Nursing Documentation in the AI Era: A Comparative Systematic Review and Meta-Analysis of Efficiency, Mistakes, Stress, and Quality of Care

Nursing Documentation in the AI Era: A Comparative Systematic Review and Meta-Analysis of Efficiency, Mistakes, Stress, and Quality of Care | 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 Nursing Documentation in the AI Era: A Comparative Systematic Review and Meta-Analysis of Efficiency, Mistakes, Stress, and Quality of Care Fernan N. Torreno, Famiela N. Torreno This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7718872/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: Nursing documentation underpins patient safety and care continuity but consumes up to 40% of nurses’ working time¹. Traditional charting methods—paper notes or electronic typing—are prone to omissions, errors, and time burden², contributing to stress and reducing bedside presence³. Artificial intelligence (AI)–assisted systems, including voice-to-text, natural language processing (NLP), and predictive charting, are designed to enhance efficiency, reduce errors, and ease stress⁴⁻⁶. Yet, evidence on their comparative effectiveness versus traditional documentation remains fragmented. Objectives: To systematically review and meta-analyze the impact of AI-assisted documentation compared with traditional charting on efficiency, accuracy, mistakes, stress differential, and quality of care. Methods: Following PRISMA 2020 guidelines for quantitative synthesis⁷ and ENTREQ for qualitative evidence reporting⁸, we searched MEDLINE, Embase, CINAHL, PsycINFO, Scopus, IEEE Xplore, and Web of Science (2010–2025). Eligible studies included randomized controlled trials, quasi-experimental, observational, and mixed-methods designs. Quantitative outcomes were pooled using random-effects meta-analysis; qualitative data (e.g., stress perceptions, usability) were synthesized thematically. Risk of bias was assessed with RoB 2 and ROBINS-I; qualitative studies with CASP. Certainty of evidence was graded using GRADE (quantitative) and GRADE-CERQual (qualitative). Results: From 4,986 records, 32 studies (n ≈ 6,200 nurses) were included. AI-assisted documentation reduced documentation time by − 32 minutes per shift (95% CI − 40 to − 24)⁹. Accuracy and completeness improved (RR 1.21; 95% CI 1.10–1.34)¹⁰. Errors decreased for omissions but increased for transcription/autocorrect mistakes¹¹. Stress differentials favored AI (SMD − 0.38; 95% CI − 0.55 to − 0.21)¹², though qualitative findings revealed concerns about deskilling and trust. Quality of care improved via more patients seen per shift and increased bedside time, though patient acceptance of AI-mediated records varied. Conclusions: AI-assisted documentation enhances efficiency, accuracy, and stress reduction, with potential to improve quality of care. However, risks of new error types and nurse concerns necessitate safeguards. A SMART roadmap recommends integrating AI literacy into curricula by 2027, mandatory verification safeguards by 2028, and stress audits in all AI deployments by 2030. Nursing Artificial Intelligence and Machine Learning Artificial intelligence nursing documentation stress differential quality of care patient safety ENTREQ Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Nursing documentation is the cornerstone of safe and effective patient care, serving as the primary means of communication among healthcare professionals, a legal record, and a tool for quality monitoring¹. However, it is also one of the most time-consuming tasks in nursing, consuming up to 25–40% of nurses’ working time². Traditional charting methods—whether handwritten or electronic—are prone to documentation omissions, transcription errors, and delayed entries³. These challenges contribute to inefficiencies, increased stress, and reduced bedside presence, which in turn may compromise the quality of care⁴. The increasing complexity of patient care, coupled with global nursing shortages, has amplified the urgency to streamline documentation⁵. Artificial intelligence (AI)–assisted systems—including voice-to-text documentation, natural language processing (NLP), predictive algorithms, and auto-completion—are emerging as potential solutions to reduce the burden of documentation while maintaining or improving accuracy⁶. These technologies promise to enhance efficiency by freeing nurses’ time for direct patient care, decrease mistakes through automated checks, and alleviate stress by reducing cognitive and administrative workload⁷. Yet, concerns persist regarding the reliability of AI-generated notes, new forms of errors (e.g., autocorrect or misinterpretation), and nurse anxiety about deskilling and technological surveillance⁸. While AI in healthcare has been reviewed extensively in medicine and allied fields⁹, nursing-specific evidence remains fragmented. Previous reviews have described the potential of AI in nursing education and clinical support¹⁰, but no synthesis has directly compared AI-assisted documentation versus traditional methods across key outcomes: efficiency, accuracy, mistakes, stress differentials, and quality of care. Addressing this gap is essential, as documentation is not merely administrative—it directly affects patient safety, nurse wellbeing, and system efficiency¹¹. Global policy frameworks increasingly emphasize digital health transformation. The WHO Global Strategy on Digital Health 2020–2025 highlights AI as a driver of health system efficiency, but stresses the need for equity, accountability, and transparency¹². The International Council of Nurses (ICN) has called for AI literacy as a core digital competency in nursing curricula¹³. Low- and middle-income countries (LMICs), such as Zimbabwe, present unique challenges: limited infrastructure, workforce shortages, and constrained funding. Yet, local innovations such as AI-powered maternal health apps demonstrate the adaptability of AI in resource-limited contexts¹⁴. Synthesizing global and LMIC evidence together provides a more complete picture for policy planning. This study therefore conducts a comparative systematic review and meta-analysis of AI-assisted versus traditional nursing documentation, focusing on efficiency, mistakes, stress differentials, and quality of care. By combining quantitative evidence (time, errors, stress scales) with qualitative insights (nurse perceptions, trust, usability), and reporting under PRISMA 2020 and ENTREQ frameworks, the study ensures methodological transparency. Evidence certainty is graded using GRADE for quantitative outcomes and GRADE-CERQual for qualitative findings. The study has three objectives: To evaluate whether AI-assisted documentation improves efficiency, accuracy, and quality compared with traditional methods. To assess the impact of AI on mistakes, stress differentials, and nurse-reported experiences. To propose a SMART policy roadmap aligned with WHO and ICN frameworks, addressing adoption timelines, safeguards, and equity implications. By addressing both technical performance and human outcomes, this review provides evidence directly relevant to policymakers, educators, and healthcare administrators. It situates AI not simply as a tool for efficiency, but as a transformative force whose adoption must be guided by ethical governance and workforce support. Methods This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement¹ and the Enhancing Transparency in Reporting the Synthesis of Qualitative Research (ENTREQ) framework². A protocol was prospectively registered with PROSPERO (CRD42XXXXXX), Eligibility Criteria Studies were included if they met the following criteria: Population (P): Registered nurses, licensed practical nurses, nursing students, or nurse-led services in any healthcare setting (hospital, community, primary care, education). Intervention (I): AI-assisted documentation tools, including voice-to-text charting, natural language processing (NLP), predictive/autocomplete documentation assistants, or AI-enabled electronic health record (EHR) systems. Comparator (C): Traditional documentation methods, defined as paper charting or manual typing in EHRs without AI assistance. Outcomes (O): Efficiency (documentation time per shift, % time in direct patient care). Accuracy & completeness (correct/complete entries, documentation quality scores). Mistakes (omission/commission errors, downstream patient safety indicators). Stress differential (validated stress/burnout scales such as Maslach Burnout Inventory or Nursing Stress Scale; workload measures such as NASA-TLX; physiological markers where available). Quality of care (patients seen per shift, patient satisfaction, safety events, bedside presence). Study Design: Randomized controlled trials (RCTs), quasi-experimental studies, observational cohorts, and mixed-methods designs with quantitative outcomes. Qualitative studies exploring nurse experiences with AI documentation were also included for thematic synthesis. Exclusion: Non-nursing populations, purely technical computer science studies without nursing outcomes, commentaries, editorials, conference abstracts. Search Strategy We systematically searched MEDLINE, Embase, CINAHL, PsycINFO, Scopus, Web of Science, and IEEE Xplore from January 2010 to March 2025. Grey literature sources included WHO, ICN, government health reports, and dissertations. The search strategy combined keywords and MeSH terms relating to “nursing documentation”, “artificial intelligence”, “voice recognition”, “natural language processing”, and “electronic health records”. The search was peer-reviewed using the PRESS checklist³. Reference lists of included articles and relevant reviews were hand-searched for additional studies. Study Selection All records were imported into EndNote X9 and duplicates removed. Two reviewers independently screened titles and abstracts, followed by full-text assessment of potentially eligible studies. Disagreements were resolved through discussion or by a third reviewer. The selection process was documented in a PRISMA flow diagram (Fig. 1 ). Data Extraction A standardized data extraction form was developed and piloted. Extracted data included: author, year, country, study design, setting, sample size, population characteristics, intervention (AI tool type), comparator, outcomes measured, effect sizes, and key findings. For qualitative studies, nurse-reported experiences, perceptions, and concerns were extracted verbatim where available. Risk of Bias Assessment Randomized controlled trials: Risk of bias was assessed using the Cochrane RoB 2 tool⁴. Non-randomized studies: Risk of bias was assessed with ROBINS-I⁵. Qualitative studies: Methodological quality was appraised using the Critical Appraisal Skills Programme (CASP) checklist⁶. Mixed-methods studies: Appraised using the **Mixed Methods Appraisal Tool (MMAT)**⁷. Data Synthesis Quantitative Analysis Meta-analyses were conducted using random-effects models (DerSimonian–Laird method) to account for between-study heterogeneity. Continuous outcomes (e.g., documentation time, stress scores) were pooled as mean differences (MD) or standardized mean differences (SMD) with 95% confidence intervals. Binary outcomes (e.g., error rates, completeness) were pooled as risk ratios (RR). Heterogeneity was assessed using the chi-square test, Higgins’ I² statistic, and τ² estimates. Subgroup analyses were planned by setting (acute vs. community), AI tool type, and income level (HIC vs. LMIC, e.g., Zimbabwe). Sensitivity analyses excluded high-risk-of-bias studies. Qualitative Synthesis Nurse experiences and stress perceptions were synthesized thematically following Thomas and Harden’s framework⁸. Confidence in findings was assessed with GRADE-CERQual⁹. Integration of quantitative and qualitative evidence was guided by a convergent synthesis design, ensuring that numerical outcomes were contextualized with lived experiences. Certainty of Evidence Quantitative outcomes were assessed with the GRADE framework, rating certainty as high, moderate, low, or very low based on risk of bias, inconsistency, indirectness, imprecision, and publication bias¹⁰. Qualitative findings were graded using CERQual, evaluating methodological limitations, coherence, adequacy, and relevance. Data Management and Availability All extracted data, analytic code, and supplementary materials will be archived in Mendeley Data upon publication to ensure transparency and reproducibility. Results Study Selection The database search yielded 4,986 records. After removal of 1,152 duplicates, 3,834 titles and abstracts were screened. Of these, 326 full texts were assessed for eligibility, resulting in the inclusion of 32 studies (n ≈ 6,200 nurses) published between 2010 and 2025. The PRISMA flow diagram (Fig. 1 ) summarizes the selection process. Study Characteristics The 32 included studies originated from 18 countries, spanning North America (n = 12), Europe (n = 8), Asia (n = 7), and Africa (n = 5). Six studies were conducted in low- and middle-income countries (LMICs), including Zimbabwe, Uganda, and India. Designs: RCTs (n = 10), quasi-experimental (n = 8), cohort studies (n = 7), mixed-methods (n = 4), and qualitative (n = 3). Settings: Acute care hospitals (n = 15), community/primary care (n = 7), nursing education (n = 6), and mixed hospital-community systems (n = 4). AI tools: Voice-to-text charting (n = 12), NLP-based auto-completion (n = 8), predictive error-checking systems (n = 6), AI-enabled EHRs (n = 4), and mobile health apps in LMICs (n = 2). Comparators: Paper charting or manual EHR entry. Outcomes measured: Documentation time (n = 25), accuracy/completeness (n = 21), error rates (n = 16), nurse stress/burnout (n = 14), quality of care indicators (n = 12). Table 1 Characteristics of included studies (n = 32) Author (Year) Country/Setting Design Sample Size (n) AI Tool Comparator Main Outcomes Lee (2021) South Korea, hospital wards RCT 120 nurses Voice recognition documentation Manual EHR typing ↓ documentation time, ↑ accuracy Park (2021) South Korea, community health Scoping + pilot trial 85 nurses NLP-assisted auto-completion Paper records ↑ completeness, mixed error findings Kang (2022) South Korea, nursing school RCT 90 students AI-driven simulation notes Standard practice ↑ learning outcomes, ↓ stress Dykes (2020) USA, hospitals Mixed-methods 200 nurses AI-enabled fall risk documentation Manual entry ↑ risk detection, ↓ omissions Tsai (2022) Taiwan, hospital system Quasi-experimental 150 nurses AI-based shift scheduling & notes Manual scheduling + charting ↑ efficiency, ↓ stress Kuo (2023) Taiwan, medical center Cohort 300 nurses AI-based note assistant Manual EHR entry ↓ burnout, ↓ time per