AI-Assisted Speech Documentation in Nursing: Identifying User Groups via Latent Class Analysis

AI-Assisted Speech Documentation in Nursing: Identifying User Groups via Latent Class Analysis | 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 Article AI-Assisted Speech Documentation in Nursing: Identifying User Groups via Latent Class Analysis Drin Ferizaj, Susann Neumann, Corina Burkhardt-Herdtle, Alexander Rau, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9557939/v1 This work is licensed under a CC BY 4.0 License Status: Under Revision Version 1 posted 3 You are reading this latest preprint version Abstract This study applied latent class analysis (LCA) to identify distinct technology acceptance groups among nursing staff using an AI speech assistant for nursing documentation (voize) in German long-term care facilities. Using a cross-sectional survey design ( N = 134) and three complementary indicator specifications derived from Unified Theory of Acceptance and Use of Technology (UTAUT2) constructs, we identified a consistent three-class solution with excellent entropy (.982): Champions (27.3%), Pragmatic Adopters (52.6%), and Reluctant Adopters (20.1%). Monte Carlo parameter-recovery simulations confirmed the robustness of this structure across simulated sample sizes from 100 to 1,000, with most recovery rates above 90%. Class profiles were validated through a Bayesian multivariate modelling approach with stacked posterior inference across 15 multiply imputed datasets, totaling 120,000 posterior draws. Compared with Pragmatic Adopters, Champions were more satisfied, perceived greater time savings, and were more willing to recommend the system, whereas Reluctant Adopters showed the opposite pattern across all practice-related outcomes. The clearest practical contrast was in likelihood to recommend: Champions rated the system 9.5 out of 10 on average, compared with 8.1 among Pragmatic Adopters and 4.7 among Reluctant Adopters. Champions endorsed nearly all three positive implementation outcomes on average—intention to continue use, better documentation, and better handovers—whereas Reluctant Adopters endorsed only about half. Age and duration of use did not meaningfully differentiate classes. These findings reveal distinct acceptance groups in AI-assisted nursing documentation and suggest that class-specific, targeted implementation strategies, rather than a uniform rollout, should be considered to support skeptical users while sustaining enthusiastic adoption. Health sciences/Health care Physical sciences/Mathematics and computing latent class analysis nursing automatic speech recognition mobile documentation Figures Figure 1 INTRODUCTION Nurses working in care homes and home care services constitute one of the largest groups of healthcare professionals in Germany, with nearly one million employed in this sector. 1 Their central role in patient care makes them primary users of health record systems and places them at the forefront of clinical and care documentation. While such documentation is critical for ensuring high-quality care, 2,3 documentation can consume up to 30% of a shift across nursing settings. 4 , 5 , 11 This considerable documentation burden, combined with fragmented work processes, frequent interruptions, and the urgent need to comply with regulatory standards, has been associated with lower job satisfaction and heightened job-related stress. 6 , 7 To reduce workload and improve both care quality and staff well-being, care facilities are increasingly considering digital tools. Their impact depends on whether nurses can and will use them in everyday practice. One promising option for nursing documentation is automatic speech recognition (ASR), which converts spoken language into structured entries or notes, enabling hands-free operation, faster capture, and greater mobility in daily work. 8 – 11 Evidence from nursing settings suggests that ASR can improve efficiency, accuracy, and user satisfaction. 9 , 11 Related mobile bedside documentation research also points to workflow and time-saving benefits. 10 Consistent with this, a recent pre-post time-motion study in German long-term care found that implementing a speech AI documentation system was associated with substantially less documentation time, fewer interruptions, and higher satisfaction with the documentation system. 11 Despite these benefits, low acceptance may still limit adoption, as shown in early computer-based documentation 12 and more recent mobile ASR documentation research. 9 To explain why some healthcare professionals adopt a given system while others hesitate, the Unified Theory of Acceptance and Use of Technology and its consumer extension (UTAUT2) offers a validated framework. It aims to explain technology adoption through seven core constructs: Performance Expectancy (perceived job benefits), Effort Expectancy (ease of learning and use), Social Influence (normative pressure from colleagues and supervisors), and Facilitating Conditions (organizational and technical support). Consumer-focused extensions add Hedonic Motivation (enjoyment of use), Habit (automaticity of adoption), and Price-Value (cost-benefit assessment). 13 In healthcare, these constructs are repeatedly associated with users’ behavioral intention and usage, 14,15 with performance expectancy and effort expectancy often emerging as influential predictors. 13 , 14 , 16 Extensions of these models in healthcare commonly add trust and satisfaction, and in some settings privacy or security concerns. 17 In general, these frameworks yield estimates at the population mean and do not reveal whether distinct subgroups of nurses hold qualitatively different configurations of these perceptions. Therefore, most acceptance research has been “variable-centered,” estimating paths among constructs and outcomes at the level of the average user. 13 , 14 Yet, in real care environments, nurses can hold qualitatively different attitudes and experiences that may cluster into distinct groups. 18 When the goal is to derive such classes from empirical data, latent class analysis (LCA) is a feasible approach. LCA is a person-centered statistical method that uncovers hidden latent subgroups within a population based on response patterns across multiple observed variables, 19–21 shifting attention from average effects to co-occurring patterns of attitudes within individuals. Specifically, LCA is a probabilistic finite-mixture model that attributes heterogeneity to a discrete set of latent classes, each defined by characteristic item-response probabilities. In this context, class membership rather than direct item-to-item associations accounts for observed correlations among indicators. 19 , 20 Evidence from LCA studies highlights the usefulness of segmentation, although most were conducted in non-nursing populations and did not specifically assess ASR. Among university students, an LCA of attitudes toward AI-enabled symptom checkers identified five profiles ranging from “tech acceptors” to “tech rejectors,” with class membership strongly associated with intention to use the tool. 22 In mental health training, psychotherapists-in-training assessed with UTAUT-derived indicators formed two acceptance classes that differed primarily on performance expectancy and effort expectancy, with little differentiation by age or gender. 23 Latent profile analysis (LPA), a method for identifying subgroups based on continuous variables, has been applied to characterize nursing competencies and related technology perceptions. In a nationwide sample of 3,610 Finnish registered nurses, three informatics-competence profiles differed in perceived usefulness of health information systems, with higher-competence profiles reporting greater usefulness. 24 Similarly, an LPA of nurses in traditional Chinese medicine hospitals identified low, moderate, and high innovative-behavior profiles, with profile membership associated with structural empowerment, adversity quotient, training, competency, and several work-related characteristics. 25 These studies collectively strengthen the case for person-centered modeling but also delineate a clear gap for long-term care nursing and ASR technologies. First, most ASR acceptance work, when framed by UTAUT, has remained variable-centered, focusing on average paths rather than qualitatively distinct user classes. 13 Second, the person-centered LCA/latent profile analysis literature has largely profiled students, the public, or constructs adjacent to acceptance such as competence or innovation behavior, rather than segmenting practicing nurses with actual exposure to a specific AI documentation tool in routine care. The only comparable LCA study from Germany targeted the general public rather than practicing nurses and assessed broader digital health services rather than a specific AI documentation workflow, leaving the question of user classes among long-term care professionals empirically open. 18 Our study addresses this gap by applying LCA to nurses in German long-term care facilities who have hands-on experience with an AI speech assistant for documentation. We combine a theory-driven set of UTAUT2 indicators—capturing perceived usefulness, effort, facilitating conditions, habit, enjoyment, social influence, and trust—with a data-driven segmentation that recognizes qualitatively different user mindsets rather than assuming homogeneity. Competing class solutions are estimated across multiple indicator specifications, with several information criteria and entropy used for class selection, and classes are validated against practical outcomes including satisfaction, frequency of use, perceived time savings, perceived improvements, and recommendation. Accordingly, we address three central research questions: 1. How many distinct user classes of AI speech-assistant adoption can be identified among nursing staff in senior care based on UTAUT2-related perceptions? 2. What are the defining response-pattern characteristics of each class across the UTAUT2 constructs? 