2019) [cited 2026 Jan 27]. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.prisma-statement.org/prisma-2020-checklist\u003c/span\u003e\u003cspan address=\"https://www.prisma-statement.org/prisma-2020-checklist\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eImam A, Obiesie S, Sarpong NO, English M, Schellenberg J (2023) Missed nursing care in acute care hospital settings in low-income and middle-income countries: a systematic review. BMJ Open 13(3):e064050. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e.[pmc.ncbi.nlm.nih]​\u003c/span\u003e\u003cspan address=\"http://.[pmc.ncbi.nlm.nih]​\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eImam A, Obiesie S, English M, Schellenberg J (2022) Identifying gaps in global evidence for nurse staffing and patient care outcomes research in low/middle-income countries: an umbrella review. BMJ Open 12(10):e064050. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e.[pmc.ncbi.nlm.nih]​\u003c/span\u003e\u003cspan address=\"http://.[pmc.ncbi.nlm.nih]​\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAtinga BE et al (2025) Optimizing the quality of nursing care in public hospitals in low- and middle-income countries: an integrative literature review. Int J Nurs Stud\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eClark M et al Critical care nursing role in low and lower middle-income countries: a scoping review. J Intensive Care Soc\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Health Organization (2020) State of the world\u0026rsquo;s nursing 2020: investing in education, jobs and leadership. WHO, Geneva\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eInternational Council of Nurses (2025) State of the world\u0026rsquo;s nursing 2025 update. ICN, Geneva\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRahman A et al (2025) Artificial intelligence in nursing: a systematic review of current applications and future directions. Front Digit Health. ;XX:1666005.[frontiersin]​\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEl Arab RA et al (2025) Nurses\u0026rsquo; perceptions and use of artificial intelligence in clinical practice: a cross-sectional study. J Nurs Manag\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHacıalioğlu N et al (2025) Nurses\u0026rsquo; attitudes towards artificial intelligence: relationship with emotion regulation and cognitive flexibility. Int Nurs Rev\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePedersen M et al Nurses\u0026rsquo; intention to integrate AI into their practice: a technology acceptance study. JMIR Nurs\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCiecierski-Holmes T et al (2022) Artificial intelligence for low- and middle-income countries: a systematic scoping review. npj Digit Med\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHussein B (2026) AI implementation in low-resource healthcare settings: bridging the digital divide. Carleton University, Ottawa\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Health Organization (2024) Digitalization of health care in low- and middle-income countries. Bull World Health Organ XX:1\u0026ndash;12\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNature Editorial (2024) Artificial intelligence for low-income countries. Humanit Soc Sci Commun. ;11\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDornan M et al Every nurse an AI nurse: a framework for integrating AI into nursing practice. J Nurs Scholarsh\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGlobal reasons for missed nursing care: a systematic review. Int Nurs Rev\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMagnitude and reasons for missed nursing care among nurses working in public hospitals. Springer Pflege\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNursing management practices on the quality nursing care among nurses in a tertiary hospital. Int J Res Sci Innov\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStrategies to optimise the quality of nursing care for hospitalised patients in African settings: an integrative review. Afr J Nurs Midwifery\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eA systematic review of the application of artificial intelligence in nursing. J Med Internet Res / J Multidiscip Healthc (Dove).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHealthcare bias in AI (2025) a systematic literature review. In: Proceedings of the International Conference on Health Informatics\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRuksakulpiwat S et al The integration of AI in nursing: addressing current applications, challenges, and future directions. J Nurs Scholarsh\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eResearch Square How do I submit a preprint? Research Square Help Center; 2024 [cited 2026 Jan 27]. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://support.researchsquare.com\u003c/span\u003e\u003cspan address=\"https://support.researchsquare.com\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJoanna Briggs Institute (2020) JBI Manual for Evidence Synthesis. JBI, Adelaide\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShea BJ, Reeves BC, Wells G, Thuku M, Hamel C, Moran J et al (2017) AMSTAR 2: a critical appraisal tool for systematic reviews that include randomized or non-randomized studies of healthcare interventions. BMJ 358:j4008\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMissed nursing care (2021) in acute care hospital settings in low-income and middle-income countries: protocol. Wellcome Open Res 6:359\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTopol EJ (2019) High-performance medicine: the convergence of human and artificial intelligence. Nat Med 25(1):44\u0026ndash;56. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e.[pubmed.ncbi.nlm.nih]​\u003c/span\u003e\u003cspan address=\"http://.[pubmed.ncbi.nlm.nih]​\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"nursing care quality, low- and middle-income countries, artificial intelligence, nurse adaptation, digital health, missed nursing care","lastPublishedDoi":"10.21203/rs.3.rs-8704137/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8704137/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eNurses in low- and middle-income countries (LMICs) deliver care under chronic resource constraints, which compromise quality and shape how new technologies, including artificial intelligence (AI), can be adopted. Evidence on how nursing care quality and nurses\u0026rsquo; adaptation to AI intersect in LMICs remains fragmented.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe conducted a systematic review of empirical and review studies published between 2022 and 2025 that examined (1) the quality of nursing care in LMICs and/or (2) nurses\u0026rsquo; perceptions, readiness, and adaptation to AI in clinical practice. Searches of major databases identified ten eligible studies, including systematic and integrative reviews, scoping reviews, cross-sectional surveys, and conceptual frameworks addressing nursing care quality, AI in nursing, or AI implementation in LMIC health systems. Data were synthesized narratively across two domains: nursing care quality and AI-related attitudes and adaptation.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eStudies on care quality reported high levels of missed or delayed nursing care in LMIC acute and critical care settings, driven by chronic understaffing, inadequate skill mix, limited supplies, and weak governance and quality systems; staffing\u0026ndash;outcome research in LMICs was sparse and methodologically heterogeneous. AI-focused studies showed nurses were cautiously open to AI\u0026rsquo;s potential for efficiency, documentation, and decision support, yet concerned about job security, role erosion, liability, data privacy, and loss of human touch. Attitudes and readiness were influenced by emotion regulation, cognitive flexibility, and digital literacy, while implementation in LMICs was constrained by unreliable infrastructure, immature data systems, and limited technical support. Conceptual frameworks proposed baseline AI competencies for all nurses but emphasized phased, context-sensitive implementation and strong governance.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eNurses in LMICs are attempting to adapt to AI while fundamental deficits in nursing workforce and care environments remain unresolved. Strengthening staffing and basic quality infrastructure, embedding AI literacy in nursing education, involving nurses in digital-health planning, and establishing clear policies on data protection and accountability are essential to ensure that AI augments rather than undermines nursing care quality in LMICs.\u003c/p\u003e","manuscriptTitle":"Quality of Nursing Care and Adaptation to Artificial Intelligence in Low‑ and Middle‑Income Countries: A Systematic Review of Empirical Studies","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-03 05:06:02","doi":"10.21203/rs.3.rs-8704137/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"16a6c00f-51ba-4479-8b38-98df2c4a4b46","owner":[],"postedDate":"February 3rd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":61782806,"name":"Artificial Intelligence and Machine Learning"},{"id":61782807,"name":"Nursing"},{"id":61782808,"name":"Hospital Medicine"},{"id":61782809,"name":"Health Economics \u0026 Outcomes Research"},{"id":61782810,"name":"Health Policy"},{"id":61782811,"name":"Health Law"}],"tags":[],"updatedAt":"2026-02-03T05:06:02+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-03 05:06:02","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8704137","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8704137","identity":"rs-8704137","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}