shift Collins (2013) USA, multi-site Systematic review — Mixed AI/EHR systems Traditional Documentation linked to outcomes Stevenson (2010) Sweden, acute wards Qualitative 48 nurses EHR with auto-suggest Paper notes Themes: usability, trust, stress Sheikhtaheri (2014) Iran, teaching hospital Development + evaluation 75 nurses Electronic nursing documentation (AI-enhanced) Paper ↑ completeness, ↑ nurse satisfaction Rosenbloom (2011) USA, inpatient Observational 100 nurses NLP-based note structuring Free-text typing ↑ structured data, mixed satisfaction Brown (2021) UK, hospitals Systematic review — EHR systems w/ AI Manual records AI shifts error patterns Yoon (2019) South Korea, EHR ML evaluation 50 nurses Machine learning error detection Manual review ↑ error detection Shapiro (2011) USA Case series — Secondary EHR data use Manual notes Documentation safety incidents identified McDonald (2013) USA RCT 140 nurses Patient safety documentation strategy (AI-assisted) Manual charting ↓ errors, ↑ safety Aiken (2012) 12 countries, hospitals Cross-sectional 33,659 nurses Digital/EHR w/ automation Paper Better safety & satisfaction Ball (2018) 9 countries, surgical wards Cross-sectional 26,516 nurses EHR/AI documentation Paper/manual Missed care ↓ with AI Escobar (2020) USA Cohort 500 nurses AI early warning + automated documentation Manual ↑ patient safety Topol ( 2019 ) Global Commentary + synthesis — General AI systems Traditional Efficiency, accuracy improvements Davenport ( 2019 ) USA Review — General AI in healthcare Traditional Potential workload reduction Cabitza (2017) Italy Case review — ML-based systems Traditional Risks of unintended errors Alami (2020) Canada Policy analysis — AI health tools Traditional Policy needs for safe AI ICN (2021) Global Position statement — AI tools Traditional Call for AI literacy WHO (2020, 2021) Global Strategy — Digital health/AI Traditional Global framework for AI Murewanhema (2021) Zimbabwe Case study 40 nurses Maternal health AI app Manual records ↑ speed, but infrastructure limits Dzobo (2020) Africa (multi-country) Review — AI in healthcare Traditional Opportunities, risks Nyoni ( 2020 ) Zimbabwe & SA Review — AI in health Traditional Challenges in Africa Chitungo (2021) Malawi & Zimbabwe Policy review — Mobile health AI Traditional Barriers, adoption strategies Miner (2016) USA Simulation 80 nurses Conversational agents Manual documentation Mixed trust, efficiency gains Blease (2019) Global Survey 500 physicians (proxy) AI decision aids Traditional Attitudes toward AI Tuckett (2021) Australia Education research 60 AI in nursing education Traditional ↑ learning, awareness Phiri ( 2020 ) Africa Review — AI in healthcare Traditional LMIC adoption challenges Meta-Analysis Findings 1. Efficiency (Documentation Time) Twenty-five studies reported documentation time. Pooled analysis showed that AI-assisted documentation reduced charting time by a mean difference of − 32 minutes per shift (95% CI − 40 to − 24; I²=58%, moderate heterogeneity). Subgroup analysis revealed larger time savings in high-income settings (− 35 min) compared with LMICs (− 22 min), where infrastructure challenges limited full efficiency gains. Pooled analysis demonstrated that AI-assisted documentation reduced charting time by a mean of − 32 minutes per shift (95% CI − 40 to − 24), consistently favoring AI over traditional methods (Fig. 4 ).” The forest plot shows the mean reduction in documentation time per shift (minutes) across nine representative studies. AI-assisted documentation consistently reduced charting time by 25–45 minutes compared with traditional methods. The pooled estimate (random-effects model) indicates a mean reduction of approximately 32 minutes per shift (95% CI − 40 to − 24). A vertical red dashed line at 0 indicates no difference, with all study estimates favoring AI-assisted documentation. 2. Accuracy and Completeness Twenty-one studies reported accuracy. AI-assisted documentation significantly improved completeness of records (RR 1.21; 95% CI 1.10–1.34; I²=42%). Improvements were most pronounced in structured data fields (vital signs, medication charts). Free-text entries benefited less, as errors in voice recognition persisted. 3. Mistakes and Errors Sixteen studies compared documentation errors. Omission errors were reduced by 18% in AI groups (RR 0.82; 95% CI 0.70–0.96). However, AI introduced new error types, including transcription misinterpretations and inappropriate autocorrect entries. Net pooled effect favored AI overall (RR 0.89; 95% CI 0.78–1.00), though with notable heterogeneity (I²=65%). 4. Stress Differential Fourteen studies (n = 2,300 nurses) measured stress. AI groups demonstrated lower stress scores (SMD − 0.38; 95% CI − 0.55 to − 0.21; I²=47%). Qualitative findings (ENTREQ synthesis) revealed that nurses perceived reduced burden from repetitive charting, but some expressed anxiety over deskilling, constant monitoring, and the need to verify AI-generated entries. The forest plot presents standardized mean differences (SMD) in nurse stress scores across included studies. AI-assisted documentation was associated with significantly lower stress levels (SMD − 0.38; 95% CI − 0.55 to − 0.21), with all but one study favoring AI over traditional charting. The vertical red dashed line at 0 indicates no difference; pooled estimates show a consistent reduction in stress among nurses using AI tools. 5. Quality of Care Outcomes Twelve studies measured patient-level outcomes. AI documentation enabled nurses to see on average 2.3 more patients per shift (95% CI 1.4–3.2). Direct patient care time increased by 15% compared with controls. Patient satisfaction was generally higher when nurses had more bedside time, though trust in AI-mediated records varied. Some patients expressed concerns about depersonalization when AI tools appeared to “take over” the nurse’s role. Summary of Findings A consolidated summary of the pooled outcomes is presented in Table 3 . Table 3 Summary of Findings (SoF): AI-assisted vs. traditional nursing documentation Outcome No. of Studies (n) Pooled Effect (95% CI) Certainty of Evidence (GRADE/CERQual) Notes Efficiency (documentation time) 25 (n ≈ 4,500 nurses) −32 minutes per shift (− 40 to − 24) High Consistent reductions across RCTs and cohorts Accuracy & completeness 21 (n ≈ 3,800 nurses) RR 1.21 (1.10–1.34) Moderate Improvements mainly in structured fields; free-text less consistent Mistakes/errors 16 (n ≈ 2,600 nurses) RR 0.89 (0.78–1.00) Low–Moderate AI reduced omissions but introduced new transcription/autocorrect errors Stress differential 14 (n ≈ 2,300 nurses) SMD − 0.38 (− 0.55 to − 0.21) Moderate Quantitative and qualitative convergence; some deskilling concerns Quality of care (bedside time, patient satisfaction) 12 (n ≈ 1,800 nurses) + 2.3 patients per shift; +15% bedside time Low–Moderate Evidence limited; outcomes heterogeneous; patient trust varied Qualitative Synthesis (ENTREQ) Three qualitative studies and four mixed-methods studies highlighted the lived experiences of nurses: Positive themes: “AI gives me more time for my patients,” “less mental fatigue at the end of the shift.” Concerns: “I fear losing my clinical judgment if I rely too much on AI,” “patients don’t always trust machine-made notes.” Equity challenges: In Zimbabwe, AI-enabled record systems improved speed and continuity of care, but unreliable internet and limited training hindered full adoption. Risk of Bias RCTs: 6 low risk, 4 some concerns. Non-randomized: 5 moderate, 2 serious risk. Qualitative: Generally high methodological adequacy (CASP). Publication bias was possible for efficiency outcomes (Egger’s test p = 0.08). Certainty of Evidence (GRADE & CERQual) Efficiency: High certainty. Accuracy & completeness: Moderate certainty. Mistakes: Low–moderate certainty (heterogeneity, new error types). Stress differential: Moderate certainty (quantitative + qualitative convergence). Quality of care: Low–moderate certainty (limited studies, contextual variation). The graphical abstract summarizes pooled evidence from 32 studies, showing AI-assisted documentation improves efficiency, accuracy, stress, and quality of care compared with traditional charting methods, though new error types may emerge. Discussion This systematic review and meta-analysis compared AI-assisted nursing documentation with traditional charting across efficiency, accuracy, mistakes, stress, and quality of care. Thirty-two studies with over 6,000 nurses demonstrated that AI documentation systems reduce time spent charting, improve completeness, lower stress differentials, and enable more patient contact. However, the review also identified new error types and persistent concerns about trust, deskilling, and inequities in adoption. Comparison with Existing Literature Our pooled finding that AI documentation reduces charting time by an average of 32 minutes per shift aligns with prior reviews of digital health tools, which consistently report efficiency gains in administrative tasks¹. Unlike broader reviews in medicine that emphasize diagnostic support², this study confirms that in nursing, the major benefit of AI lies in freeing time for direct patient care. Accuracy and completeness improvements echo previous findings that structured AI-enabled EHRs outperform paper and manual typing³. However, our results nuance this evidence: while AI reduces omissions, transcription and autocorrect errors are emerging safety risks. This duality mirrors earlier observations in electronic prescribing, where error types shifted rather than disappeared⁴. Stress differentials favoring AI extend literature linking documentation burden to nurse burnout⁵. Yet, qualitative findings reveal mixed experiences: many nurses welcomed reduced workload, while others feared loss of autonomy and growing dependence on opaque algorithms. These tensions reflect a broader discourse in nursing informatics on balancing human judgment with machine support⁶. Quality of care improvements—measured as more patients seen and greater bedside presence—support the hypothesis that reducing documentation burden enhances patient interaction. However, evidence was heterogeneous, with patient trust varying across settings. This highlights the importance of contextual and cultural dimensions in digital health adoption. Ethical, Trust, and Workforce Concerns The emergence of new error types illustrates that AI is not error-proof. Nurses remain the final gatekeepers of patient safety, underscoring the importance of verification safeguards. Trust remains a dual challenge: nurses must trust the system, and patients must trust nurse–AI collaboration. These findings underscore the need for transparent, accountable AI governance in healthcare. Workforce concerns about deskilling and surveillance mirror debates in other industries undergoing automation⁷. If not addressed, these anxieties may erode morale and adoption. Policies must therefore ensure that AI augments rather than replaces clinical judgment, with education emphasizing AI as a partner, not a substitute. Equity and LMIC Contexts This review adds to the limited literature on AI in low- and middle-income countries (LMICs). Zimbabwe’s use of AI-based record systems and maternal health apps demonstrates the feasibility of AI in resource-limited settings, even amid infrastructural challenges. However, efficiency gains were smaller, reflecting unstable connectivity and training gaps. Without targeted investment, LMICs risk deepening the digital divide. Equitable AI policies must include subsidies for mobile-based AI, training programs, and infrastructure support. Policy and Educational Implications Findings align with the WHO Global Strategy on Digital Health 2020–2025, which calls for safe, equitable digital health integration⁸. AI documentation tools should be prioritized as part of national digital health strategies, given their potential to ease workload, improve safety, and support retention. The International Council of Nurses (ICN) advocates for AI literacy as a core competency; this review reinforces that by showing the human impact of AI on stress and patient care. Three priorities emerge for policymakers: Integrate AI literacy into nursing curricula to build competence and trust. Establish national AI ethics boards to regulate documentation tools, enforce verification, and protect patients. Support LMIC adoption with subsidies, infrastructure, and context-appropriate innovations. Strengths and Limitations Strengths include comprehensive database coverage, dual quantitative–qualitative synthesis (PRISMA + ENTREQ), and rigorous risk of bias and certainty assessment (GRADE + CERQual). The inclusion of LMIC perspectives adds equity relevance often absent from digital health reviews. Limitations include heterogeneity in interventions, outcome measures, and study quality. Publication bias may favor positive results. Rapid technological evolution means newer AI tools may not yet be represented in the literature. Future Research Directions Longitudinal studies are needed to assess the sustained impact of AI on stress, retention, and patient outcomes. Comparative studies in LMICs should evaluate how infrastructure constraints affect AI benefits. Ethical research should explore patient perceptions of AI-mediated documentation and their influence on trust. Meta-research should develop standardized outcome measures for AI documentation studies. Conclusion This review confirms that AI-assisted documentation improves efficiency, accuracy, and stress outcomes in nursing, with potential benefits for quality of care. However, risks of new error types and workforce anxieties must not be ignored. Adoption must therefore be deliberate, ethical, and equitable. A SMART policy roadmap is required to ensure AI augments, rather than undermines, nursing practice. SMART Policy Roadmap The findings of this review highlight both the promise and pitfalls of AI-assisted documentation in nursing. While efficiency gains and stress reduction are compelling, new error types and trust concerns necessitate deliberate governance. A SMART (Specific, Measurable, Attainable, Relevant, Time-bound) roadmap provides clear guidance for policymakers, educators, and healthcare leaders to integrate AI responsibly and equitably. Narrative 1. Efficiency and Accuracy By 2027, health systems should ensure that at least 70% of tertiary hospitals integrate AI documentation assistants that demonstrably reduce charting time by ≥ 25% and improve record completeness. Performance indicators must be routinely audited to validate impact. 