3. How do classes differ on auxiliary acceptance and usage outcomes, including satisfaction, frequency of use, perceived time savings, recommendation, and intention to continue use? METHOD Study Design and Participants This study employed a cross-sectional survey design to examine user acceptance patterns of an ASR documentation system (“voize,” provided by voize GmbH, Berlin, Germany) among nursing staff in German senior care facilities between August 2024 and January 2025. The research was conducted as part of the PYSA project (“Pflegedokumentation mit hybridem Sprachassistenten”), 26 funded by the German Federal Ministry of Education and Research (now called Federal Ministry of Research, Technology and Space). Data were collected via a voluntary and anonymous online survey administered through the REDCap platform. Recruitment was conducted exclusively through the voize app, ensuring that all participants had direct experience with the speech documentation assistant. In addition to purposive sampling, a convenience component was used to maximize participation across varying care contexts. The survey was open to staff at four cooperating care facilities in Germany. Eligible participants were adults aged 18 years or older who had used the voize documentation assistant in their daily work within the previous six months and could understand written German instructions. Before completing the questionnaire, all participants received study information and provided written informed consent. To protect privacy and reduce identifiability, the survey did not collect information about respondents’ specific care facilities and included only minimal sociodemographic data. Participation was voluntary, anonymous, noninterventional, and not monetarily incentivized. All respondents who completed at least 10% of the questionnaire were included in the final analysis. No formal ethics vote was obtained because the study involved only adult professional users and only collected no patient-related data. The Ethics Committee of Charité - Universitätsmedizin Berlin would generally be the responsible institutional ethics body for the underlying research project. All procedures followed established ethical principles, including the Declaration of Helsinki and Good Clinical Practice guidelines, and ensured informed consent, voluntary participation, anonymity, and the right to discontinue participation at any time. Measures and Survey Instrument The questionnaire was developed collaboratively by two nursing experts and three of the authors. An initial set of items was drafted based on previous research. 8 Five reviewers evaluated each item on a binary scale (0 = reject, 1 = keep). Items scoring below three were excluded, those scoring five were retained, and those scoring four were discussed until consensus was reached. The final version was pretested by two nurses under realistic conditions, and no major modifications were made. The full translated English and German versions as well as an overview of included constructs, individual items, and response formats are provided in Appendix A. The instrument operationalized UTAUT2 constructs alongside demographic variables and auxiliary measures. Multi-item UTAUT2 constructs were Performance Expectancy (three items; e.g., “Using voize reduces my documentation workload”), Effort Expectancy (four items; e.g., “voize is easy to learn”), Facilitating Conditions (two items; e.g., “I can get quick help if needed”), and Habit (two items; e.g., “I use voize almost automatically during my shift”). Single-item indicators captured Hedonic Motivation (“Using voize is enjoyable”), Social Influence (“My team supports me using voize”), and Trust in Data Security (“Data security is ensured”). All UTAUT2 items were rated on a five-point Likert scale (1 = strongly disagree to 5 = strongly agree). Price Value was not included in this study as the costs for the software were covered by the employer. Higher construct scores indicate, respectively, stronger perceived benefit, lower perceived effort, greater organizational support, more ingrained use, greater enjoyment, stronger social influence, and higher trust. The following self-reported auxiliary outcomes were collected: satisfaction with the documentation system (1–5), usage frequency (1–5), perceived time saved per shift (rated on a five-point scale: 1 = no time saved, 2 = 5–10 min, 3 = 10–20 min, 4 = 20–30 min, 5 = 30–45 min), and likelihood to recommend the documentation system (0–10; higher values indicating higher agreement). Initial reaction was rated on a four-point scale (1 = not happy at all; 4 = very happy about the implementation). Technology affinity was rated on a seven-point scale (1–7). Age was recorded in six ordinal categories (1 = under 21 years, 2 = 21–30, 3 = 31–40, 4 = 41–50, 5 = 51–60, 6 = over 60 years). Three binary items (0 = no, 1 = yes) assessed perceived improvements in documentation quality, handover quality, and intention to continue using the system. Internal consistency and validity were evaluated via Cronbach’s α, composite reliability, average variance extracted, and confirmatory factor analysis. Reliability was good and overall model fit was acceptable (see Appendix B). Statistical Analysis Latent Class Analysis Specifications The goal of the LCA was to identify distinct acceptance classes from observed responses. We estimated a series of latent class models with K ∈ {1, 2, 3, 4, 5} classes under three complementary indicator specifications: (1) the full-item specification retained all 14 original indicators; (2) the statistically sparse specification reduced redundancy by removing one variable from pairs with absolute correlations exceeding r > .70 and collapsing rare response categories below 10%, yielding 12 indicators; and (3) the theory-driven UTAUT2 specification used multi-item composites for Performance Expectancy, Effort Expectancy, Facilitating Conditions, and Habit, and single-item indicators for Hedonic Motivation, Social Influence, and Trust. Analyses were conducted using the poLCA package in R, 21 with models estimated via maximum likelihood with 250 random starts to avoid local maxima. Model Selection Model selection was guided by multiple fit indices: Bayesian Information Criterion (BIC), Akaike Information Criterion (AIC), Sample-size Adjusted BIC (SABIC), BIC-approximated Bayes Factor comparing adjacent models (BF; values > 10 indicating strong evidence for adding a class), and Entropy (values closer to 1 indicating better class separation). Primary consideration was given to BIC (lower = better), Entropy, and the Bayes Factor comparing K vs. K − 1 class solutions. Monte Carlo Simulations Latent class analysis typically benefits from large samples (N > 500). 20 To verify our identified solution’s robustness at N = 134, we conducted Monte Carlo parameter-recovery simulations. Parameters were extracted from the retained three-class LCA models and used to generate synthetic datasets: 1,500 replications at N = 100, 134, 200, 300, 500, and 1,000 for both the UTAUT2 and full-indicator models. Each dataset tested 1 to 5 classes with 250 random starts each, with BIC as the primary criterion for class extraction. Class Validation With Auxiliary Variables Missing data for auxiliary outcomes ranged from 4.5% to 20.9% (overall 15.1%; 77.6% complete cases), with moderate associations between missingness and UTAUT2 indicators (correlations ranging from − .30 to .33). We created m = 15 multiply-imputed datasets using Random Forest Predictive Mean Matching with 10 iterations, K = 5 nearest neighbors, and 250 trees. Eight outcomes were modeled jointly in a single Bayesian multivariate normal regression in Python 3.12 using PyMC 5.28.2 34 and ArviZ 1.0.0 35 : usage frequency, perceived time saved per shift, satisfaction, age group, likelihood to recommend, technology affinity, voize usage duration, and a three-item formative index indicating overall perceived benefits (the unweighted sum of three binary outcomes: intention to continue use, perceived documentation quality improvement, and perceived handover improvement; range 0–3). The sole predictor across all outcomes was latent class membership, with Pragmatic Adopters serving as the reference category. Class membership was operationalized via soft class weights, propagating posterior membership probabilities from the LCA directly into the likelihood. Residual covariance used a Lewandowski-Kurowicka-Joe (LKJ) prior (η = 3), and outcome-specific residual standard deviations received half-Student-t priors (ν = 4, σ = 1.0). Contrast coefficients were assigned hierarchical shrinkage priors (half-Student-t global scale τ ~ ν = 3, σ = 0.35). Models were estimated within each imputed dataset using the JAX NumPyro NUTS backend (4 chains, 2,000 post-warmup draws per chain, 2,500 warmup steps, target acceptance = 0.99). Posteriors were stacked across imputations, yielding 15 × 4 × 2,000 = 120,000 total draws. Convergence was assessed via R̂, Bayesian Fraction of Missing Information (BFMI), and effective sample size (ESS). For each class contrast we report stacked posterior means, 95% highest-density intervals (HDIs), and the direction of the posterior as indicators of directional robustness. RESULTS Sample Description The study sample comprised 134 participants, predominantly qualified nurses (39.4%) and nursing assistants (35.1%). The remaining roles included nursing service managers (8.5%), caregivers (8.2%), ward managers (4.4%), and trainees (4.4%). Regarding age, the largest cohort fell within the 41–50 year range (26.9%), followed by those aged 51–60 (23.1%) and 31–40 (20.9%); fewer participants were aged 21–30 (13.4%), over 60 (12.7%), or under 21 (3.0%). Usage duration ranged from 0–4 weeks (11.9%), 5–8 weeks (32.1%), 9–12 weeks (29.9%), to over 12 weeks (26.1%). Descriptive Results Table 1 presents descriptive statistics for all study variables before and after multiple imputation. Overall, participants reported positive initial reactions about the implementation of the speech assistant (M = 3.69, SD = 1.06 on a 4-point scale) and strong technology affinity (M = 5.52, SD = 1.42 on a 7-point scale); multiple imputation produced negligible changes across all variables (all |Δ| < 1.6%), supporting the adequacy of the imputation procedure. Table 1 Descriptive Statistics for All Study Variables: Complete and Imputed Data Variable M (SD) Median Min Max M (SD) Median Δ Complete Data Imputed Data Sociodemographic Variables Age Group (1–6) 3.92 (1.33) 4.0 1 6 3.92 (1.33) 4.0 0.0% Technology Affinity (1–7) 5.52 (1.42) 6.0 1 7 5.52 (1.42) 6.0 0.0% UTAUT2 Variables Performance Expectancy (α = .86) Work Relief (1–5) 3.82 (1.08) 4.0 1 5 3.81 (1.08) 4.0 −0.2% Documents More (1–5) 3.62 (1.06) 4.0 1 5 3.58 (1.08) 4.0 −1.2% More Time for Residents (1–5) 3.88 (1.06) 4.0 1 5 3.91 (1.03) 4.0 + 0.8% Effort Expectancy (α = .83) Easy to Learn (1–5) 3.93 (0.94) 4.0 1 5 3.95 (0.92) 4.0 + 0.4% Ease of Use (1–5) 3.97 (0.90) 4.0 1 5 3.97 (0.88) 4.0 −0.1% Reliable Speech Recognition (1–5) 3.58 (1.01) 4.0 1 5 3.54 (1.01) 4.0 −1.1% System Reliability (1–5) 3.80 (1.06) 4.0 1 5 3.76 (1.07) 4.0 −1.0% Facilitating Conditions (α = .80) Quick Help Available (1–5) 4.02 (0.93) 4.0 1 5 4.01 (0.93) 4.0 −0.4% Enough Time to Learn (1–5) 3.84 (1.01) 4.0 1 5 3.83 (0.99) 4.0 −0.3% Habit (α = .73) Habit Formation (1–5) 4.03 (1.03) 