General Medicine · Medical Literature
Cognitive Workload and Mental Burden in Health Care Professionals Interacting With AI: Systematic Review and Meta-Analysis.
Abstract
AI adoption in health care has accelerated rapidly, with ambient documentation tools, diagnostic imaging AI, and clinical decision support systems (CDSSs) entering routine practice. However, the cognitive demands placed on clinicians supervising these systems remain understudied. Specifically, the concept of verification burden requires closer examination. Consequently, institutional decision-makers lack a structured, certainty-graded evidence base regarding the true impact of AI on clinician workload and burnout. This study aimed to systematically review evidence on cognitive workload and burnout in health care professionals that use AI-powered clinical tools, quantify pooled effects under a conservative inferential framework, and assess certainty of evidence by AI category. The study was registered in PROSPERO (CRD420261284298) and reported per PRISMA 2020 and PRISMA-S guidelines. We searched MEDLINE, Embase, Web of Science, and Cochrane CENTRAL (January 2015-2026) for studies measuring cognitive workload or burnout using validated instruments (NASA Task Load Index [NASA-TLX] and Professional Fulfillment Index [PFI]) among health care professionals using clinical AI. Risk of bias was assessed using ROB 2.0 and ROBINS-I; certainty was rated using GRADE. Meta-analyses applied Hartung-Knapp-Sidik-Jonkman adjustment with restricted maximum likelihood estimation, incorporating prediction intervals (PIs). We included 21 studies representing 2885 health care professionals across 7 countries. The synthesis demonstrated that the cognitive impact of clinical AI varies according to its specific application. Pooled analyses of ambient AI documentation showed statistically significant reductions in NASA-TLX temporal demand (SMD -1.46, 95% CI -2.81 to -0.11; k=2; I2=31.1%) and effort (SMD -1.29, 95% CI -2.16 to -0.42; k=2; I2=0%), PFI work exhaustion (MD -0.35, 95% CI -0.58 to -0.12; k=3; I2=0%; 95% PI -1.03 to 0.33), and burnout prevalence (OR 0.47, 95% CI 0.25-0.86; k=3; I2=0%; 95% PI 0.06-3.82). Two pools favored ambient AI but did not reach significance at k=2: NASA-TLX mental demand (SMD -1.29, 95% CI -3.64 to 1.07) and documentation time (SMD -0.24, 95% CI -1.10 to 0.61). Diagnostic imaging AI and CDSS showed mixed or paradoxically increased workload. GRADE certainty was moderate for cognitive workload reduction with ambient AI, low for burnout reduction with ambient AI, and very low for imaging AI and CDSS outcomes. This review combines validated workload instruments, meta-analysis, and PIs in health care AI, delivering a GRADE certainty assessment across 5 AI categories that prior accuracy- or efficiency-focused reviews have not provided. Ambient AI documentation was associated with reduced cognitive workload and burnout, but only in voluntary early-adopter cohorts and based on few studies; the conservative CIs were wide and, where estimable, PIs crossed the null. Findings inform institutional pilots with prospective workload measurement, regulatory human-factors evaluation of AI medical devices, and human-centered AI design. Net benefit on the health care workforce remains an open empirical question.