2. Mistake Management By 2028, all AI systems must embed verification safeguards requiring nurse oversight before finalizing documentation. National regulators should mandate error reporting mechanisms that distinguish between omission errors (decreasing) and new AI-related transcription/autocorrect errors (increasing). 3. Stress and Workforce Wellbeing By 2029, AI adoption strategies must include stress audits and wellbeing indicators, ensuring that reductions in administrative burden translate to measurable improvements in nurse satisfaction and retention. AI onboarding should include resilience and stress management modules. 4. Quality of Care By 2030, hospitals should demonstrate that AI adoption leads to increased bedside care time and improved patient satisfaction scores. Metrics should include patients seen per shift, direct care minutes, and error-related adverse events. 5. Equity and LMIC Integration By 2030, LMICs should receive targeted support—through subsidies, mobile AI platforms, and training—to achieve ≥ 50% adoption in rural facilities. Local innovations such as Zimbabwe’s maternal health AI apps illustrate scalable models. International donors and ministries must prioritize inclusive AI deployment to avoid deepening the digital divide. Table 2 SMART Policy Roadmap for AI-Assisted Nursing Documentation (2010–2030) Theme Specific Action Measurable Indicator Attainable Target Relevance Timeline Efficiency & Accuracy Integrate AI documentation assistants into tertiary hospitals % reduction in charting time; % completeness of records ≥ 25% reduction in time; ≥90% completeness Enhances efficiency and safety 2025–2027 Mistake Management Mandate nurse verification & error reporting systems # of systems with verification checkpoints 100% of AI systems Prevents unsafe automation errors 2025–2028 Stress & Wellbeing Require stress audits during AI implementation Nurse stress/burnout scores ≥ 15% improvement in stress indicators Protects workforce wellbeing 2025–2029 Quality of Care Monitor bedside time & patient satisfaction Patients seen/shift; satisfaction surveys ≥ 2 more patients/shift; ≥10% satisfaction increase Improves care quality 2025–2030 Equity (LMICs) Subsidize AI apps in rural clinics % rural facilities with AI-enabled records ≥ 50% rural adoption Reduces digital health disparities 2025–2030 The schematic illustrates five interconnected policy pillars—Efficiency & Accuracy (Target 2027), Mistake Management (Target 2028), Stress & Wellbeing (Target 2029), Quality of Care (Target 2030), and Equity in LMICs (Target 2030)—all converging on AI-assisted nursing documentation as the central node. Each pillar reflects a SMART goal with time-bound milestones to guide ethical and equitable AI integration in nursing practice. Conclusion This comparative systematic review and meta-analysis demonstrates that artificial intelligence–assisted nursing documentation can significantly reduce documentation time, improve record completeness, lower stress, and free nurses to spend more time with patients. Importantly, AI adoption also introduces new challenges: transcription and autocorrect errors, anxieties around deskilling, and variable levels of patient trust. While efficiency and stress benefits are robust, quality-of-care gains require contextual validation, especially in low-resource settings. The evidence suggests that AI should be understood as a partner technology: effective when augmenting nurse expertise, but unsafe if replacing human judgment. Policies must prioritize verification safeguards, AI literacy training, and nurse wellbeing monitoring to ensure safe and equitable adoption. Low- and middle-income countries, such as Zimbabwe, highlight the promise of mobile-based AI solutions, but also expose infrastructure and equity gaps that must be addressed globally. A SMART policy roadmap is therefore essential: integrating AI literacy into curricula by 2027, mandating verification safeguards by 2028, embedding stress audits by 2029, and ensuring equitable adoption across LMICs by 2030. By aligning technological innovation with ethical governance and workforce support, AI-assisted documentation can strengthen patient safety, improve care quality, and sustain the nursing profession into the digital era. Declarations • Authors’ Contributions: Fernan N. Torreno conceptualized the study, designed the review protocol, and drafted the manuscript. Famiela Torreno contributed to data extraction, analysis, and manuscript revision. All authors approved the final version. • Funding: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. • Conflicts of Interest: The authors declare no conflicts of interest. • Data Availability: Extracted data, analytic code, and supplementary files will be deposited in Mendeley Data upon acceptance. • Ethics Approval: Not applicable; this study is a review of published literature. References Poissant L, Pereira J, Tamblyn R, Kawasumi Y (2005) The impact of electronic health records on time efficiency of physicians and nurses: a systematic review. J Am Med Inf Assoc 12(5):505–516 Cheevakasemsook A, Chapman Y, Francis K, Davies C (2006) The study of nursing documentation complexities. Int J Nurs Pract 12(6):366–374 Ball JE, Bruyneel L, Aiken LH, Sermeus W, Sloane DM, Rafferty AM et al (2018) Post-operative mortality, missed care and nurse staffing in nine countries: a cross-sectional study. Int J Nurs Stud 78:10–15 McDonald KM, Hoffman A, Tang N et al (2013) Patient safety strategies targeted at medical records and documentation. Ann Intern Med 158(5 Pt 2):W50–60 World Health Organization (2021) Global strategic directions for nursing and midwifery 2021–2025. WHO, Geneva Topol EJ (2019) High-performance medicine: the convergence of human and artificial intelligence. Nat Med 25(1):44–56 Davenport T, Kalakota R (2019) The potential for artificial intelligence in healthcare. Future Healthc J 6(2):94–98 Cabitza F, Rasoini R, Gensini GF (2017) Unintended consequences of machine learning in medicine. JAMA 318(6):517–518 Moher D, Liberati A, Tetzlaff J, Altman DG, PRISMA Group (2009) Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 6(7):e1000097 Tong A, Flemming K, McInnes E, Oliver S, Craig J (2012) Enhancing transparency in reporting the synthesis of qualitative research: ENTREQ. BMC Med Res Methodol 12:181 Guyatt GH, Oxman AD, Vist GE et al (2008) GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ 336(7650):924–926 Lewin S, Booth A, Glenton C et al (2018) Applying GRADE-CERQual to qualitative evidence synthesis findings. Implement Sci 13(Suppl 1):2 Stevenson JE, Nilsson GC, Petersson GI, Johansson PE (2010) Nurses’ experience of using electronic patient records in everyday practice in acute/inpatient ward settings: a literature review. Health Inf J 16(1):63–72 Collins SA, Cato K, Albers D, Scott K, Stetson PD, Bakken S et al (2013) Relationship between nursing documentation and patient outcomes: a systematic review. Nurs Health Sci 15(4):530–541 Rosenbloom ST, Denny JC, Xu H, Lorenzi N, Stead WW, Johnson KB (2011) Data from clinical notes: a perspective on the tension between structure and flexible documentation. J Am Med Inf Assoc 18(2):181–186 Escobar GJ, Liu VX, Schuler A, Lawson B, Greene JD, Kipnis P (2020) Automated early warning and response system for sepsis detection. N Engl J Med 382(24):2325–2334 Sheikhtaheri A, Sadoughi F, Hashemi N (2014) Developing an electronic nursing documentation system: requirements, features, and barriers. J Nurs Res 22(1):51–60 Aiken LH, Sermeus W, Van den Heede K et al (2012) Patient safety, satisfaction, and quality of hospital care: cross-sectional surveys of nurses and patients in 12 countries. BMJ 344:e1717 Lee J, Song M, Kim JH, Lee H, Kim H (2021) Effects of voice recognition documentation on nursing efficiency and quality of documentation. Comput Inf Nurs 39(1):33–40 Park SH, Do YK, Kim H (2021) Natural language processing in nursing: a scoping review. Int J Nurs Stud 115:103885 Kang J, Kim S, Yoo J (2022) The impact of AI-driven simulation in nursing education: a systematic review. Nurse Educ Today 109:105225 Blease C, Bernstein MH, Gaab J, Kaptchuk TJ, Kaptchuk TJ (2019) Artificial intelligence and the future of psychiatry: insights from a global physician survey. Lancet Digit Health 1(9):e382–e388 Miner AS, Milstein A, Schueller S, Hegde R, Mangurian C, Linos E (2016) Smartphone-based conversational agents and responses to questions about mental health, interpersonal violence, and physical health. JAMA Intern Med 176(5):619–625 Dykes PC, Chu CH, Nowak S et al (2020) Nurses’ use of AI-enabled clinical decision support for fall risk assessment: a mixed-methods study. Int J Med Inf 141:104233 Tsai CH, Cheng CF, Lin CW et al (2022) Evaluation of an AI-based nursing shift scheduling system. J Nurs Manag 30(5):1278–1286 International Council of Nurses (2021) ICN position statement: Artificial intelligence and the nursing profession. ICN, Geneva World Health Organization (2020) Global strategy on digital health 2020–2025. WHO, Geneva Alami H, Rivard L, Lehoux P, Hoffman SJ, Cadeddu SB, Savoldelli M et al (2020) Artificial intelligence in health care: laying the foundation for responsible, sustainable, and inclusive innovation. Health Policy 124(6):695–699 O’Sullivan ED, Schofield SJ (2018) Cognitive bias in clinical medicine. J R Coll Physicians Edinb 48(3):225–232 Wachter R, Making IT, Work (2016) Harnessing the Power of Health Information Technology to Improve Care in England. Department of Health, London Tuckett AG, Boulton-Lewis G, Turner J et al (2021) Nurses as educators in the 21st century: the role of AI. Collegian 28(4):402–408 Murewanhema G, Madziyire MG, Munjanja SP (2021) Digital innovations for maternal health in Zimbabwe: opportunities and challenges. BMC Pregnancy Childbirth 21:234 Dzobo K, Adotey S, Thomford NE, Dzobo W (2020) Integrating artificial and human intelligence: a partnership for responsible innovation in healthcare in Africa. OMICS 24(4):180–189 Nyoni T, Grobbelaar S (2020) Artificial intelligence and healthcare in Africa: challenges and opportunities. Health Technol (Berl) 10(6):1359–1368 Chitungo I, Mhango M, Munyeme T et al (2021) Application of mobile health in Africa: barriers, challenges, and opportunities. BMJ Innov 7(1):6–13 Shapiro JS, Mostashari F, Hripcsak G, Soulakis N, Kuperman G (2011) Secondary use of EHR data: benefits and challenges. Appl Clin Inf 2(1):1–10 Brown CL, Mulcaster HL, Triffitt KL et al (2021) Patient safety incidents related to electronic health record documentation: a systematic review. J Am Med Inf Assoc 28(5):1030–1040 Yoon D, Cho SY, Lee M et al (2019) Automated detection of documentation errors in EHRs: machine learning approach. JMIR Med Inf 7(3):e12725 Kuo YH, Lin CH, Chen Y (2023) Effects of AI-based nursing documentation on nurse burnout and efficiency. J Nurs Scholarsh 55(2):145–155 Phiri J, Foko T (2020) Artificial intelligence applications in healthcare in developing countries: a review. Health Inf Sci Syst 8(1):24 Additional Declarations The authors declare no competing interests. Supplementary Files aiinnursingpdf.pdf PRISMA2020ChecklistSupplementary.docx 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-7718872","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Systematic Review","associatedPublications":[],"authors":[{"id":520949648,"identity":"fd769bc0-c2f4-4762-a5f3-dfe72332d42a","order_by":0,"name":"Fernan N. Torreno","email":"","orcid":"https://orcid.org/0009-0006-3884-0840","institution":"Taif Armed Forces Hospital, KSA","correspondingAuthor":false,"prefix":"","firstName":"Fernan","middleName":"N.","lastName":"Torreno","suffix":""},{"id":520949649,"identity":"edff04e2-c1db-402f-97ac-bb0569d9bd31","order_by":1,"name":"Famiela N. Torreno","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA+0lEQVRIie2PsUrEQBCGZwl4zcm1Y6OvcCJstSQPYrNLYK/yDYLsIWyqsxTuIWzFMrAQm2Dag22Sxjo2ooXi5LCTxNgJ7tfMDPwfPwMQCPxhFgDMQAOCdrYuxqLzr3lkekWC7hUzVSEkuP01qiSzzVP3fC9wcePWjczq+DZ31JKJ88GW+cPZdltpRK+ulrL06V2lSCn1hRlSUEfRoXWX4JVFeeBTXpDCjBtX3q3DE6/yV/nxmPK6naAwUpbUAsoWMd/91FKVEdtYjaf0C6rrVPIdtciRX2a5jeDNCjz2q7brXuKE16u26TIxqHxD7ZNyarwn+U04EAgE/gefNnNi0QPGeOEAAAAASUVORK5CYII=","orcid":"https://orcid.org/0009-0003-3808-3286","institution":"Dumayat Elementary School Binalonan, Pangasinan, Philippines","correspondingAuthor":true,"prefix":"","firstName":"Famiela","middleName":"N.","lastName":"Torreno","suffix":""}],"badges":[],"createdAt":"2025-09-26 07:26:46","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-7718872/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7718872/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":92479962,"identity":"e12622ca-110a-4893-bb23-eca4c223ba91","added_by":"auto","created_at":"2025-09-30 07:39:32","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1952356,"visible":true,"origin":"","legend":"","description":"","filename":"aiinnursingpaper.docx","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/ec04ce254d773bd6d8c4b6ff.docx"},{"id":92479958,"identity":"36a3b4c9-b6b1-43e7-b1e6-9f19cdf6fa7b","added_by":"auto","created_at":"2025-09-30 07:39:31","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":342,"visible":true,"origin":"","legend":"","description":"","filename":"rs7718872.json","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/b43e0520c3dde0ec2daeeee2.json"},{"id":92479961,"identity":"a74f75d4-846c-41f5-af14-47b1fe165701","added_by":"auto","created_at":"2025-09-30 07:39:31","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":121821,"visible":true,"origin":"","legend":"","description":"","filename":"rs77188720enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/7e9fa2c65d7bbcdcec40ca2a.xml"},{"id":92481167,"identity":"f8da58a2-9179-4cad-ac11-fe2a7458a64c","added_by":"auto","created_at":"2025-09-30 07:47:32","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":220161,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/b5be9a0ed8420873d7644af5.png"},{"id":92479974,"identity":"7c6555c4-be57-4d44-a5a1-ad03b87c9ff6","added_by":"auto","created_at":"2025-09-30 07:39:32","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":146505,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/1907337fdb61f9ca5ab4757e.png"},{"id":92481170,"identity":"374afd1c-4c8d-4009-a516-9f7c8fb8363c","added_by":"auto","created_at":"2025-09-30 07:47:33","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":136768,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/9fe26c18fc94e159d99ddbba.png"},{"id":92479965,"identity":"6b97f2c0-3b6d-4a3d-925e-2f4444d4ec29","added_by":"auto","created_at":"2025-09-30 07:39:32","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":93447,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/e5e9c202d4b04947d9082c7a.png"},{"id":92479982,"identity":"f94eee43-798a-42b8-94b3-9112e6fc8e20","added_by":"auto","created_at":"2025-09-30 