4.0 1 5 4.04 (1.01) 4.0 + 0.2% Independent Use (1–5) 4.15 (0.96) 4.0 1 5 4.14 (0.97) 4.0 −0.3% Additional UTAUT2 Variables Enjoyment (1–5) 3.84 (1.09) 4.0 1 5 3.81 (1.10) 4.0 −0.8% Team Supports voize (1–5) 3.62 (0.89) 4.0 1 5 3.59 (0.88) 3.5 −0.8% Trust in Data Security (1–5) 4.08 (0.90) 4.0 2 5 4.07 (0.90) 4.0 −0.3% Auxiliary Variables Initial Reaction (1–4) 3.69 (1.06) 4.0 1 4 3.69 (1.06) 4.0 −0.2% Usage Frequency (1–5) 4.19 (1.20) 5.0 1 5 4.18 (1.22) 5.0 −0.3% Usage Duration (1–4) 2.70 (0.99) 3.0 1 4 2.72 (1.00) 3.0 + 0.5% Time Saved per Shift (1–5) 3.03 (1.37) 3.0 1 5 3.07 (1.30) 3.0 + 1.1% Satisfaction (1–5) 4.01 (0.95) 4.0 1 5 3.98 (0.92) 4.0 −0.7% Recommendation (0–10) 7.89 (2.68) 9.0 0 10 7.77 (2.82) 9.0 −1.4% Binary Auxiliary Variables (% Yes) Quality Improved (%) 84.3% — — — 83.5% — −0.8% Handover Improved (%) 64.2% — — — 65.8% — + 1.6% Intention to Continue (%) 90.9% — — — 91.5% — + 0.6% Note. Δ = percentage difference between original and pooled, imputed means. Usage duration categories: 1 = 0–4 weeks, 2 = 5–8 weeks, 3 = 9–12 weeks, 4 = over 12 weeks. All UTAUT2 items were rated on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree). Dashes indicate that median/range are not applicable. Participants showed generally positive responses, with high recommendation scores (M = 7.89, SD = 2.68) and strong intentions to continue use (90.9%). Most reported quality improvements (84.3%) and better handovers (64.2%). The majority reported saving time per shift: 22.7% saved 5–10 minutes, 19.3% saved 10–20 minutes, 22.7% saved 20–30 minutes, 18.5% saved 30–45 minutes, and 16.8% reported saving no time. Participants demonstrated regular usage patterns (usage frequency M = 4.19, SD = 1.20 on a 5-point scale). Latent Class Analysis Model Selection Table 2 presents model fit indices for the UTAUT2 specification across all class solutions; full results for the sparse and full-item specifications are reported in Appendix C. The three-class solution was consistently supported as optimal across all three indicator specifications. The three-class UTAUT2 solution yielded the lowest BIC (BIC = 2,114.55), with the Bayes Factor providing strong evidence for three over two classes (BF = 1.38 × 10¹²) and overwhelming evidence against four classes (BF = 4.92 × 10⁻²¹). Table 2 Model Fit Criteria Across Class Solutions: UTAUT2 Specification Model Classes LL AIC BIC SABIC BF UTAUT2 1 −1,131.13 2,316.25 2,394.50 2,309.09 — 2 −950.54 2,011.08 2,170.46 1,996.48 4.46e + 48 3* −854.02 1,874.03 2,114.55 1,852.01 1.38e + 12 4 −832.21 1,886.42 2,208.08 1,856.96 4.92e − 21 5 −815.30 1,908.59 2,311.39 1,871.70 3.68e − 23 Note. LL = log-likelihood. AIC = Akaike Information Criterion (lower = better). BIC = Bayesian Information Criterion (lower = better); primary selection criterion. SABIC = sample-size adjusted BIC. BF = BIC-approximated Bayes Factor comparing K vs. K − 1 class solutions; values > 10 indicate strong evidence for the larger solution, values < 0.1 indicate evidence against. * Retained solution. Full results for the full-item and sparse indicator specifications are reported in Appendix C. Diagnostic criteria further supported the three-class solution (Table 3 ). Entropy values for the three-class UTAUT2 solution were excellent (.982), indicating near-perfect class separation (with values approaching 1.0 reflecting minimal classification uncertainty). Average latent class posterior probabilities (ALCPP) exceeded .97 in all cases, meaning that on average, individuals were assigned to their most probable class with over 97% certainty. The smallest class in the retained three-class solution contained 27 participants (20.1%), exceeding recommended minimum class sizes. Consistent results across all three indicator specifications are reported in Appendix C. Table 3 Diagnostic Criteria Across Class Solutions: UTAUT2 Specification Model Classes Smallest n Smallest % Entropy ALCPP UTAUT2 1 134 100.0 — — 2 43 32.0 0.951 0.986 3* 27 20.1 0.982 0.978 4 15 11.5 0.952 0.972 5 13 9.5 0.956 0.971 Note. ALCPP = Average Latent Class Posterior Probability; values closer to 1.0 indicate that individuals are assigned to their modal class with high certainty (values ≥ .90 are considered acceptable). Entropy ranges from 0 to 1; values ≥ .80 indicate good class separation, values ≥ .90 indicate excellent separation. * Retained solution. Comparable results for the full-item and sparse specifications are reported in Appendix C. Monte Carlo Simulations Monte Carlo simulations confirmed the robustness of the three-class solution. Recovery rates exceeded .90 in all conditions for both the UTAUT2 and full-indicator specifications, with the sole exception of the UTAUT2 model at N = 100 (recovery = 83%; full-indicator model at N = 100: 96%). From N = 134 onward, both models achieved 100% recovery, near-perfect convergence, and entropy > .90. The optimal three-class solution showed excellent class separation across both specifications (UTAUT2 entropy = .962). These results indicate that the three-class structure is a robust feature of the data and not a statistical artifact of the sample size. Class Profiles The three-class solution was consistent across indicator specifications, with highly similar class proportions and within-class response profiles. Figure 1 illustrates the mean item responses by class for the UTAUT2 specification. Champions (27.3%) These users embrace voice documentation wholeheartedly. Champions demonstrate exceptional trust in data security (M = 4.98 out of 5), derive strong pleasure from the system (Hedonic Motivation: 4.91), and perceive substantial performance benefits (Work Relief: 4.87). Their relationship with the technology appears to transcend mere utility: they have thoroughly integrated voice documentation into their daily routines (Habit: 4.85) and experience pronounced workload reduction. Even their most modestly rated constructs—perceived team support (4.72) and system reliability (4.71)—remain strongly positive, suggesting comprehensive satisfaction across all facets of the system. Pragmatic Adopters (52.6%) Representing the majority, Pragmatic Adopters display practical acceptance without extraordinary enthusiasm. Their profile emphasizes facilitating conditions (M = 4.76 out of 5) and trust (3.98), with effort expectancy moderately positive (Ease of Use: 3.91). Performance benefits (3.73) and enjoyment (3.73) receive overall positive but measured ratings. These users appear to view voice documentation as a functional tool that adequately meets professional needs - valuing organizational support and (technical) reliability without the strong emotional engagement characteristic of Champions. Reluctant Adopters (20.1%) Among the three classes, Reluctant Adopters show the most restrained engagement with the mobile voice documentation. Compared with the other groups, they score lowest on hedonic motivation (M = 2.50 out of 5), perceived ease of use (2.63), and performance expectancy (Work Relief: M = 2.66), all of which fall below the scale midpoint. Habit and trust, their relatively highest-rated constructs, nonetheless remain substantially below the levels of the other groups (Habit: M = 3.15; Trust: M = 3.13). Furthermore, this indicates that these users do engage with the system on a regular basis. Class Validation: Bayesian Multivariate Modelling across Auxiliary Outcomes Convergence and Sampling Quality Sampling converged cleanly across all 15 imputed datasets. Zero divergent transitions were observed in every imputation, and maximum R̂ did not exceed 1.002 in any run. BFMI ranged from 0.85 to 0.89 across imputations, all within acceptable bounds. Minimum bulk effective sample sizes exceeded 4,200 per imputation, well above the recommended threshold of 400. Stacking the 15 × 4 × 2,000 = 120,000 posterior draws yielded a well-characterized posterior suitable for robust inference. Posterior Class Means Table 4 presents the stacked posterior means and 95% HDIs for each class across all eight outcomes. The ordering of classes on substantive outcomes was consistent and theoretically coherent: Champions had the highest posterior means across implementation-relevant outcomes, and Reluctant Adopters the lowest. Age-group differences were small and directionally uncertain across classes. Pragmatic Adopters occupied an intermediate position—closer to Champions on usage frequency, satisfaction, and recommendation, but closer to Reluctant Adopters on time saved. Table 4 Stacked Posterior Class Means and 95% HDIs for Auxiliary Outcomes Outcome Champions Reluctant Adopters Pragmatic Adopters (Reference) Scale Usage Frequency 4.51 [4.17, 4.85] 3.61 [3.14, 4.10] 4.22 [3.97, 4.48] 1–5 Time Saved per Shift 3.93 [3.53, 4.33] 2.29 [1.83, 2.74] 2.90 [2.63, 3.18] 1–5 Satisfaction 4.75 [4.58, 4.92] 2.72 [2.47, 2.98] 3.96 [3.83, 4.09] 1–5 Recommendation 9.45 [8.83, 10.00] 4.65 [3.76, 5.52] 8.14 [7.66, 8.59] 0–10 Technology Affinity 6.37 [6.00, 6.74] 4.00 [3.51, 4.50] 5.56 [5.28, 5.84] 1–7 Age Group 4.13 [3.73, 4.56] 3.85 [3.40, 4.29] 3.81 [3.50, 4.12] 1–6 Usage Duration 2.86 [2.57, 3.16] 2.40 [1.99, 2.80] 2.73 [2.52, 2.94] 1–4 Three-Item Formative Index 2.77 [2.55, 3.00] 1.53 [1.21, 1.85] 2.48 [2.31, 2.65] 0–3 Note. Posterior means and 95% highest-density intervals (HDIs) from 120,000 stacked draws (m = 15 imputations × 4 chains × 2,000 draws). Three-Item Formative Index = unweighted sum of three binary outcomes (intention to continue use + quality improved + handover improved; range 0–3). Pragmatic Adopters served as the reference group in contrast models. Posterior Contrasts Table 5 presents the posterior contrasts for each class relative to Pragmatic Adopters (reference). Table 5 Stacked Posterior Contrasts for Auxiliary Outcomes (Reference: Pragmatic Adopters) Outcome (Scale) Champions vs. Pragmatic Adopters Mean [95% HDI] Reluctant Adopters vs. Pragmatic Adopters Mean [95% HDI] Direction Interpretation Usage Frequency (1–5) + 0.29 [− 0.10, 0.70] −0.61 [− 1.15, − 0.07] ~ / ↓ Reluctant Adopters use the tool less frequently; Champions show no clear frequency advantage Time Saved per Shift (1–5) + 1.03 [0.53, 1.52] −0.62 [− 1.13, − 0.09] ↑ / ↓ Champions save meaningfully more time; Reluctant Adopters save less Satisfaction (1–5) + 0.79 [0.57, 1.00] −1.24 [− 1.52, − 0.95] ↑ / ↓ Clear gradient across all three classes Recommendation (0–10) + 1.31 [0.55, 2.09] −3.49 [− 4.49, − 2.49] ↑ / ↓ Largest absolute effect; Reluctant Adopters score 3.5 points below Pragmatic Adopters Technology Affinity (1–7) + 0.81 [0.34, 1.27] −1.56 [− 2.12, − 0.98] ↑ / ↓ Class membership co-varies with broader dispositional orientation toward technology Age Group (1–6) + 0.32 [− 0.12, 0.88] + 0.03 [− 0.46, 0.54] ~ / ~ No meaningful age differences; age cannot guide segmentation Usage Duration (1–4) + 0.13 [− 0.17, 0.49] −0.33 [− 0.77, 0.08] ~ / ~ Class membership is not explained by length of exposure Three-Item Formative Index (0–3) + 0.30 [0.02, 0.57] −0.95 [− 1.32, − 0.59] ↑ / ↓ Champions endorse nearly all adoption outcomes; Reluctant Adopters endorse fewer than half Note. Effects are stacked posterior mean differences on the original scale with 95% highest-density intervals (HDIs). Direction column: ↑ = Champions robustly higher than Pragmatic Adopters (HDI entirely above zero); ↓ = Reluctant Adopters robustly lower than Pragmatic Adopters (HDI entirely below zero); ~ = inconclusive (HDI crosses zero). Format: Champions / Reluctant Adopters direction. Three-Item Formative Index = unweighted sum of three binary outcomes (range 0–3). Champions consistently scored higher than Pragmatic Adopters, with clear differences in satisfaction (+ 0.79 [0.57, 1.00]), time saved per shift (+ 1.03 [0.53, 1.52]), recommendation (+ 1.31 [0.55, 2.09]), technology affinity (+ 0.81 [0.34, 1.27]), and the three-item formative index (+ 0.30 [0.02, 0.57]). Usage frequency (+ 0.29 [− 0.10, 0.70]) and usage duration (+ 0.13 [− 0.17, 0.49]) were directionally positive but uncertain, suggesting that Champions and Pragmatic Adopters use the system at comparable rates but differ qualitatively in their experience. Reluctant Adopters showed the inverse pattern with clear deficits across most outcomes. Recommendation was markedly lower (− 3.49 [− 4.49, − 2.49]), with a mean of 4.65 representing a 3.5-point gap below Pragmatic Adopters’ 8.14; satisfaction was also substantially reduced (− 1.24 [− 1.52, − 0.95]), falling below the scale midpoint (M = 2.72); and technology affinity showed a robust negative contrast (− 1.56 [− 2.12, − 0.98]). The three-item formative index contrast of − 0.95 [− 1.32, − 0.59] indicates that Reluctant Adopters—with a mean of 1.53 out of 3.0—endorse only about half the binary adoption outcomes, compared with Pragmatic Adopters (2.48) and Champions (2.77). Directly comparing Champions and Reluctant Adopters, Champions exceeded Reluctant Adopters by 2.03 satisfaction units, 4.80 recommendation points, 2.37 technology affinity units, 1.64 time-saved units, and 1.25 three-item formative index points, underscoring the qualitative distinctiveness of these profiles for implementation planning. Critically, neither age group nor usage duration showed directionally robust contrasts, suggesting that these variables were not strong correlates of class membership in this sample. Interestingly, the identified Champions had the highest descriptive mean age across the classes. Posterior predictive checks confirmed adequate model fit, with model predictions matching observed data within 3% for all outcomes. DISCUSSION Principal Findings The latent class analysis identified three acceptance classes among practicing nurses who used an AI-based speech assistant for documentation. Champions, approximately one quarter of participants, combined the highest scores on all UTAUT2-derived constructs—including performance expectancy, effort expectancy, hedonic motivation, habit, facilitating conditions, social influence, and trust—with the strongest behavioral and attitudinal outcomes: the highest perceived improvements (as indicated by the formative index) and recommendation rates, greater satisfaction, and greater perceived time savings. Pragmatic Adopters, approximately half of the sample, endorsed facilitating conditions, trust, and ease of use, while reporting moderate enjoyment and perceived performance gains. Their outcomes were solid and characteristic of functional adoption. Reluctant Adopters, approximately one fifth of participants, reported lower performance and effort expectancy and lower enjoyment, alongside clearly weaker outcomes across satisfaction, recommendation, and usage behavior. Taken together, the pattern suggests that core UTAUT2 constructs do not operate in isolation but cluster into coherent user classes that align with meaningful differences in actual and perceived use. Comparison With Previous Work The Champions class conceptually resembles higher-acceptance or innovation-ready classes reported in adjacent literatures. In radiology, latent class analysis distinguished high- versus low-acceptance clinicians, and acceptance interacted with workload and AI use in explaining burnout, indicating that acceptance heterogeneity has clinical consequences. 28 In higher education, a faculty segmentation identified optimistic, critical, critically reflected, and neutral classes. In this context, the optimistic class showed the strongest link between AI self-efficacy and AI usage, echoing the high-engagement orientation similar to the Champions class. 29 Within nursing, related findings have recovered a class characterized by comparatively higher informatics-competence scores associated with greater perceived usefulness of health information systems, 24 an optimal innovator class among U.S. nurses, 32 and a class reflecting higher innovative behavior among nurses in Chinese hospitals. 25 A population-level study in Germany also identified an active user class for digitalized services. Although that analysis concerned citizens rather than professionals, the active class parallels a high-adoption pattern. 18 Across adjacent findings, the prevalence of high-acceptance classes in regard to technology varies substantially by population and context: the active class in the German citizen study comprised 18.3%, and the optimistic faculty profile roughly one third — broadly consistent with Champions here. In the purposively sampled U.S. innovation-engaged nurse cohort, the 'optimal innovator' class was far larger with about 83%, underscoring that prevalence is highly context-dependent 18 , 29 , 32 Pragmatic Adopters align with two streams of evidence. At the citizen level, the largest class in the German population study comprised 'Potential' users who had not yet used many digitalized healthcare services but were in principle willing to do so. They were younger, more educated, more frequently employed, and less skeptical than rejecters. 18 This functional profile maps closely onto the present Pragmatic Adopters, who likewise reported moderate but positive perceptions under supportive conditions. Another study among Finnish registered nurses identified groups with comparatively lower, moderate, and higher informatics competence, with higher-competence groups reporting greater perceived usefulness of the health information system.²⁴ While competence is related but not equivalent to acceptance, 30 the finding that capability groupings align with perceived usefulness helps explain why Pragmatic Adopters may sustain steady use when facilitating conditions are present. Reluctant Adopters mirror lower-acceptance segments described in prior research. In the German citizen study, a “Rejecting” class was characterized by skepticism and lower readiness to use digitalized services. 18 Hesitancy in adoption is generally accompanied by diminished trust in system reliability, elevated concern over data security, lower enjoyment, and more negative evaluations of the technology such as perceived performance expectancy and effort expectancy. 23 Wang and colleagues 31 found that even participants with high levels of objective AI knowledge could retain strongly negative attitudes and moderate behavioral intentions if trust or risk perception was unfavorable, suggesting that knowledge alone does not alleviate skepticism. Similarly, Leary and colleagues 32 identified four classes among nurses engaged in innovative behaviors and emphasized the importance of organizational support, innovation exposure, and protected time for fostering innovativeness. Theoretical and Practical Implications The findings extend UTAUT2 by showing that core constructs tend to cluster into stable user groups rather than operating as isolated levers. Champions combined high perceived usefulness, ease, trust, habit, and enjoyment; Pragmatic Adopters exhibited a functional pattern in which facilitating conditions and ease were present but affective engagement and perceived gains were moderate; Reluctant Adopters scored consistently lower across all domains. This suggests that positive perceptions of usefulness, ease of use, trust, habit, and enjoyment tend to co-occur and highlights hedonic motivation as a practically important correlate of higher acceptance in clinical documentation when systems are reliable and are aligned with routine workflows. The speech modality may be experienced as pleasant to use in routine care, which, together with perceived workload relief, can strengthen sustained engagement. 8 , 9 , 11 Implementation strategies can be tailored accordingly. Champions could serve as peer mentors and early validators who model effective use in realistic workflows. 11 Pragmatic Adopters benefit from robust facilitating conditions—including rapid support and opportunities for practice—that maintain perceptions of ease. 24 , 28 , 31 Reluctant Adopters require confidence-building that addresses perceived reliability, effort, and risk directly. 24 Staged onboarding, protected time to integrate voice commands, and transparent feedback on recognition accuracy are likely to be more effective than generic training programs. Qualitative fieldwork on AI speech documentation highlights the importance of training, social influence, continuous multimodal technical support, and iterative refinement based on user feedback for successful adoption. 8 , 33 Limitations Several limitations should be considered. First, the cross-sectional design precludes causal inference about how classes emerge or evolve with experience. Second, the sampling frame included active users in organizations where the speech assistant was already deployed, so non-users and early refusers were likely not captured, which may bias estimates toward more favorable experiences. Third, outcomes were self-reported and may be affected by social desirability or recall bias; time-motion measures would provide stronger validation. 