Clinical relevance
Among health care professionals using AI-powered clinical tools (ambient documentation, diagnostic imaging AI, and clinical decision support systems), what is the pooled effect on cognitive workload and burnout as measured by validated instruments, and does that effect differ by AI category? __SNAPSHOT__ {"studyDesign":"Systematic review and meta-analysis (PROSPERO-registered: CRD420261284298; reported per PRISMA 2020 and PRISMA-S guidelines)","setting":"International; multi-database search (MEDLINE, Embase, Web of Science, Cochrane CENTRAL); specific institutional settings determined by included primary studies","sampleSize":"not reported in available abstract (number of included studies and total participant count not provided in the supplied text)","population":"Health care professionals using clinical AI tools; specific disciplines (physicians, nurses, allied health) and specialties determined by included primary studies","intervention":"Use of AI-powered clinical tools across three categories: ambient documentation tools, diagnostic imaging AI, and clinical decision support systems (CDSSs)","comparator":"Usual practice or pre-implementation baseline (as determined by included primary studies; not explicitly stated in the supplied abstract)","primaryOutcome":"Cognitive workload and burnout in health care professionals, measured with validated instruments: NASA Task Load Index (NASA-TLX) and Professional Fulfillment Index (PFI); effects quantified under a conservative inferential framework with certainty of evidence graded by AI category","followUp":"not reported in available abstract","fundingSource":"not reported in available abstract"}
Key findings
Specific pooled effect estimates (standardized mean differences, odds ratios, confidence intervals, p-values, I² heterogeneity statistics, GRADE certainty ratings by AI subcategory, and NNT/NNH) are not present in the truncated abstract provided and therefore cannot be reported without risk of fabrication. The study's stated aim was to quantify pooled effects under a conservative inferential framework and to assess certainty of evidence separately for ambient documentation tools, diagnostic imaging AI, and CDSSs—suggesting subgroup analyses were planned or conducted, but results are not available from the supplied text. The concept of 'verification burden'—the cognitive overhead incurred when clinicians must actively supervise and validate AI-generated outputs—was identified as an understudied and potentially adverse dimension of AI adoption. This framing is consistent with published evidence that poorly designed human–AI interaction can increase rather than reduce cognitive load in certain contexts. All specific numerical results are not reported from the available abstract.
Practice implications
IMPORTANT LIMITATION: Only a truncated abstract was supplied. No quantitative results, GRADE certainty ratings, or subgroup findings are available for synthesis; this summary is therefore necessarily incomplete and must not be used to guide clinical or institutional decisions without access to the full publication. What the study appears to have established: it provides a PROSPERO-registered, PRISMA-compliant framework for evaluating AI's impact on cognitive workload and burnout across heterogeneous AI categories using two validated psychometric instruments (NASA-TLX, PFI), addressing a gap in the structured evidence base available to institutional decision-makers. What it did not establish (from the available text): whether AI tools uniformly reduce or increase cognitive burden across all professional roles and AI types; whether any observed effects are durable over time; the magnitude, direction, and certainty of effects for each AI subcategory; and whether findings translate to patient safety or clinical outcomes. Patients and clinical contexts not represented: specific disciplines (e.g., nursing, pharmacy) and settings (e.g., low-resource environments, rural practice, non-Anglophone institutions) may not be adequately represented depending on the primary literature included. Practice implications: No change in clinical or institutional practice should be inferred from the abstract alone. The study's framework—requiring certainty grading by AI category and attention to verification burden as a potential harm—represents a methodologically important addition to the field if the full results support its aims. __LIMITATIONS__ The following limitations can be identified from the abstract and expected methodological considerations: (1) Only the truncated abstract is available; the full results, risk-of-bias assessments, heterogeneity statistics, and GRADE certainty ratings cannot be evaluated. (2) Reliance on self-reported, subjective instruments (NASA-TLX, PFI) introduces response bias and limits objectivity; these tools do not capture physiologic or performance-based workload. (3) Primary study heterogeneity in AI type, clinical setting, clinician role, follow-up duration, and outcome operationalization is expected to be substantial, potentially limiting interpretability of pooled estimates. (4) Search window (January 2015–2026) may capture an early and rapidly evolving literature with predominantly small, short-duration, non-randomized studies. (5) Publication bias likely favors studies reporting workload reduction. (6) The concept of 'verification burden' as a distinct, potentially adverse cognitive construct requires standardized operationalization not yet established in the literature. (7) Generalizability is constrained by the specialties, health systems, and AI platforms represented in included primary studies. (8) Funding source is not reported, precluding assessment of industry influence. (9) Without longer-term follow-up data in primary studies, effects on chronic burnout, deskilling, and automation dependence cannot be assessed. __BOTTOM_LINE_CATEGORY__ low_certainty __BOTTOM_LINE__ This PROSPERO-registered systematic review and meta-analysis (Gong, Bang, Lee; JMIR 2026) aims to provide the first certainty-graded, AI-category-specific evidence synthesis on cognitive workload and burnout in clinicians using AI tools, with particular attention to the underrecognized risk of verification burden. Quantitative results are not available from the truncated abstract supplied, and no practice-directing conclusions can be drawn at this time. Clinicians and institutional leaders should await the full publication before citing this work to support or oppose AI implementation decisions.