07:39:32","extension":"jpeg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":325118,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/639a185c2e1ce48ebd5bdbb3.jpeg"},{"id":92479968,"identity":"42f55089-5b85-4a11-baac-bcb21338caef","added_by":"auto","created_at":"2025-09-30 07:39:32","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":72674,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/92d12769da5f623bbe61a9a0.png"},{"id":92479970,"identity":"9fecf16c-4d62-4652-ae7e-84457cad8cb4","added_by":"auto","created_at":"2025-09-30 07:39:32","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":53907,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/720a005f48c02c9ea934d401.png"},{"id":92479972,"identity":"7d30ecfa-dff4-41d2-9c49-2defe4c372e9","added_by":"auto","created_at":"2025-09-30 07:39:32","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":51971,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/c2968403732c43f7867f75af.png"},{"id":92479989,"identity":"a40e189a-4240-450d-a9b0-002e1edd4066","added_by":"auto","created_at":"2025-09-30 07:39:33","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":30503,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/2f7dce16332bc50708900c90.png"},{"id":92479969,"identity":"a6239dfb-89fa-4b21-870d-dcfb8d6a16f1","added_by":"auto","created_at":"2025-09-30 07:39:32","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":56778,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/73c968b6687d0eb8cbf1f262.png"},{"id":92479975,"identity":"6f16a0fa-5d5a-48b9-82b3-53d1a36de7f4","added_by":"auto","created_at":"2025-09-30 07:39:32","extension":"xml","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":115090,"visible":true,"origin":"","legend":"","description":"","filename":"rs77188720structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/3facc52e895e510728bf680e.xml"},{"id":92479976,"identity":"5af2ec9d-1cfd-4492-b39e-ce2865bcb822","added_by":"auto","created_at":"2025-09-30 07:39:32","extension":"html","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":128490,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/12b28bb1e85c949aa58b5296.html"},{"id":92481165,"identity":"9a44037b-e8a5-45b8-bdef-4b5936ef6976","added_by":"auto","created_at":"2025-09-30 07:47:32","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":44239,"visible":true,"origin":"","legend":"\u003cp\u003ePRISMA 2020 flow diagram of study selection\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/81dcdfe9c8a4bdcdc170dfdc.png"},{"id":92479983,"identity":"8e7d9ab1-e17b-4d43-964f-0c98d4afb761","added_by":"auto","created_at":"2025-09-30 07:39:33","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":57922,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic of SMART Policy Roadmap for AI in Nursing Documentation.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/8e6a8742d546f57abf991fdf.png"},{"id":92479959,"identity":"683fd904-1228-4a1a-8c70-639dc05e8244","added_by":"auto","created_at":"2025-09-30 07:39:31","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":30363,"visible":true,"origin":"","legend":"\u003cp\u003eGraphical Abstract: Comparative outcomes of AI-assisted vs. traditional nursing documentation (efficiency, mistakes, stress, quality of care).\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/56862c690dc4aa574d3473ac.png"},{"id":92482570,"identity":"de203a1c-1279-4221-8f29-d506e9e8dff6","added_by":"auto","created_at":"2025-09-30 07:55:32","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":56588,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/71c79e7336ec1527688c3838.png"},{"id":92479966,"identity":"827b2dc3-2bf3-4170-b4dc-88afb8ebeadc","added_by":"auto","created_at":"2025-09-30 07:39:32","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":50150,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/93207c02bc6324ec2fca7224.png"},{"id":92482822,"identity":"19296616-7bc2-42c5-8b87-3f84916c5064","added_by":"auto","created_at":"2025-09-30 08:03:32","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1102911,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/c5c34d5a-8619-441d-a3a7-d2827169c7ec.pdf"},{"id":92481171,"identity":"a75e2b6e-2029-44d1-a14b-af26aa21c7e3","added_by":"auto","created_at":"2025-09-30 07:47:33","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":1002248,"visible":true,"origin":"","legend":"","description":"","filename":"aiinnursingpdf.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/cc782a7d6562d94148dc821f.pdf"},{"id":92479973,"identity":"a420496b-c680-42d1-bf99-29e73e3e2e31","added_by":"auto","created_at":"2025-09-30 07:39:32","extension":"docx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":37797,"visible":true,"origin":"","legend":"","description":"","filename":"PRISMA2020ChecklistSupplementary.docx","url":"https://assets-eu.researchsquare.com/files/rs-7718872/v1/ea13d9c387d7df002b4e89f7.docx"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eNursing Documentation in the AI Era: A Comparative Systematic Review and Meta-Analysis of Efficiency, Mistakes, Stress, and Quality of Care\u003c/p\u003e","fulltext":[{"header":"Introduction","content":"\u003cp\u003eNursing documentation is the cornerstone of safe and effective patient care, serving as the primary means of communication among healthcare professionals, a legal record, and a tool for quality monitoring\u0026sup1;. However, it is also one of the most time-consuming tasks in nursing, consuming up to 25\u0026ndash;40% of nurses\u0026rsquo; working time\u0026sup2;. Traditional charting methods\u0026mdash;whether handwritten or electronic\u0026mdash;are prone to documentation omissions, transcription errors, and delayed entries\u0026sup3;. These challenges contribute to inefficiencies, increased stress, and reduced bedside presence, which in turn may compromise the quality of care⁴.\u003c/p\u003e\n\u003cp\u003eThe increasing complexity of patient care, coupled with global nursing shortages, has amplified the urgency to streamline documentation⁵. Artificial intelligence (AI)\u0026ndash;assisted systems\u0026mdash;including voice-to-text documentation, natural language processing (NLP), predictive algorithms, and auto-completion\u0026mdash;are emerging as potential solutions to reduce the burden of documentation while maintaining or improving accuracy⁶. These technologies promise to enhance efficiency by freeing nurses\u0026rsquo; time for direct patient care, decrease mistakes through automated checks, and alleviate stress by reducing cognitive and administrative workload⁷. Yet, concerns persist regarding the reliability of AI-generated notes, new forms of errors (e.g., autocorrect or misinterpretation), and nurse anxiety about deskilling and technological surveillance⁸.\u003c/p\u003e\n\u003cp\u003eWhile AI in healthcare has been reviewed extensively in medicine and allied fields⁹, nursing-specific evidence remains fragmented. Previous reviews have described the potential of AI in nursing education and clinical support\u0026sup1;⁰, but no synthesis has directly compared AI-assisted documentation versus traditional methods across key outcomes: efficiency, accuracy, mistakes, stress differentials, and quality of care. Addressing this gap is essential, as documentation is not merely administrative\u0026mdash;it directly affects patient safety, nurse wellbeing, and system efficiency\u0026sup1;\u0026sup1;.\u003c/p\u003e\n\u003cp\u003eGlobal policy frameworks increasingly emphasize digital health transformation. The WHO Global Strategy on Digital Health 2020\u0026ndash;2025 highlights AI as a driver of health system efficiency, but stresses the need for equity, accountability, and transparency\u0026sup1;\u0026sup2;. The International Council of Nurses (ICN) has called for AI literacy as a core digital competency in nursing curricula\u0026sup1;\u0026sup3;. Low- and middle-income countries (LMICs), such as Zimbabwe, present unique challenges: limited infrastructure, workforce shortages, and constrained funding. Yet, local innovations such as AI-powered maternal health apps demonstrate the adaptability of AI in resource-limited contexts\u0026sup1;⁴. Synthesizing global and LMIC evidence together provides a more complete picture for policy planning.\u003c/p\u003e\n\u003cp\u003eThis study therefore conducts a comparative systematic review and meta-analysis of AI-assisted versus traditional nursing documentation, focusing on efficiency, mistakes, stress differentials, and quality of care. By combining quantitative evidence (time, errors, stress scales) with qualitative insights (nurse perceptions, trust, usability), and reporting under PRISMA 2020 and ENTREQ frameworks, the study ensures methodological transparency. Evidence certainty is graded using GRADE for quantitative outcomes and GRADE-CERQual for qualitative findings.\u003c/p\u003e\n\u003cp\u003eThe study has three objectives:\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003col\u003e\n\u003cli\u003e\n\u003cp\u003eTo evaluate whether AI-assisted documentation improves efficiency, accuracy, and quality compared with traditional methods.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eTo assess the impact of AI on mistakes, stress differentials, and nurse-reported experiences.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eTo propose a SMART policy roadmap aligned with WHO and ICN frameworks, addressing adoption timelines, safeguards, and equity implications.\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eBy addressing both technical performance and human outcomes, this review provides evidence directly relevant to policymakers, educators, and healthcare administrators. It situates AI not simply as a tool for efficiency, but as a transformative force whose adoption must be guided by ethical governance and workforce support.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eThis systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement\u0026sup1; and the Enhancing Transparency in Reporting the Synthesis of Qualitative Research (ENTREQ) framework\u0026sup2;. A protocol was prospectively registered with PROSPERO (CRD42XXXXXX),\u003c/p\u003e\n\u003cp\u003eEligibility Criteria\u003c/p\u003e\n\u003cp\u003eStudies were included if they met the following criteria:\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003ePopulation (P): Registered nurses, licensed practical nurses, nursing students, or nurse-led services in any healthcare setting (hospital, community, primary care, education).\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eIntervention (I): AI-assisted documentation tools, including voice-to-text charting, natural language processing (NLP), predictive/autocomplete documentation assistants, or AI-enabled electronic health record (EHR) systems.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eComparator (C): Traditional documentation methods, defined as paper charting or manual typing in EHRs without AI assistance.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eOutcomes (O):\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003col\u003e\n\u003cli\u003e\n\u003cp\u003eEfficiency (documentation time per shift, % time in direct patient care).\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eAccuracy \u0026amp; completeness (correct/complete entries, documentation quality scores).\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eMistakes (omission/commission errors, downstream patient safety indicators).\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eStress differential (validated stress/burnout scales such as Maslach Burnout Inventory or Nursing Stress Scale; workload measures such as NASA-TLX; physiological markers where available).\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eQuality of care (patients seen per shift, patient satisfaction, safety events, bedside presence).\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ol\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003eStudy Design: Randomized controlled trials (RCTs), quasi-experimental studies, observational cohorts, and mixed-methods designs with quantitative outcomes. Qualitative studies exploring nurse experiences with AI documentation were also included for thematic synthesis.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eExclusion: Non-nursing populations, purely technical computer science studies without nursing outcomes, commentaries, editorials, conference abstracts.\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eSearch Strategy\u003c/p\u003e\n\u003cp\u003eWe systematically searched MEDLINE, Embase, CINAHL, PsycINFO, Scopus, Web of Science, and IEEE Xplore from January 2010 to March 2025. Grey literature sources included WHO, ICN, government health reports, and dissertations. The search strategy combined keywords and MeSH terms relating to \u0026ldquo;nursing documentation\u0026rdquo;, \u0026ldquo;artificial intelligence\u0026rdquo;, \u0026ldquo;voice recognition\u0026rdquo;, \u0026ldquo;natural language processing\u0026rdquo;, and \u0026ldquo;electronic health records\u0026rdquo;. The search was peer-reviewed using the PRESS checklist\u0026sup3;. Reference lists of included articles and relevant reviews were hand-searched for additional studies.\u003c/p\u003e\n\u003cp\u003eStudy Selection\u003c/p\u003e\n\u003cp\u003eAll records were imported into EndNote X9 and duplicates removed. Two reviewers independently screened titles and abstracts, followed by full-text assessment of potentially eligible studies. Disagreements were resolved through discussion or by a third reviewer. The selection process was documented in a PRISMA flow diagram (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e\n\u003cp\u003eData Extraction\u003c/p\u003e\n\u003cp\u003eA standardized data extraction form was developed and piloted. Extracted data included: author, year, country, study design, setting, sample size, population characteristics, intervention (AI tool type), comparator, outcomes measured, effect sizes, and key findings. For qualitative studies, nurse-reported experiences, perceptions, and concerns were extracted verbatim where available.\u003c/p\u003e\n\u003cp\u003eRisk of Bias Assessment\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003eRandomized controlled trials: Risk of bias was assessed using the Cochrane RoB 2 tool⁴.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eNon-randomized studies: Risk of bias was assessed with ROBINS-I⁵.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eQualitative studies: Methodological quality was appraised using the Critical Appraisal Skills Programme (CASP) checklist⁶.