10 , 33 Fourth, trust and hedonic motivation were assessed with single-item indicators, limiting measurement depth. Fifth, generalizability is bounded by the focus on German long-term care facilities and a single AI speech assistant. Outlook and Future Directions Longitudinal designs would add considerable depth by testing whether users transition across classes as accuracy, training, and workflow fit evolve over time. Latent transition or growth mixture models can quantify movement between profiles and identify predictors of shifts from Reluctant Adopter to Pragmatic Adopter or Champion status. With larger samples, Bayesian network or structural equation approaches could test hypothesized directional relationships among UTAUT2 constructs, moving beyond descriptive classification toward models that specify which constructs actually drive acceptance and represent bridges between constructs. Complementarily, once classes have been identified in larger samples, network analysis methods could be applied within and across classes to identify central (hub) variables that sustain positive acceptance and bridge variables whose strengthening could facilitate movement between acceptance profiles. Such analyses would help prioritize intervention targets within each segment. Class-tailored interventions could also be developed and trialed: peer-mentor networks and co-design workshops with Champions to identify core drivers for adoption, reinforced support and micro-learning exercises for Pragmatic Adopters to strengthen existing capabilities, and enhanced support and requirements analysis for Reluctant Adopters, with class-specific success criteria defined in advance. Finally, integrating objective system analytics alongside self-report measures would strengthen the validity of future evaluations. Conclusions The proposed three-class taxonomy suggests how nurses in long-term care relate to AI-based speech documentation in routine practice. Champions combined high perceived usefulness, ease of use, trust, habit, and enjoyment with strong adoption and perceived benefits. Pragmatic Adopters adopted the system functionally and emphasized supportive conditions. The smallest group, Reluctant Adopters, remained a measured profile despite regular usage. The taxonomy offers a practical basis for developing and tailoring implementation strategies in real care settings: pairing functional reliability with visible benefits, sustaining fast social and organizational support, providing practice opportunities, and addressing concerns directly for hesitant users. When adoption strategies are aligned with user classes’ specific needs and characteristics, AI speech documentation is more likely to translate from technical promise to perceptible relief in documentation and handovers, without assuming that technology is a benefit by default. Class-aware implementation provides a path toward inclusive, evidence-based digital transformation in nursing. Abbreviations AI-ASSISTED SPEECH DOCUMENTATION USER GROUPS Declarations Competing Interests K.S. was employed by voize GmbH, the developer of the mobile AI speech documentation system evaluated in this study, at the time of data collection and manuscript preparation. The study was conducted within the joint research project PYSA. All other authors declare that they have no competing interests. Ethics Declaration This study involved a voluntary and anonymous online survey of adult nursing staff in German senior care facilities. Prior to participation, all respondents received study information and provided electronic informed consent. No patient data was collected. To protect participants’ privacy and reduce identifiability, no information on the specific care facility was collected and only a minimal set of sociodemographic variables was recorded. The questionnaire was reviewed and pretested by experienced nurses, and the study procedure was approved by the employee representative committees of the participating care facilities. The study was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines. Funding This work received funding through the project PYSA (“Pflegedokumentation mit Hybridem Sprachassistenten”), a joint research project funded by the German Federal Ministry of Education and Research (now the Federal Ministry of Research, Technology and Space). Author Contribution Conceptualization: D.F., S.N., C.B.-H., A.R., K.S., N.L. and S.S.-L.; methodology and study design: D.F. and S.N.; investigation and data acquisition, D.F., S.N., C.B.-H., A.R. and K.S.; formal analysis: D.F.; data interpretation: D.F., S.N., and S.S.-L.; writing-original draft preparation: D.F.; writing-review and editing: S.N., L.P., N.R., C.B.-H., A.R., K.S., N.L. and S.S.-L.; visualization: D.F.; supervision: N.L. and S.S.-L. All authors reviewed the manuscript, approved the final submitted version, and agree to be accountable for their own contributions. Acknowledgement We sincerely thank all participants and the care facilities that took part in this study for their valuable feedback, support, and engagement throughout the project. Data Availability The datasets generated and analyzed during the study are not publicly available. The datasets may be available from the corresponding author on reasonable request, subject to institutional, legal, and ethical requirements. References Federal Statistical Office. Number of people working in hospital nursing care up 18% in ten years. (2022). https://www.destatis.de/EN/Press/2022/05/PE22_N026_2313.html McCarthy, B. et al. Electronic nursing documentation interventions to promote or improve patient safety and quality care: A systematic review. J. Nurs. Adm. Manag. 27 (3), 491–501. https://doi.org/10.1111/jonm.12727 (2019). Heislbetz, C. Die Dokumentation in der Altenpflege. Universität Bamberg. (2008). https://fis.uni-bamberg.de/handle/uniba/190 Schenk, E. et al. Time motion analysis of nursing work in ICU, telemetry and medical-surgical units. J. Nurs. Adm. Manag. 25 (8), 640–646. https://doi.org/10.1111/jonm.12502 (2017). Yen, P. Y. et al. Nurses’ time allocation and multitasking of nursing activities: A time motion study. AMIA Annual Symposium Proceedings, 2018, 1137–1146. (2018). Gesner, E., Dykes, P. C., Zhang, L. & Gazarian, P. Documentation burden in nursing and its role in clinician burnout syndrome. Appl. Clin. Inf. 13 (5), 983–990. https://doi.org/10.1055/s-0042-1757157 (2022). Moy, A. J. et al. Measurement of clinical documentation burden among physicians and nurses using electronic health records: A scoping review. J. Am. Med. Inform. Assoc. 28 (5), 998–1008. https://doi.org/10.1093/jamia/ocaa325 (2021). Ferizaj, D. & Neumann, S. Assessing perceptions and experiences of an AI-driven speech assistant for nursing documentation: A qualitative study in German nursing homes. In M. Kurosu & A. Hashizume (Eds.), Human-Computer Interaction (pp. 17–34). Springer Nature Switzerland. (2024). https://doi.org/10.1007/978-3-031-60449-2_2 Joseph, J., Moore, Z. E. H., Patton, D., O’Connor, T. & Nugent, L. E. The impact of implementing speech recognition technology on the accuracy and efficiency of clinical documentation by nurses: A systematic review. J. Clin. Nurs. 29 (13–14), 2125–2137. https://doi.org/10.1111/jocn.15261 (2020). Ehrler, F., Wu, D. T. Y., Ducloux, P. & Blondon, K. A mobile application to support bedside nurse documentation and care: A time and motion study. JAMIA Open. 4 (3), ooab046. https://doi.org/10.1093/jamiaopen/ooab046 (2021). Schwabe, K., Ferizaj, D., Neumann, S., Strube-Lahmann, S. & Lahmann, N. Time Savings Through an AI Speech Assistant for Nursing Documentation: A Pre-Post Time-Motion Study in German Long-Term Care. J. Med. Internet Res. https://doi.org/10.2196/86078 Ammenwerth, E., Mansmann, U., Iller, C. & Eichstädter, R. Factors affecting and affected by user acceptance of computer-based nursing documentation: Results of a two-year study. J. Am. Med. Inform. Assoc. 10 (1), 69–84. https://doi.org/10.1197/jamia.m1118 (2003). Venkatesh, V., Thong, J. Y. L. & Xu, X. Consumer acceptance and use of information technology: Extending the unified theory of acceptance and use of technology. MIS Q. 36 (1), 157–178. https://doi.org/10.2307/41410412 (2012). Lee, A. T., Ramasamy, R. K. & Subbarao, A. Understanding psychosocial barriers to healthcare technology adoption: A review of TAM and UTAUT frameworks. Healthcare 13 (3), 250. https://doi.org/10.3390/healthcare13030250 (2025). Venkatesh, V., Morris, M. G., Davis, G. B. & Davis, F. D. User acceptance of information technology: Toward a unified view. MIS Q. 27 (3), 425–478. https://doi.org/10.2307/30036540 (2003). Hussain, A. et al. The mediating effects of perceived usefulness and perceived ease of use on nurses’ intentions to adopt advanced technology. BMC Nurs. 24 (1), 33. 10.1186/s12912-024-02648-8 (2025). Lee, W. I., Fu, H. P., Mendoza, N. & Liu, T. Y. Determinants impacting user behavior towards emergency use intentions of m-health services in Taiwan. Healthcare 9 (5), 535. https://doi.org/10.3390/healthcare9050535 (2021). Knöchelmann, A. et al. User profiles in digitalized healthcare: Active, potential, and rejecting—A cross-sectional study using latent class analysis. BMC Health Serv. Res. 24 (1), 1083. https://doi.org/10.1186/s12913-024-11523-w (2024). Lanza, S. T. & Rhoades, B. L. Latent class analysis: An alternative perspective on subgroup analysis in prevention and treatment. Prev. Sci. 14 (2), 157–168. https://doi.org/10.1007/s11121-011-0201-1 (2013). Sinha, P., Calfee, C. S. & Delucchi, K. L. Practitioner’s guide to latent class analysis: Methodological considerations and common pitfalls. Crit. Care Med. 49 (1), e63–e79. https://doi.org/10.1097/CCM.0000000000004710 (2021). Linzer, D. A. & Lewis, J. B. poLCA: An R package for polytomous variable latent class analysis. J. Stat. Softw. 42 , 1–29. https://doi.org/10.18637/jss.v042.i10 (2011). Aboueid, S., Meyer, S. B., Wallace, J. & Chaurasia, A. Latent classes associated with the intention to use a symptom checker for self-triage. PLOS ONE . 16 (11), e0259547. https://doi.org/10.1371/journal.pone.0259547 (2021). Staeck, R., Stüble, M. & Drüge, M. Exploring subgroups of acceptance prediction for e-mental health among psychotherapists-in-training: A latent class analysis. Front. Psychiatry . 15 , 1296449. https://doi.org/10.3389/fpsyt.2024.1296449 (2024). Kaihlanen, A. M. et al. Nursing informatics competence profiles and perceptions of health information system usefulness among registered nurses: A latent profile analysis. J. Adv. Nurs. 79 (10), 4022–4033. https://doi.org/10.1111/jan.15718 (2023). Fu, L., Xie, Y., Zhu, Y., Zhang, C. & Ge, Y. Innovative behavior profile and its associated factors among nurses in China: A cross-sectional study based on latent profile analysis. BMJ Open. 14 (6), e084932. https://doi.org/10.1136/bmjopen-2024-084932 (2024). Bundesministerium für Forschung, Technologie und Raumfahrt. PYSA: Pflege entlasten - Pflegedokumentation mit hybridem Sprachassistenten. (2025). https://www.interaktive-technologien.de/projekte/pysa Wurpts, I. C. & Geiser, C. Is adding more indicators to a latent class analysis beneficial or detrimental? Results of a Monte-Carlo study. Front. Psychol. https://doi.org/10.3389/fpsyg.2014.00920 (2014). 