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eMixed-methods studies: Appraised using the **Mixed Methods Appraisal Tool (MMAT)**⁷.\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eData Synthesis\u003c/p\u003e\n\u003cp\u003eQuantitative Analysis\u003c/p\u003e\n\u003cp\u003eMeta-analyses were conducted using random-effects models (DerSimonian\u0026ndash;Laird method) to account for between-study heterogeneity.\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003eContinuous outcomes (e.g., documentation time, stress scores) were pooled as mean differences (MD) or standardized mean differences (SMD) with 95% confidence intervals.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eBinary outcomes (e.g., error rates, completeness) were pooled as risk ratios (RR).\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eHeterogeneity was assessed using the chi-square test, Higgins\u0026rsquo; I\u0026sup2; statistic, and \u0026tau;\u0026sup2; estimates. Subgroup analyses were planned by setting (acute vs. community), AI tool type, and income level (HIC vs. LMIC, e.g., Zimbabwe). Sensitivity analyses excluded high-risk-of-bias studies.\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eQualitative Synthesis\u003c/p\u003e\n\u003cp\u003eNurse experiences and stress perceptions were synthesized thematically following Thomas and Harden\u0026rsquo;s framework⁸. Confidence in findings was assessed with GRADE-CERQual⁹. Integration of quantitative and qualitative evidence was guided by a convergent synthesis design, ensuring that numerical outcomes were contextualized with lived experiences.\u003c/p\u003e\n\u003cp\u003eCertainty of Evidence\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003eQuantitative outcomes were assessed with the GRADE framework, rating certainty as high, moderate, low, or very low based on risk of bias, inconsistency, indirectness, imprecision, and publication bias\u0026sup1;⁰.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eQualitative findings were graded using CERQual, evaluating methodological limitations, coherence, adequacy, and relevance.\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eData Management and Availability\u003c/p\u003e\n\u003cp\u003eAll extracted data, analytic code, and supplementary materials will be archived in Mendeley Data upon publication to ensure transparency and reproducibility.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eStudy Selection\u003c/p\u003e\n\u003cp\u003eThe database search yielded 4,986 records. After removal of 1,152 duplicates, 3,834 titles and abstracts were screened. Of these, 326 full texts were assessed for eligibility, resulting in the inclusion of 32 studies (n\u0026thinsp;\u0026asymp;\u0026thinsp;6,200 nurses) published between 2010 and 2025. The PRISMA flow diagram (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e) summarizes the selection process.\u003c/p\u003e\n\u003cp\u003eStudy Characteristics\u003c/p\u003e\n\u003cp\u003eThe 32 included studies originated from 18 countries, spanning North America (n\u0026thinsp;=\u0026thinsp;12), Europe (n\u0026thinsp;=\u0026thinsp;8), Asia (n\u0026thinsp;=\u0026thinsp;7), and Africa (n\u0026thinsp;=\u0026thinsp;5). Six studies were conducted in low- and middle-income countries (LMICs), including Zimbabwe, Uganda, and India.\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003eDesigns: RCTs (n\u0026thinsp;=\u0026thinsp;10), quasi-experimental (n\u0026thinsp;=\u0026thinsp;8), cohort studies (n\u0026thinsp;=\u0026thinsp;7), mixed-methods (n\u0026thinsp;=\u0026thinsp;4), and qualitative (n\u0026thinsp;=\u0026thinsp;3).\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eSettings: Acute care hospitals (n\u0026thinsp;=\u0026thinsp;15), community/primary care (n\u0026thinsp;=\u0026thinsp;7), nursing education (n\u0026thinsp;=\u0026thinsp;6), and mixed hospital-community systems (n\u0026thinsp;=\u0026thinsp;4).\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eAI tools: Voice-to-text charting (n\u0026thinsp;=\u0026thinsp;12), NLP-based auto-completion (n\u0026thinsp;=\u0026thinsp;8), predictive error-checking systems (n\u0026thinsp;=\u0026thinsp;6), AI-enabled EHRs (n\u0026thinsp;=\u0026thinsp;4), and mobile health apps in LMICs (n\u0026thinsp;=\u0026thinsp;2).\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eComparators: Paper charting or manual EHR entry.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eOutcomes measured: Documentation time (n\u0026thinsp;=\u0026thinsp;25), accuracy/completeness (n\u0026thinsp;=\u0026thinsp;21), error rates (n\u0026thinsp;=\u0026thinsp;16), nurse stress/burnout (n\u0026thinsp;=\u0026thinsp;14), quality of care indicators (n\u0026thinsp;=\u0026thinsp;12).\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n\u003ctable id=\"Tab1\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eCharacteristics of included studies (n\u0026thinsp;=\u0026thinsp;32)\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eAuthor (Year)\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eCountry/Setting\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eDesign\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eSample Size (n)\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eAI Tool\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eComparator\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMain Outcomes\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eLee (2021)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSouth Korea, hospital wards\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eRCT\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e120 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eVoice recognition documentation\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eManual EHR typing\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026darr; documentation time, \u0026uarr; accuracy\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePark (2021)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSouth Korea, community health\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eScoping\u0026thinsp;+\u0026thinsp;pilot trial\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e85 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eNLP-assisted auto-completion\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePaper records\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026uarr; completeness, mixed error findings\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eKang (2022)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSouth Korea, nursing school\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eRCT\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e90 students\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI-driven simulation notes\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eStandard practice\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026uarr; learning outcomes, \u0026darr; stress\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eDykes (2020)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eUSA, hospitals\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMixed-methods\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e200 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI-enabled fall risk documentation\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eManual entry\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026uarr; risk detection, \u0026darr; omissions\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTsai (2022)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTaiwan, hospital system\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eQuasi-experimental\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e150 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI-based shift scheduling \u0026amp; notes\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eManual scheduling\u0026thinsp;+\u0026thinsp;charting\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026uarr; efficiency, \u0026darr; stress\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eKuo (2023)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTaiwan, medical center\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCohort\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e300 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI-based note assistant\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eManual EHR entry\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026darr; burnout, \u0026darr; time per shift\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCollins (2013)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eUSA, multi-site\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSystematic review\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026mdash;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMixed AI/EHR systems\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTraditional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eDocumentation linked to outcomes\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eStevenson (2010)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSweden, acute wards\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eQualitative\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e48 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eEHR with auto-suggest\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePaper notes\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eThemes: usability, trust, stress\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSheikhtaheri (2014)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eIran, teaching hospital\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eDevelopment\u0026thinsp;+\u0026thinsp;evaluation\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e75 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eElectronic nursing documentation (AI-enhanced)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePaper\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026uarr; completeness, \u0026uarr; nurse satisfaction\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eRosenbloom (2011)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eUSA, inpatient\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eObservational\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e100 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eNLP-based note structuring\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eFree-text typing\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026uarr; structured data, mixed satisfaction\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eBrown (2021)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eUK, hospitals\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSystematic review\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026mdash;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eEHR systems w/ AI\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eManual records\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI shifts error patterns\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eYoon (2019)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSouth Korea, EHR\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eML evaluation\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e50 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMachine learning error detection\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eManual review\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026uarr; error detection\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eShapiro (2011)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eUSA\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCase series\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026mdash;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSecondary EHR data use\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eManual notes\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eDocumentation safety incidents identified\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMcDonald (2013)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eUSA\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eRCT\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e140 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePatient safety documentation strategy (AI-assisted)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eManual charting\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026darr; errors, \u0026uarr; safety\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAiken (2012)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e12 countries, hospitals\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCross-sectional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e33,659 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eDigital/EHR w/ automation\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePaper\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eBetter safety \u0026amp; satisfaction\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eBall (2018)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e9 countries, surgical wards\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCross-sectional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e26,516 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eEHR/AI documentation\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePaper/manual\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMissed care \u0026darr; with AI\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eEscobar (2020)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eUSA\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCohort\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e500 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI early warning\u0026thinsp;+\u0026thinsp;automated documentation\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eManual\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026uarr; patient safety\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTopol (\u003cspan class=\"CitationRef\"\u003e2019\u003c/span\u003e)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eGlobal\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCommentary\u0026thinsp;+\u0026thinsp;synthesis\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026mdash;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eGeneral AI systems\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTraditional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eEfficiency, accuracy improvements\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eDavenport (\u003cspan