5. Liu, H. et al. Artificial intelligence and radiologist burnout. JAMA Netw. Open. 7 (11), e2448714. https://doi.org/10.1001/jamanetworkopen.2024.48714 (2024). Mah, D. K. & Groß, N. Artificial intelligence in higher education: Exploring faculty use, self-efficacy, distinct profiles, and professional development needs. Int. J. Educational Technol. High. Educ. 21 (1), 58. https://doi.org/10.1186/s41239-024-00490-1 (2024). Ferizaj, D., Perotti, L., Dahms, R. & Heimann-Steinert, A. Use of technology in old age: Associations between acceptance, competence, control, interest and social indicators in individuals over 60 years old. Z. für Gerontologie und Geriatrie . https://doi.org/10.1007/s00391-023-02225-9 (2023). Wang, H., Sun, Q., Gu, L., Lai, K. & He, L. Diversity in people’s reluctance to use medical artificial intelligence: Identifying subgroups through latent profile analysis. Front. Artif. Intell. 5 , 1006173. https://doi.org/10.3389/frai.2022.1006173 (2022). Leary, M. et al. Creating a phenotype and taxonomy of nurses engaging in innovative behaviors. Online. J. Issues. Nurs. 30 (1). https://doi.org/10.3912/OJIN.Vol30No01Man02 (2025). Schwabe, K., Ferizaj, D. & Neumann, S. Reducing Nurses' Workload with an AI Speech Assistant for Documentation (2025). In: Proceedings of the First International Conference on AI in Medicine and Healthcare (AiMH' 2025) , pp. 119–122. Abril-Pla, O. et al. PyMC: A modern and comprehensive probabilistic programming framework in Python. PeerJ Comput. Sci. 9 , e1516. https://doi.org/10.7717/peerj-cs.1516 (2023). Kumar, R., Carroll, C., Hartikainen, A. & Martin, O. ArviZ: A unified library for exploratory analysis of Bayesian models in Python. J. Open. Source Softw. 4 (33), 1143. https://doi.org/10.21105/joss.01143 (2019). Additional Declarations Competing interest reported. K.S. was employed by voize GmbH, the developer of the mobile AI speech documentation system evaluated in this study, at the time of data collection and manuscript preparation. The study was conducted within the joint research project PYSA. All other authors declare that they have no competing interests. Supplementary Files AppendixAQuestionnaire.docx AppendixBMeasurementModel.docx AppendixCutaut.docx Cite Share Download PDF Status: Under Revision Version 1 posted Editorial decision: Revision requested 04 May, 2026 Submission checks completed at journal 30 Apr, 2026 First submitted to journal 30 Apr, 2026 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-9557939","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":633970832,"identity":"b5ad6857-2d85-416b-ae01-cd74d7581445","order_by":0,"name":"Drin Ferizaj","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA/0lEQVRIie3Pu2rDMBSA4SME8mLs1Z3yCsdoKAE3fRUFg6eWFgqZMmhSlkJXD3mOzA4H3CX4BdrBXTJr9Jb6shVcdQxU/6Ab+hAC8PmusYppgAxA9Gu0UAHw4Vi4SDHdScuRcBcZhmJa83DcOkj0QaYFlT1FMbUv2fZzcbuLjy1sslly06x3CKpYGqhRPtTndE+cIzTFLMETMwnrCAXTKB81sZJzkTBDvxNQFxQ8sHKp6b4nQcfMxUUqFCJEyTSth1eAmWr+LwNRKkcRhs/pa015T2SimnyWRKfgnFi1wsXb+wG7Ld2V8fHL2s1qloypaRL488QZb/940efz+f5Z34W8TZKtAadIAAAAAElFTkSuQmCC","orcid":"","institution":"Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität","correspondingAuthor":true,"prefix":"","firstName":"Drin","middleName":"","lastName":"Ferizaj","suffix":""},{"id":633970833,"identity":"e7a091d8-1479-4ca4-a77b-5084ba871d47","order_by":1,"name":"Susann Neumann","email":"","orcid":"","institution":"Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität","correspondingAuthor":false,"prefix":"","firstName":"Susann","middleName":"","lastName":"Neumann","suffix":""},{"id":633970834,"identity":"d8ab570b-8a61-4c20-afa0-3f15a6f473a8","order_by":2,"name":"Corina Burkhardt-Herdtle","email":"","orcid":"","institution":"Kleeblatt Pflegeheime gGmbH","correspondingAuthor":false,"prefix":"","firstName":"Corina","middleName":"","lastName":"Burkhardt-Herdtle","suffix":""},{"id":633970835,"identity":"433bc9c7-0df2-47d3-b18a-870870960722","order_by":3,"name":"Alexander Rau","email":"","orcid":"","institution":"Johannesstift Diakonie Pflege \u0026 Wohnen Berlin Brandenburg gGmbH","correspondingAuthor":false,"prefix":"","firstName":"Alexander","middleName":"","lastName":"Rau","suffix":""},{"id":633970837,"identity":"4e9bd1a3-a4ff-4fad-b06e-4b0eaed0e86c","order_by":4,"name":"Luis Perotti","email":"","orcid":"","institution":"Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität","correspondingAuthor":false,"prefix":"","firstName":"Luis","middleName":"","lastName":"Perotti","suffix":""},{"id":633970839,"identity":"e22bdc26-0a49-4020-9e11-cac4d714a7b9","order_by":5,"name":"Nikolai Ratajczak","email":"","orcid":"","institution":"Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Berlin, Germany","correspondingAuthor":false,"prefix":"","firstName":"Nikolai","middleName":"","lastName":"Ratajczak","suffix":""},{"id":633970841,"identity":"f70d47df-be58-471b-9afe-ddecec5c8dd8","order_by":6,"name":"Katja Schwabe","email":"","orcid":"","institution":"voize GmbH","correspondingAuthor":false,"prefix":"","firstName":"Katja","middleName":"","lastName":"Schwabe","suffix":""},{"id":633970844,"identity":"289239e1-1e5b-4d40-bbce-665e6f4a64be","order_by":7,"name":"Nils Lahmann","email":"","orcid":"","institution":"Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität","correspondingAuthor":false,"prefix":"","firstName":"Nils","middleName":"","lastName":"Lahmann","suffix":""},{"id":633970847,"identity":"7631ee9e-fdf4-4b7b-ad9e-6b5dcc881c57","order_by":8,"name":"Sandra Strube-Lahmann","email":"","orcid":"","institution":"Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität","correspondingAuthor":false,"prefix":"","firstName":"Sandra","middleName":"","lastName":"Strube-Lahmann","suffix":""}],"badges":[],"createdAt":"2026-04-28 19:53:28","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9557939/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9557939/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108570559,"identity":"5dddd4b2-cd05-4a59-9ff5-1a29bc085cc7","added_by":"auto","created_at":"2026-05-06 06:10:23","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":198904,"visible":true,"origin":"","legend":"\u003cp\u003eMean item response profiles across three latent classes identified via UTAUT2-based latent class analysis. Panel A displays construct-level averages on a radar plot for seven UTAUT2-related constructs. Panel B shows item-level response means grouped by construct: performance expectancy (PE, items 1–3), effort expectancy (EE, items 4–7), facilitating conditions (FC, items 8–9), habit (HB, items 10–11), hedonic motivation (HM), social influence (SI), and trust (TR). Dashed vertical lines mark construct boundaries. Class 1 (Champions, 27.3%) shows uniformly high technology acceptance across all constructs. Class 2 (Pragmatic Adopters, 52.6%), the largest class, exhibits moderate-to-high endorsement. Class 3 (Reluctant Adopters, 20.1%) scores lowest on nearly all indicators, with Habit and Trust showing the smallest inter-class gaps.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-9557939/v1/3eb4073d71334d459874cc99.png"},{"id":108805035,"identity":"85207aab-c810-422d-90b4-041b15aa7f20","added_by":"auto","created_at":"2026-05-08 15:24:35","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":752672,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9557939/v1/84bff9b8-5cc5-4522-a819-5cac3518125c.pdf"},{"id":108570581,"identity":"c66a0681-6206-4756-980a-56fba7b2115a","added_by":"auto","created_at":"2026-05-06 06:10:27","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":25547,"visible":true,"origin":"","legend":"","description":"","filename":"AppendixAQuestionnaire.docx","url":"https://assets-eu.researchsquare.com/files/rs-9557939/v1/c529fc97bf68c3e1547a212c.docx"},{"id":108570589,"identity":"c70eb642-73b4-42a6-92f2-45441cb08f8e","added_by":"auto","created_at":"2026-05-06 06:10:33","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":20843,"visible":true,"origin":"","legend":"","description":"","filename":"AppendixBMeasurementModel.docx","url":"https://assets-eu.researchsquare.com/files/rs-9557939/v1/e2b23518b446d7f3ab92fcfb.docx"},{"id":108570634,"identity":"e3052aec-b25e-45e2-992e-e23e5a142a45","added_by":"auto","created_at":"2026-05-06 06:10:38","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":26613,"visible":true,"origin":"","legend":"","description":"","filename":"AppendixCutaut.docx","url":"https://assets-eu.researchsquare.com/files/rs-9557939/v1/16ea7c1f826a2f7db9f7f812.docx"}],"financialInterests":"Competing interest reported. K.S. was employed by voize GmbH, the developer of the mobile AI speech documentation system evaluated in this study, at the time of data collection and manuscript preparation. The study was conducted within the joint research project PYSA. All other authors declare that they have no competing interests.","formattedTitle":"AI-Assisted Speech Documentation in Nursing: Identifying User Groups via Latent Class Analysis","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eNurses working in care homes and home care services constitute one of the largest groups of healthcare professionals in Germany, with nearly one million employed in this sector.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e Their central role in patient care makes them primary users of health record systems and places them at the forefront of clinical and care documentation. While such documentation is critical for ensuring high-quality care,\u003csup\u003e2,3\u003c/sup\u003e documentation can consume up to 30% of a shift across nursing settings.\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e,\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e This considerable documentation burden, combined with fragmented work processes, frequent interruptions, and the urgent need to comply with regulatory standards, has been associated with lower job satisfaction and heightened job-related stress.\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eTo reduce workload and improve both care quality and staff well-being, care facilities are increasingly considering digital tools. Their impact depends on whether nurses can and will use them in everyday practice. One promising option for nursing documentation is automatic speech recognition (ASR), which converts spoken language into structured entries or notes, enabling hands-free operation, faster capture, and greater mobility in daily work.\u003csup\u003e\u003cspan additionalcitationids=\"CR9 CR10\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e Evidence from nursing settings suggests that ASR can improve efficiency, accuracy, and user satisfaction.\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e Related mobile bedside documentation research also points to workflow and time-saving benefits.