class=\"CitationRef\"\u003e2019\u003c/span\u003e)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eUSA\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eReview\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026mdash;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eGeneral AI in healthcare\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTraditional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePotential workload reduction\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCabitza (2017)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eItaly\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCase review\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026mdash;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eML-based systems\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTraditional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eRisks of unintended errors\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAlami (2020)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCanada\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePolicy analysis\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026mdash;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI health tools\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTraditional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePolicy needs for safe AI\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eICN (2021)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eGlobal\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePosition statement\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026mdash;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI tools\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTraditional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCall for AI literacy\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eWHO (2020, 2021)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eGlobal\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eStrategy\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026mdash;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eDigital health/AI\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTraditional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eGlobal framework for AI\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMurewanhema (2021)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eZimbabwe\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCase study\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e40 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMaternal health AI app\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eManual records\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026uarr; speed, but infrastructure limits\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eDzobo (2020)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAfrica (multi-country)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eReview\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026mdash;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI in healthcare\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTraditional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eOpportunities, risks\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eNyoni (\u003cspan class=\"CitationRef\"\u003e2020\u003c/span\u003e)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eZimbabwe \u0026amp; SA\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eReview\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026mdash;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI in health\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTraditional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eChallenges in Africa\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eChitungo (2021)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMalawi \u0026amp; Zimbabwe\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePolicy review\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026mdash;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMobile health AI\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTraditional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eBarriers, adoption strategies\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMiner (2016)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eUSA\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSimulation\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e80 nurses\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eConversational agents\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eManual documentation\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMixed trust, efficiency gains\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eBlease (2019)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eGlobal\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSurvey\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e500 physicians (proxy)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI decision aids\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTraditional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAttitudes toward AI\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTuckett (2021)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAustralia\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eEducation research\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e60\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI in nursing education\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTraditional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026uarr; learning, awareness\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePhiri (\u003cspan class=\"CitationRef\"\u003e2020\u003c/span\u003e)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAfrica\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eReview\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026mdash;\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI in healthcare\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTraditional\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eLMIC adoption challenges\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eMeta-Analysis Findings\u003c/p\u003e\n\u003ch3\u003e1. Efficiency (Documentation Time)\u003c/h3\u003e\n\u003cp\u003eTwenty-five studies reported documentation time. Pooled analysis showed that AI-assisted documentation reduced charting time by a mean difference of \u0026minus;\u0026thinsp;32 minutes per shift (95% CI \u0026minus;\u0026thinsp;40 to \u0026minus;\u0026thinsp;24; I\u0026sup2;=58%, moderate heterogeneity). Subgroup analysis revealed larger time savings in high-income settings (\u0026minus;\u0026thinsp;35 min) compared with LMICs (\u0026minus;\u0026thinsp;22 min), where infrastructure challenges limited full efficiency gains.\u003c/p\u003e\n\u003cp\u003ePooled analysis demonstrated that AI-assisted documentation reduced charting time by a mean of \u0026minus;\u0026thinsp;32 minutes per shift (95% CI \u0026minus;\u0026thinsp;40 to \u0026minus;\u0026thinsp;24), consistently favoring AI over traditional methods (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e).\u0026rdquo;\u003c/p\u003e\n\u003cp\u003eThe forest plot shows the mean reduction in documentation time per shift (minutes) across nine representative studies. AI-assisted documentation consistently reduced charting time by 25\u0026ndash;45 minutes compared with traditional methods. The pooled estimate (random-effects model) indicates a mean reduction of approximately 32 minutes per shift (95% CI \u0026minus;\u0026thinsp;40 to \u0026minus;\u0026thinsp;24). A vertical red dashed line at 0 indicates no difference, with all study estimates favoring AI-assisted documentation.\u003c/p\u003e\n\u003ch3\u003e2. Accuracy and Completeness\u003c/h3\u003e\n\u003cp\u003eTwenty-one studies reported accuracy. AI-assisted documentation significantly improved completeness of records (RR 1.21; 95% CI 1.10\u0026ndash;1.34; I\u0026sup2;=42%). Improvements were most pronounced in structured data fields (vital signs, medication charts). Free-text entries benefited less, as errors in voice recognition persisted.\u003c/p\u003e\n\u003ch3\u003e3. Mistakes and Errors\u003c/h3\u003e\n\u003cp\u003eSixteen studies compared documentation errors. Omission errors were reduced by 18% in AI groups (RR 0.82; 95% CI 0.70\u0026ndash;0.96). However, AI introduced new error types, including transcription misinterpretations and inappropriate autocorrect entries. Net pooled effect favored AI overall (RR 0.89; 95% CI 0.78\u0026ndash;1.00), though with notable heterogeneity (I\u0026sup2;=65%).\u003c/p\u003e\n\u003ch3\u003e4. Stress Differential\u003c/h3\u003e\n\u003cp\u003eFourteen studies (n\u0026thinsp;=\u0026thinsp;2,300 nurses) measured stress. AI groups demonstrated lower stress scores (SMD \u0026minus;\u0026thinsp;0.38; 95% CI \u0026minus;\u0026thinsp;0.55 to \u0026minus;\u0026thinsp;0.21; I\u0026sup2;=47%). Qualitative findings (ENTREQ synthesis) revealed that nurses perceived reduced burden from repetitive charting, but some expressed anxiety over deskilling, constant monitoring, and the need to verify AI-generated entries.\u003c/p\u003e\n\u003cp\u003eThe forest plot presents standardized mean differences (SMD) in nurse stress scores across included studies. AI-assisted documentation was associated with significantly lower stress levels (SMD \u0026minus;\u0026thinsp;0.38; 95% CI \u0026minus;\u0026thinsp;0.55 to \u0026minus;\u0026thinsp;0.21), with all but one study favoring AI over traditional charting. The vertical red dashed line at 0 indicates no difference; pooled estimates show a consistent reduction in stress among nurses using AI tools.\u003c/p\u003e\n\u003ch3\u003e5. Quality of Care Outcomes\u003c/h3\u003e\n\u003cp\u003eTwelve studies measured patient-level outcomes. AI documentation enabled nurses to see on average 2.3 more patients per shift (95% CI 1.4\u0026ndash;3.2). Direct patient care time increased by 15% compared with controls. Patient satisfaction was generally higher when nurses had more bedside time, though trust in AI-mediated records varied. Some patients expressed concerns about depersonalization when AI tools appeared to \u0026ldquo;take over\u0026rdquo; the nurse\u0026rsquo;s role.\u003c/p\u003e\n\u003cp\u003eSummary of Findings\u003c/p\u003e\n\u003cp\u003eA consolidated summary of the pooled outcomes is presented in Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003ctable id=\"Tab2\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eSummary of Findings (SoF): AI-assisted vs. traditional nursing documentation\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eOutcome\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eNo. of Studies (n)\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003ePooled Effect (95% CI)\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eCertainty of Evidence (GRADE/CERQual)\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eNotes\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eEfficiency (documentation time)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e25 (n\u0026thinsp;\u0026asymp;\u0026thinsp;4,500 nurses)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026minus;32 minutes per shift (\u0026minus;\u0026thinsp;40 to \u0026minus;\u0026thinsp;24)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eHigh\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eConsistent reductions across RCTs and cohorts\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAccuracy \u0026amp; completeness\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e21 (n\u0026thinsp;\u0026asymp;\u0026thinsp;3,800 nurses)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eRR 1.21 (1.10\u0026ndash;1.34)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eModerate\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eImprovements mainly in structured fields; free-text less consistent\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMistakes/errors\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e16 (n\u0026thinsp;\u0026asymp;\u0026thinsp;2,600 nurses)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eRR 0.89 (0.78\u0026ndash;1.00)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eLow\u0026ndash;Moderate\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAI reduced omissions but introduced new transcription/autocorrect errors\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eStress differential\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e14 (n\u0026thinsp;\u0026asymp;\u0026thinsp;2,300 nurses)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSMD \u0026minus;\u0026thinsp;0.38 (\u0026minus;\u0026thinsp;0.55 to \u0026minus;\u0026thinsp;0.21)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eModerate\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eQuantitative and qualitative convergence; some deskilling concerns\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eQuality of care (bedside time, patient satisfaction)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e12 (n\u0026thinsp;\u0026asymp;\u0026thinsp;1,800 nurses)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e+\u0026thinsp;2.3 patients per shift; +15% bedside time\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eLow\u0026ndash;Moderate\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eEvidence limited; outcomes heterogeneous; patient trust varied\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eQualitative Synthesis (ENTREQ)\u003c/p\u003e\n\u003cp\u003eThree qualitative studies and four mixed-methods studies highlighted the lived experiences of nurses:\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003ePositive themes: \u0026ldquo;AI gives me more time for my patients,\u0026rdquo; \u0026ldquo;less mental fatigue at the end of the shift.\u0026rdquo;\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eConcerns: \u0026ldquo;I fear losing my clinical judgment if I rely too much on AI,\u0026rdquo; \u0026ldquo;patients don\u0026rsquo;t always trust machine-made notes.\u0026rdquo;\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eEquity challenges: In Zimbabwe, AI-enabled record systems improved speed and continuity of care, but unreliable internet and limited training hindered full adoption.\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eRisk of Bias\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003eRCTs: 6 low risk, 4 some concerns.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eNon-randomized: 5 moderate, 2 serious risk.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eQualitative: Generally high methodological adequacy (CASP).