\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e Consistent with this, a recent pre-post time-motion study in German long-term care found that implementing a speech AI documentation system was associated with substantially less documentation time, fewer interruptions, and higher satisfaction with the documentation system.\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e Despite these benefits, low acceptance may still limit adoption, as shown in early computer-based documentation\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e and more recent mobile ASR documentation research.\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e To explain why some healthcare professionals adopt a given system while others hesitate, the Unified Theory of Acceptance and Use of Technology and its consumer extension (UTAUT2) offers a validated framework. It aims to explain technology adoption through seven core constructs: Performance Expectancy (perceived job benefits), Effort Expectancy (ease of learning and use), Social Influence (normative pressure from colleagues and supervisors), and Facilitating Conditions (organizational and technical support). Consumer-focused extensions add Hedonic Motivation (enjoyment of use), Habit (automaticity of adoption), and Price-Value (cost-benefit assessment).\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e In healthcare, these constructs are repeatedly associated with users\u0026rsquo; behavioral intention and usage,\u003csup\u003e14,15\u003c/sup\u003e with performance expectancy and effort expectancy often emerging as influential predictors.\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e,\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e,\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e Extensions of these models in healthcare commonly add trust and satisfaction, and in some settings privacy or security concerns.\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eIn general, these frameworks yield estimates at the population mean and do not reveal whether distinct subgroups of nurses hold qualitatively different configurations of these perceptions. Therefore, most acceptance research has been \u0026ldquo;variable-centered,\u0026rdquo; estimating paths among constructs and outcomes at the level of the average user.\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e,\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e Yet, in real care environments, nurses can hold qualitatively different attitudes and experiences that may cluster into distinct groups.\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e When the goal is to derive such classes from empirical data, latent class analysis (LCA) is a feasible approach. LCA is a person-centered statistical method that uncovers hidden latent subgroups within a population based on response patterns across multiple observed variables,\u003csup\u003e19\u0026ndash;21\u003c/sup\u003e shifting attention from average effects to co-occurring patterns of attitudes within individuals. Specifically, LCA is a probabilistic finite-mixture model that attributes heterogeneity to a discrete set of latent classes, each defined by characteristic item-response probabilities. In this context, class membership rather than direct item-to-item associations accounts for observed correlations among indicators.\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e,\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eEvidence from LCA studies highlights the usefulness of segmentation, although most were conducted in non-nursing populations and did not specifically assess ASR. Among university students, an LCA of attitudes toward AI-enabled symptom checkers identified five profiles ranging from \u0026ldquo;tech acceptors\u0026rdquo; to \u0026ldquo;tech rejectors,\u0026rdquo; with class membership strongly associated with intention to use the tool.\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e In mental health training, psychotherapists-in-training assessed with UTAUT-derived indicators formed two acceptance classes that differed primarily on performance expectancy and effort expectancy, with little differentiation by age or gender.\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eLatent profile analysis (LPA), a method for identifying subgroups based on continuous variables, has been applied to characterize nursing competencies and related technology perceptions. In a nationwide sample of 3,610 Finnish registered nurses, three informatics-competence profiles differed in perceived usefulness of health information systems, with higher-competence profiles reporting greater usefulness.\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e Similarly, an LPA of nurses in traditional Chinese medicine hospitals identified low, moderate, and high innovative-behavior profiles, with profile membership associated with structural empowerment, adversity quotient, training, competency, and several work-related characteristics.\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eThese studies collectively strengthen the case for person-centered modeling but also delineate a clear gap for long-term care nursing and ASR technologies. First, most ASR acceptance work, when framed by UTAUT, has remained variable-centered, focusing on average paths rather than qualitatively distinct user classes. \u003csup\u003e13\u003c/sup\u003e Second, the person-centered LCA/latent profile analysis literature has largely profiled students, the public, or constructs adjacent to acceptance such as competence or innovation behavior, rather than segmenting practicing nurses with actual exposure to a specific AI documentation tool in routine care. The only comparable LCA study from Germany targeted the general public rather than practicing nurses and assessed broader digital health services rather than a specific AI documentation workflow, leaving the question of user classes among long-term care professionals empirically open.\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eOur study addresses this gap by applying LCA to nurses in German long-term care facilities who have hands-on experience with an AI speech assistant for documentation. We combine a theory-driven set of UTAUT2 indicators\u0026mdash;capturing perceived usefulness, effort, facilitating conditions, habit, enjoyment, social influence, and trust\u0026mdash;with a data-driven segmentation that recognizes qualitatively different user mindsets rather than assuming homogeneity. Competing class solutions are estimated across multiple indicator specifications, with several information criteria and entropy used for class selection, and classes are validated against practical outcomes including satisfaction, frequency of use, perceived time savings, perceived improvements, and recommendation.\u003c/p\u003e \u003cp\u003eAccordingly, we address three central research questions:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003e1. How many distinct user classes of AI speech-assistant adoption can be identified among nursing staff in senior care based on UTAUT2-related perceptions?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003e2. What are the defining response-pattern characteristics of each class across the UTAUT2 constructs?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003e3. How do classes differ on auxiliary acceptance and usage outcomes, including satisfaction, frequency of use, perceived time savings, recommendation, and intention to continue use?\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e"},{"header":"METHOD","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy Design and Participants\u003c/h2\u003e \u003cp\u003eThis study employed a cross-sectional survey design to examine user acceptance patterns of an ASR documentation system (\u0026ldquo;voize,\u0026rdquo; provided by voize GmbH, Berlin, Germany) among nursing staff in German senior care facilities between August 2024 and January 2025. The research was conducted as part of the PYSA project (\u0026ldquo;Pflegedokumentation mit hybridem Sprachassistenten\u0026rdquo;),\u003csup\u003e26\u003c/sup\u003e funded by the German Federal Ministry of Education and Research (now called Federal Ministry of Research, Technology and Space).\u003c/p\u003e \u003cp\u003eData were collected via a voluntary and anonymous online survey administered through the REDCap platform. Recruitment was conducted exclusively through the voize app, ensuring that all participants had direct experience with the speech documentation assistant. In addition to purposive sampling, a convenience component was used to maximize participation across varying care contexts. The survey was open to staff at four cooperating care facilities in Germany.\u003c/p\u003e \u003cp\u003eEligible participants were adults aged 18 years or older who had used the voize documentation assistant in their daily work within the previous six months and could understand written German instructions. Before completing the questionnaire, all participants received study information and provided written informed consent. To protect privacy and reduce identifiability, the survey did not collect information about respondents\u0026rsquo; specific care facilities and included only minimal sociodemographic data. Participation was voluntary, anonymous, noninterventional, and not monetarily incentivized. All respondents who completed at least 10% of the questionnaire were included in the final analysis. No formal ethics vote was obtained because the study involved only adult professional users and only collected no patient-related data. The Ethics Committee of Charit\u0026eacute; - Universit\u0026auml;tsmedizin Berlin would generally be the responsible institutional ethics body for the underlying research project. All procedures followed established ethical principles, including the Declaration of Helsinki and Good Clinical Practice guidelines, and ensured informed consent, voluntary participation, anonymity, and the right to discontinue participation at any time.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eMeasures and Survey Instrument\u003c/h3\u003e\n\u003cp\u003eThe questionnaire was developed collaboratively by two nursing experts and three of the authors. An initial set of items was drafted based on previous research.\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e Five reviewers evaluated each item on a binary scale (0\u0026thinsp;=\u0026thinsp;reject, 1\u0026thinsp;=\u0026thinsp;keep). Items scoring below three were excluded, those scoring five were retained, and those scoring four were discussed until consensus was reached. The final version was pretested by two nurses under realistic conditions, and no major modifications were made. The full translated English and German versions as well as an overview of included constructs, individual items, and response formats are provided in Appendix A.