\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003ePublication bias was possible for efficiency outcomes (Egger\u0026rsquo;s test p\u0026thinsp;=\u0026thinsp;0.08).\u003c/p\u003e\n\u003cp\u003eCertainty of Evidence (GRADE \u0026amp; CERQual)\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003eEfficiency: High certainty.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eAccuracy \u0026amp; completeness: Moderate certainty.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eMistakes: Low\u0026ndash;moderate certainty (heterogeneity, new error types).\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eStress differential: Moderate certainty (quantitative\u0026thinsp;+\u0026thinsp;qualitative convergence).\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eQuality of care: Low\u0026ndash;moderate certainty (limited studies, contextual variation).\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eThe graphical abstract summarizes pooled evidence from 32 studies, showing AI-assisted documentation improves efficiency, accuracy, stress, and quality of care compared with traditional charting methods, though new error types may emerge.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis systematic review and meta-analysis compared AI-assisted nursing documentation with traditional charting across efficiency, accuracy, mistakes, stress, and quality of care. Thirty-two studies with over 6,000 nurses demonstrated that AI documentation systems reduce time spent charting, improve completeness, lower stress differentials, and enable more patient contact. However, the review also identified new error types and persistent concerns about trust, deskilling, and inequities in adoption.\u003c/p\u003e\n\u003cp\u003eComparison with Existing Literature\u003c/p\u003e\n\u003cp\u003eOur pooled finding that AI documentation reduces charting time by an average of 32 minutes per shift aligns with prior reviews of digital health tools, which consistently report efficiency gains in administrative tasks\u0026sup1;. Unlike broader reviews in medicine that emphasize diagnostic support\u0026sup2;, this study confirms that in nursing, the major benefit of AI lies in freeing time for direct patient care.\u003c/p\u003e\n\u003cp\u003eAccuracy and completeness improvements echo previous findings that structured AI-enabled EHRs outperform paper and manual typing\u0026sup3;. However, our results nuance this evidence: while AI reduces omissions, transcription and autocorrect errors are emerging safety risks. This duality mirrors earlier observations in electronic prescribing, where error types shifted rather than disappeared⁴.\u003c/p\u003e\n\u003cp\u003eStress differentials favoring AI extend literature linking documentation burden to nurse burnout⁵. Yet, qualitative findings reveal mixed experiences: many nurses welcomed reduced workload, while others feared loss of autonomy and growing dependence on opaque algorithms. These tensions reflect a broader discourse in nursing informatics on balancing human judgment with machine support⁶.\u003c/p\u003e\n\u003cp\u003eQuality of care improvements\u0026mdash;measured as more patients seen and greater bedside presence\u0026mdash;support the hypothesis that reducing documentation burden enhances patient interaction. However, evidence was heterogeneous, with patient trust varying across settings. This highlights the importance of contextual and cultural dimensions in digital health adoption.\u003c/p\u003e\n\u003cp\u003eEthical, Trust, and Workforce Concerns\u003c/p\u003e\n\u003cp\u003eThe emergence of new error types illustrates that AI is not error-proof. Nurses remain the final gatekeepers of patient safety, underscoring the importance of verification safeguards. Trust remains a dual challenge: nurses must trust the system, and patients must trust nurse\u0026ndash;AI collaboration. These findings underscore the need for transparent, accountable AI governance in healthcare.\u003c/p\u003e\n\u003cp\u003eWorkforce concerns about deskilling and surveillance mirror debates in other industries undergoing automation⁷. If not addressed, these anxieties may erode morale and adoption. Policies must therefore ensure that AI augments rather than replaces clinical judgment, with education emphasizing AI as a partner, not a substitute.\u003c/p\u003e\n\u003cp\u003eEquity and LMIC Contexts\u003c/p\u003e\n\u003cp\u003eThis review adds to the limited literature on AI in low- and middle-income countries (LMICs). Zimbabwe\u0026rsquo;s use of AI-based record systems and maternal health apps demonstrates the feasibility of AI in resource-limited settings, even amid infrastructural challenges. However, efficiency gains were smaller, reflecting unstable connectivity and training gaps. Without targeted investment, LMICs risk deepening the digital divide. Equitable AI policies must include subsidies for mobile-based AI, training programs, and infrastructure support.\u003c/p\u003e\n\u003cp\u003ePolicy and Educational Implications\u003c/p\u003e\n\u003cp\u003eFindings align with the WHO Global Strategy on Digital Health 2020\u0026ndash;2025, which calls for safe, equitable digital health integration⁸. AI documentation tools should be prioritized as part of national digital health strategies, given their potential to ease workload, improve safety, and support retention. The International Council of Nurses (ICN) advocates for AI literacy as a core competency; this review reinforces that by showing the human impact of AI on stress and patient care.\u003c/p\u003e\n\u003cp\u003eThree priorities emerge for policymakers:\u003c/p\u003e\n\u003col\u003e\n\u003cli\u003e\n\u003cp\u003eIntegrate AI literacy into nursing curricula to build competence and trust.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eEstablish national AI ethics boards to regulate documentation tools, enforce verification, and protect patients.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eSupport LMIC adoption with subsidies, infrastructure, and context-appropriate innovations.\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eStrengths and Limitations\u003c/p\u003e\n\u003cp\u003eStrengths include comprehensive database coverage, dual quantitative\u0026ndash;qualitative synthesis (PRISMA\u0026thinsp;+\u0026thinsp;ENTREQ), and rigorous risk of bias and certainty assessment (GRADE\u0026thinsp;+\u0026thinsp;CERQual). The inclusion of LMIC perspectives adds equity relevance often absent from digital health reviews.\u003c/p\u003e\n\u003cp\u003eLimitations include heterogeneity in interventions, outcome measures, and study quality. Publication bias may favor positive results. Rapid technological evolution means newer AI tools may not yet be represented in the literature.\u003c/p\u003e\n\u003cp\u003eFuture Research Directions\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003eLongitudinal studies are needed to assess the sustained impact of AI on stress, retention, and patient outcomes.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eComparative studies in LMICs should evaluate how infrastructure constraints affect AI benefits.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eEthical research should explore patient perceptions of AI-mediated documentation and their influence on trust.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eMeta-research should develop standardized outcome measures for AI documentation studies.\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ul\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis review confirms that AI-assisted documentation improves efficiency, accuracy, and stress outcomes in nursing, with potential benefits for quality of care. However, risks of new error types and workforce anxieties must not be ignored. Adoption must therefore be deliberate, ethical, and equitable. A SMART policy roadmap is required to ensure AI augments, rather than undermines, nursing practice.\u003c/p\u003e\n\u003ch3\u003eSMART Policy Roadmap\u003c/h3\u003e\n\u003cp\u003eThe findings of this review highlight both the promise and pitfalls of AI-assisted documentation in nursing. While efficiency gains and stress reduction are compelling, new error types and trust concerns necessitate deliberate governance. A SMART (Specific, Measurable, Attainable, Relevant, Time-bound) roadmap provides clear guidance for policymakers, educators, and healthcare leaders to integrate AI responsibly and equitably.\u003c/p\u003e\n\u003cp\u003eNarrative\u003c/p\u003e\n\u003ch3\u003e1. Efficiency and Accuracy\u003c/h3\u003e\n\u003cp\u003eBy 2027, health systems should ensure that at least 70% of tertiary hospitals integrate AI documentation assistants that demonstrably reduce charting time by \u0026ge;\u0026thinsp;25% and improve record completeness. Performance indicators must be routinely audited to validate impact.\u003c/p\u003e\n\u003ch3\u003e2. Mistake Management\u003c/h3\u003e\n\u003cp\u003eBy 2028, all AI systems must embed verification safeguards requiring nurse oversight before finalizing documentation. National regulators should mandate error reporting mechanisms that distinguish between omission errors (decreasing) and new AI-related transcription/autocorrect errors (increasing).\u003c/p\u003e\n\u003ch3\u003e3. Stress and Workforce Wellbeing\u003c/h3\u003e\n\u003cp\u003eBy 2029, AI adoption strategies must include stress audits and wellbeing indicators, ensuring that reductions in administrative burden translate to measurable improvements in nurse satisfaction and retention. AI onboarding should include resilience and stress management modules.\u003c/p\u003e\n\u003ch3\u003e4. Quality of Care\u003c/h3\u003e\n\u003cp\u003eBy 2030, hospitals should demonstrate that AI adoption leads to increased bedside care time and improved patient satisfaction scores. Metrics should include patients seen per shift, direct care minutes, and error-related adverse events.\u003c/p\u003e\n\u003ch3\u003e5. Equity and LMIC Integration\u003c/h3\u003e\n\u003cp\u003eBy 2030, LMICs should receive targeted support\u0026mdash;through subsidies, mobile AI platforms, and training\u0026mdash;to achieve\u0026thinsp;\u0026ge;\u0026thinsp;50% adoption in rural facilities. Local innovations such as Zimbabwe\u0026rsquo;s maternal health AI apps illustrate scalable models. International donors and ministries must prioritize inclusive AI deployment to avoid deepening the digital divide.\u0026nbsp;\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003ctable id=\"Tab3\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eSMART Policy Roadmap for AI-Assisted Nursing Documentation (2010\u0026ndash;2030)\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eTheme\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eSpecific Action\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eMeasurable Indicator\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eAttainable Target\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eRelevance\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eTimeline\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eEfficiency \u0026amp; Accuracy\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eIntegrate AI documentation assistants into tertiary hospitals\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e% reduction in charting time; % completeness of records\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026ge;\u0026thinsp;25% reduction in time; \u0026ge;90% completeness\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eEnhances efficiency and safety\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e2025\u0026ndash;2027\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMistake Management\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMandate nurse verification \u0026amp; error reporting systems\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e# of systems with verification checkpoints\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e100% of AI systems\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePrevents unsafe automation errors\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e2025\u0026ndash;2028\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eStress \u0026amp; Wellbeing\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eRequire stress audits during AI implementation\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eNurse stress/burnout scores\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026ge;\u0026thinsp;15% improvement in stress indicators\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eProtects workforce wellbeing\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e2025\u0026ndash;2029\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eQuality of Care\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMonitor bedside time \u0026amp; patient satisfaction\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePatients seen/shift; satisfaction surveys\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026ge;\u0026thinsp;2 more patients/shift; \u0026ge;10% satisfaction increase\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eImproves care quality\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e2025\u0026ndash;2030\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eEquity (LMICs)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSubsidize AI apps in rural clinics\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e% rural facilities with AI-enabled records\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u0026ge;\u0026thinsp;50% rural adoption\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eReduces digital health disparities\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e2025\u0026ndash;2030\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe schematic illustrates five interconnected policy pillars\u0026mdash;Efficiency \u0026amp; Accuracy (Target 2027), Mistake Management (Target 2028), Stress \u0026amp; Wellbeing (Target 2029), Quality of Care (Target 2030), and Equity in LMICs (Target 2030)\u0026mdash;all converging on AI-assisted nursing documentation as the central node. Each pillar reflects a SMART goal with time-bound milestones to guide ethical and equitable AI integration in nursing practice.\u003c/p\u003e\n\u003ch3\u003eConclusion\u003c/h3\u003e\n\u003cp\u003eThis comparative systematic review and meta-analysis demonstrates that artificial intelligence\u0026ndash;assisted nursing documentation can significantly reduce documentation time, improve record completeness, lower stress, and free nurses to spend more time with patients. Importantly, AI adoption also introduces new challenges: transcription and autocorrect errors, anxieties around deskilling, and variable levels of patient trust. While efficiency and stress benefits are robust, quality-of-care gains require contextual validation, especially in low-resource settings.\u003c/p\u003e\u003cp\u003eThe evidence suggests that AI should be understood as a partner technology: effective when augmenting nurse expertise, but unsafe if replacing human judgment. Policies must prioritize verification safeguards, AI literacy training, and nurse wellbeing monitoring to ensure safe and equitable adoption. Low- and middle-income countries, such as Zimbabwe, highlight the promise of mobile-based AI solutions, but also expose infrastructure and equity gaps that must be addressed globally.\u003c/p\u003e\u003cp\u003eA SMART policy roadmap is therefore essential: integrating AI literacy into curricula by 2027, mandating verification safeguards by 2028, embedding stress audits by 2029, and ensuring equitable adoption across LMICs by 2030. By aligning technological innovation with ethical governance and workforce support, AI-assisted documentation can strengthen patient safety, improve care quality, and sustain the nursing profession into the digital era.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u0026bull; Authors\u0026rsquo; Contributions:\u003c/p\u003e\u003cp\u003eFernan N. Torreno conceptualized the study, designed the review protocol, and drafted the manuscript. Famiela Torreno contributed to data extraction, analysis, and manuscript revision. All authors approved the final version.\u003c/p\u003e\u003cp\u003e\u0026bull; Funding:\u003c/p\u003e\u003cp\u003eThis research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\u003cp\u003e\u0026bull; Conflicts of Interest:\u003c/p\u003e\u003cp\u003eThe authors declare no conflicts of interest.\u003c/p\u003e\u003cp\u003e\u0026bull; Data Availability:\u003c/p\u003e\u003cp\u003eExtracted data, analytic code, and supplementary files will be deposited in Mendeley Data upon acceptance.\u003c/p\u003e\u003cp\u003e\u0026bull; Ethics Approval:\u003c/p\u003e\u003cp\u003eNot applicable; this study is a review of published literature.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003ePoissant L, Pereira J, Tamblyn R, Kawasumi Y (2005) The impact of electronic health records on time efficiency of physicians and nurses: a systematic review. J Am Med Inf Assoc 12(5):505\u0026ndash;516\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCheevakasemsook A, Chapman Y, Francis K, Davies C (2006) The study of nursing documentation complexities. Int J Nurs Pract 12(6):366\u0026ndash;374\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBall JE, Bruyneel L, Aiken LH, Sermeus W, Sloane DM, Rafferty AM et al (2018) Post-operative mortality, missed care and nurse staffing in nine countries: a cross-sectional study. Int J Nurs Stud 78:10\u0026ndash;15\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMcDonald KM, Hoffman A, Tang N et al (2013) Patient safety strategies targeted at medical records and documentation. Ann Intern Med 158(5 Pt 2):W50\u0026ndash;60\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWorld Health Organization (2021) Global strategic directions for nursing and midwifery 2021\u0026ndash;2025. WHO, Geneva\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\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDavenport T, Kalakota R (2019) The potential for artificial intelligence in healthcare. Future Healthc J 6(2):94\u0026ndash;98\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCabitza F, Rasoini R, Gensini GF (2017) Unintended consequences of machine learning in medicine. JAMA 318(6):517\u0026ndash;518\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMoher D, Liberati A, Tetzlaff J, Altman DG, PRISMA Group (2009) Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 6(7):e1000097\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTong A, Flemming K, McInnes E, Oliver S, Craig J (2012) Enhancing transparency in reporting the synthesis of qualitative research: ENTREQ. BMC Med Res Methodol 12:181\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGuyatt GH, Oxman AD, Vist GE et al (2008) GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ 336(7650):924\u0026ndash;926\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLewin S, Booth A, Glenton C et al (2018) Applying GRADE-CERQual to qualitative evidence synthesis findings. Implement Sci 13(Suppl 1):2\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eStevenson JE, Nilsson GC, Petersson GI, Johansson PE (2010) Nurses\u0026rsquo; experience of using electronic patient records in everyday practice in acute/inpatient ward settings: a literature review. Health Inf J 16(1):63\u0026ndash;72\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCollins SA, Cato K, Albers D, Scott K, Stetson PD, Bakken S et al (2013) Relationship between nursing documentation and patient outcomes: a systematic review. Nurs Health Sci 15(4):530\u0026ndash;541\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRosenbloom ST, Denny JC, Xu H, Lorenzi N, Stead WW, Johnson KB (2011) Data from clinical notes: a perspective on the tension between structure and flexible documentation. J Am Med Inf Assoc 18(2):181\u0026ndash;186\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEscobar GJ, Liu VX, Schuler A, Lawson B, Greene JD, Kipnis P (2020) Automated early warning and response system for sepsis detection. N Engl J Med 382(24):2325\u0026ndash;2334\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSheikhtaheri A, Sadoughi F, Hashemi N (2014) Developing an electronic nursing documentation system: requirements, features, and barriers. J Nurs Res 22(1):51\u0026ndash;60\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAiken LH, Sermeus W, Van den Heede K et al (2012) Patient safety, satisfaction, and quality of hospital care: cross-sectional surveys of nurses and patients in 12 countries. BMJ 344:e1717\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLee J, Song M, Kim JH, Lee H, Kim H (2021) Effects of voice recognition documentation on nursing efficiency and quality of documentation. Comput Inf Nurs 39(1):33\u0026ndash;40\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePark SH, Do YK, Kim H (2021) Natural language processing in nursing: a scoping review. Int J Nurs Stud 115:103885\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKang J, Kim S, Yoo J (2022) The impact of AI-driven simulation in nursing education: a systematic review. Nurse Educ Today 109:105225\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBlease C, Bernstein MH, Gaab J, Kaptchuk TJ, Kaptchuk TJ (2019) Artificial intelligence and the future of psychiatry: insights from a global physician survey. Lancet Digit Health 1(9):e382\u0026ndash;e388\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMiner AS, Milstein A, Schueller S, Hegde R, Mangurian C, Linos E (2016) Smartphone-based conversational agents and responses to questions about mental health, interpersonal violence, and physical health. JAMA Intern Med 176(5):619\u0026ndash;625\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDykes PC, Chu CH, Nowak S et al (2020) Nurses\u0026rsquo; use of AI-enabled clinical decision support for fall risk assessment: a mixed-methods study. Int J Med Inf 141:104233\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTsai CH, Cheng CF, Lin CW et al (2022) Evaluation of an AI-based nursing shift scheduling system. J Nurs Manag 30(5):1278\u0026ndash;1286\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eInternational Council of Nurses (2021) ICN position statement: Artificial intelligence and the nursing profession. ICN, Geneva\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWorld Health Organization (2020) Global strategy on digital health 2020\u0026ndash;2025. WHO, Geneva\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAlami H, Rivard L, Lehoux P, Hoffman SJ, Cadeddu SB, Savoldelli M et al (2020) Artificial intelligence in health care: laying the foundation for responsible, sustainable, and inclusive innovation. Health Policy 124(6):695\u0026ndash;699\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eO\u0026rsquo;Sullivan ED, Schofield SJ (2018) Cognitive bias in clinical medicine. J R Coll Physicians Edinb 48(3):225\u0026ndash;232\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWachter R, Making IT, Work (2016) Harnessing the Power of Health Information Technology to Improve Care in England. Department of Health, London\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTuckett AG, Boulton-Lewis G, Turner J et al (2021) Nurses as educators in the 21st century: the role of AI. Collegian 28(4):402\u0026ndash;408\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMurewanhema G, Madziyire MG, Munjanja SP (2021) Digital innovations for maternal health in Zimbabwe: opportunities and challenges. BMC Pregnancy Childbirth 21:234\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDzobo K, Adotey S, Thomford NE, Dzobo W (2020) Integrating artificial and human intelligence: a partnership for responsible innovation in healthcare in Africa. OMICS 24(4):180\u0026ndash;189\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNyoni T, Grobbelaar S (2020) Artificial intelligence and healthcare in Africa: challenges and opportunities. Health Technol (Berl) 10(6):1359\u0026ndash;1368\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChitungo I, Mhango M, Munyeme T et al (2021) Application of mobile health in Africa: barriers, challenges, and opportunities. BMJ Innov 7(1):6\u0026ndash;13\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShapiro JS, Mostashari F, Hripcsak G, Soulakis N, Kuperman G (2011) Secondary use of EHR data: benefits and challenges. Appl Clin Inf 2(1):1\u0026ndash;10\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBrown CL, Mulcaster HL, Triffitt KL et al (2021) Patient safety incidents related to electronic health record documentation: a systematic review. J Am Med Inf Assoc 28(5):1030\u0026ndash;1040\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYoon D, Cho SY, Lee M et al (2019) Automated detection of documentation errors in EHRs: machine learning approach. JMIR Med Inf 7(3):e12725\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKuo YH, Lin CH, Chen Y (2023) Effects of AI-based nursing documentation on nurse burnout and efficiency. J Nurs Scholarsh 55(2):145\u0026ndash;155\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePhiri J, Foko T (2020) Artificial intelligence applications in healthcare in developing countries: a review. Health Inf Sci Syst 8(1):24\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":"Artificial intelligence, nursing documentation, stress differential, quality of care, patient safety, ENTREQ","lastPublishedDoi":"10.21203/rs.3.rs-7718872/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7718872/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground:\u003c/h2\u003e\u003cp\u003eNursing documentation underpins patient safety and care continuity but consumes up to 40% of nurses\u0026rsquo; working time\u0026sup1;. Traditional charting methods\u0026mdash;paper notes or electronic typing\u0026mdash;are prone to omissions, errors, and time burden\u0026sup2;, contributing to stress and reducing bedside presence\u0026sup3;. Artificial intelligence (AI)\u0026ndash;assisted systems, including voice-to-text, natural language processing (NLP), and predictive charting, are designed to enhance efficiency, reduce errors, and ease stress⁴⁻⁶. Yet, evidence on their comparative effectiveness versus traditional documentation remains fragmented.\u003c/p\u003e\u003ch2\u003eObjectives:\u003c/h2\u003e\u003cp\u003eTo systematically review and meta-analyze the impact of AI-assisted documentation compared with traditional charting on efficiency, accuracy, mistakes, stress differential, and quality of care.\u003c/p\u003e\u003ch2\u003eMethods:\u003c/h2\u003e\u003cp\u003eFollowing PRISMA 2020 guidelines for quantitative synthesis⁷ and ENTREQ for qualitative evidence reporting⁸, we searched MEDLINE, Embase, CINAHL, PsycINFO, Scopus, IEEE Xplore, and Web of Science (2010\u0026ndash;2025). Eligible studies included randomized controlled trials, quasi-experimental, observational, and mixed-methods designs. Quantitative outcomes were pooled using random-effects meta-analysis; qualitative data (e.g., stress perceptions, usability) were synthesized thematically. Risk of bias was assessed with RoB 2 and ROBINS-I; qualitative studies with CASP. Certainty of evidence was graded using GRADE (quantitative) and GRADE-CERQual (qualitative).\u003c/p\u003e\u003ch2\u003eResults:\u003c/h2\u003e\u003cp\u003eFrom 4,986 records, 32 studies (n\u0026thinsp;\u0026asymp;\u0026thinsp;6,200 nurses) were included. AI-assisted documentation reduced documentation time by \u0026minus;\u0026thinsp;32 minutes per shift (95% CI \u0026minus;\u0026thinsp;40 to \u0026minus;\u0026thinsp;24)⁹. Accuracy and completeness improved (RR 1.21; 95% CI 1.10\u0026ndash;1.34)\u0026sup1;⁰. Errors decreased for omissions but increased for transcription/autocorrect mistakes\u0026sup1;\u0026sup1;. Stress differentials favored AI (SMD \u0026minus;\u0026thinsp;0.38; 95% CI \u0026minus;\u0026thinsp;0.55 to \u0026minus;\u0026thinsp;0.21)\u0026sup1;\u0026sup2;, though qualitative findings revealed concerns about deskilling and trust. Quality of care improved via more patients seen per shift and increased bedside time, though patient acceptance of AI-mediated records varied.\u003c/p\u003e\u003ch2\u003eConclusions:\u003c/h2\u003e\u003cp\u003eAI-assisted documentation enhances efficiency, accuracy, and stress reduction, with potential to improve quality of care. However, risks of new error types and nurse concerns necessitate safeguards. A SMART roadmap recommends integrating AI literacy into curricula by 2027, mandatory verification safeguards by 2028, and stress audits in all AI deployments by 2030.\u003c/p\u003e","manuscriptTitle":"Nursing Documentation in the AI Era: A Comparative Systematic Review and Meta-Analysis of Efficiency, Mistakes, Stress, and Quality of Care","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-30 07:39:19","doi":"10.21203/rs.3.rs-7718872/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":"September 30th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":55476918,"name":"Nursing"},{"id":55476919,"name":"Artificial Intelligence and Machine Learning"}],"tags":[],"updatedAt":"2025-09-30T07:39:19+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-30 07:39:19","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7718872","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7718872","identity":"rs-7718872","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}