\u003c/p\u003e \u003cp\u003eThe instrument operationalized UTAUT2 constructs alongside demographic variables and auxiliary measures. Multi-item UTAUT2 constructs were Performance Expectancy (three items; e.g., \u0026ldquo;Using voize reduces my documentation workload\u0026rdquo;), Effort Expectancy (four items; e.g., \u0026ldquo;voize is easy to learn\u0026rdquo;), Facilitating Conditions (two items; e.g., \u0026ldquo;I can get quick help if needed\u0026rdquo;), and Habit (two items; e.g., \u0026ldquo;I use voize almost automatically during my shift\u0026rdquo;). Single-item indicators captured Hedonic Motivation (\u0026ldquo;Using voize is enjoyable\u0026rdquo;), Social Influence (\u0026ldquo;My team supports me using voize\u0026rdquo;), and Trust in Data Security (\u0026ldquo;Data security is ensured\u0026rdquo;). All UTAUT2 items were rated on a five-point Likert scale (1\u0026thinsp;=\u0026thinsp;strongly disagree to 5\u0026thinsp;=\u0026thinsp;strongly agree). Price Value was not included in this study as the costs for the software were covered by the employer. Higher construct scores indicate, respectively, stronger perceived benefit, lower perceived effort, greater organizational support, more ingrained use, greater enjoyment, stronger social influence, and higher trust.\u003c/p\u003e \u003cp\u003eThe following self-reported auxiliary outcomes were collected: satisfaction with the documentation system (1\u0026ndash;5), usage frequency (1\u0026ndash;5), perceived time saved per shift (rated on a five-point scale: 1\u0026thinsp;=\u0026thinsp;no time saved, 2\u0026thinsp;=\u0026thinsp;5\u0026ndash;10 min, 3\u0026thinsp;=\u0026thinsp;10\u0026ndash;20 min, 4\u0026thinsp;=\u0026thinsp;20\u0026ndash;30 min, 5\u0026thinsp;=\u0026thinsp;30\u0026ndash;45 min), and likelihood to recommend the documentation system (0\u0026ndash;10; higher values indicating higher agreement). Initial reaction was rated on a four-point scale (1\u0026thinsp;=\u0026thinsp;not happy at all; 4\u0026thinsp;=\u0026thinsp;very happy about the implementation). Technology affinity was rated on a seven-point scale (1\u0026ndash;7). Age was recorded in six ordinal categories (1\u0026thinsp;=\u0026thinsp;under 21 years, 2\u0026thinsp;=\u0026thinsp;21\u0026ndash;30, 3\u0026thinsp;=\u0026thinsp;31\u0026ndash;40, 4\u0026thinsp;=\u0026thinsp;41\u0026ndash;50, 5\u0026thinsp;=\u0026thinsp;51\u0026ndash;60, 6\u0026thinsp;=\u0026thinsp;over 60 years). Three binary items (0\u0026thinsp;=\u0026thinsp;no, 1\u0026thinsp;=\u0026thinsp;yes) assessed perceived improvements in documentation quality, handover quality, and intention to continue using the system. Internal consistency and validity were evaluated via Cronbach\u0026rsquo;s α, composite reliability, average variance extracted, and confirmatory factor analysis. Reliability was good and overall model fit was acceptable (see Appendix B).\u003c/p\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eStatistical Analysis\u003c/h2\u003e \u003cdiv id=\"Sec6\" class=\"Section3\"\u003e \u003ch2\u003eLatent Class Analysis Specifications\u003c/h2\u003e \u003cp\u003eThe goal of the LCA was to identify distinct acceptance classes from observed responses. We estimated a series of latent class models with K \u0026isin; {1, 2, 3, 4, 5} classes under three complementary indicator specifications: (1) the full-item specification retained all 14 original indicators; (2) the statistically sparse specification reduced redundancy by removing one variable from pairs with absolute correlations exceeding r \u0026gt; .70 and collapsing rare response categories below 10%, yielding 12 indicators; and (3) the theory-driven UTAUT2 specification used multi-item composites for Performance Expectancy, Effort Expectancy, Facilitating Conditions, and Habit, and single-item indicators for Hedonic Motivation, Social Influence, and Trust. Analyses were conducted using the poLCA package in R,\u003csup\u003e21\u003c/sup\u003e with models estimated via maximum likelihood with 250 random starts to avoid local maxima.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e\n\u003ch3\u003eModel Selection\u003c/h3\u003e\n\u003cp\u003eModel selection was guided by multiple fit indices: Bayesian Information Criterion (BIC), Akaike Information Criterion (AIC), Sample-size Adjusted BIC (SABIC), BIC-approximated Bayes Factor comparing adjacent models (BF; values\u0026thinsp;\u0026gt;\u0026thinsp;10 indicating strong evidence for adding a class), and Entropy (values closer to 1 indicating better class separation). Primary consideration was given to BIC (lower\u0026thinsp;=\u0026thinsp;better), Entropy, and the Bayes Factor comparing K vs. K\u0026thinsp;\u0026minus;\u0026thinsp;1 class solutions.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eMonte Carlo Simulations\u003c/h2\u003e \u003cp\u003eLatent class analysis typically benefits from large samples (N\u0026thinsp;\u0026gt;\u0026thinsp;500).\u003csup\u003e20\u003c/sup\u003e To verify our identified solution\u0026rsquo;s robustness at \u003cem\u003eN\u003c/em\u003e\u0026thinsp;=\u0026thinsp;134, we conducted Monte Carlo parameter-recovery simulations. Parameters were extracted from the retained three-class LCA models and used to generate synthetic datasets: 1,500 replications at N\u0026thinsp;=\u0026thinsp;100, 134, 200, 300, 500, and 1,000 for both the UTAUT2 and full-indicator models. Each dataset tested 1 to 5 classes with 250 random starts each, with BIC as the primary criterion for class extraction.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eClass Validation With Auxiliary Variables\u003c/h3\u003e\n\u003cp\u003eMissing data for auxiliary outcomes ranged from 4.5% to 20.9% (overall 15.1%; 77.6% complete cases), with moderate associations between missingness and UTAUT2 indicators (correlations ranging from \u0026minus;\u0026thinsp;.30 to .33). We created m\u0026thinsp;=\u0026thinsp;15 multiply-imputed datasets using Random Forest Predictive Mean Matching with 10 iterations, K\u0026thinsp;=\u0026thinsp;5 nearest neighbors, and 250 trees.\u003c/p\u003e \u003cp\u003eEight outcomes were modeled jointly in a single Bayesian multivariate normal regression in Python 3.12 using PyMC 5.28.2\u003csup\u003e34\u003c/sup\u003e and ArviZ 1.0.0\u003csup\u003e35\u003c/sup\u003e: usage frequency, perceived time saved per shift, satisfaction, age group, likelihood to recommend, technology affinity, voize usage duration, and a three-item formative index indicating overall perceived benefits (the unweighted sum of three binary outcomes: intention to continue use, perceived documentation quality improvement, and perceived handover improvement; range 0\u0026ndash;3).\u003c/p\u003e \u003cp\u003eThe sole predictor across all outcomes was latent class membership, with Pragmatic Adopters serving as the reference category. Class membership was operationalized via soft class weights, propagating posterior membership probabilities from the LCA directly into the likelihood. Residual covariance used a Lewandowski-Kurowicka-Joe (LKJ) prior (η\u0026thinsp;=\u0026thinsp;3), and outcome-specific residual standard deviations received half-Student-t priors (ν\u0026thinsp;=\u0026thinsp;4, σ\u0026thinsp;=\u0026thinsp;1.0). Contrast coefficients were assigned hierarchical shrinkage priors (half-Student-t global scale τ\u0026thinsp;~\u0026thinsp;ν\u0026thinsp;=\u0026thinsp;3, σ\u0026thinsp;=\u0026thinsp;0.35). Models were estimated within each imputed dataset using the JAX NumPyro NUTS backend (4 chains, 2,000 post-warmup draws per chain, 2,500 warmup steps, target acceptance\u0026thinsp;=\u0026thinsp;0.99). Posteriors were stacked across imputations, yielding 15 \u0026times; 4 \u0026times; 2,000\u0026thinsp;=\u0026thinsp;120,000 total draws. Convergence was assessed via R̂, Bayesian Fraction of Missing Information (BFMI), and effective sample size (ESS). For each class contrast we report stacked posterior means, 95% highest-density intervals (HDIs), and the direction of the posterior as indicators of directional robustness.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eSample Description\u003c/h2\u003e \u003cp\u003eThe study sample comprised 134 participants, predominantly qualified nurses (39.4%) and nursing assistants (35.1%). The remaining roles included nursing service managers (8.5%), caregivers (8.2%), ward managers (4.4%), and trainees (4.4%). Regarding age, the largest cohort fell within the 41\u0026ndash;50 year range (26.9%), followed by those aged 51\u0026ndash;60 (23.1%) and 31\u0026ndash;40 (20.9%); fewer participants were aged 21\u0026ndash;30 (13.4%), over 60 (12.7%), or under 21 (3.0%). Usage duration ranged from 0\u0026ndash;4 weeks (11.9%), 5\u0026ndash;8 weeks (32.1%), 9\u0026ndash;12 weeks (29.9%), to over 12 weeks (26.1%).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eDescriptive Results\u003c/h2\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e presents descriptive statistics for all study variables before and after multiple imputation. Overall, participants reported positive initial reactions about the implementation of the speech assistant (M\u0026thinsp;=\u0026thinsp;3.69, SD\u0026thinsp;=\u0026thinsp;1.06 on a 4-point scale) and strong technology affinity (M\u0026thinsp;=\u0026thinsp;5.52, SD\u0026thinsp;=\u0026thinsp;1.42 on a 7-point scale); multiple imputation produced negligible changes across all variables (all |Δ| \u0026lt; 1.6%), supporting the adequacy of the imputation procedure.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cem\u003eDescriptive Statistics for All Study Variables: Complete and Imputed Data\u003c/em\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"11\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eM (SD)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eMedian\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003eMin\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eMax\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eM (SD)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003eMedian\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c11\"\u003e \u003cp\u003eΔ\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eComplete Data\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eImputed Data\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSociodemographic Variables\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge Group (1\u0026ndash;6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.92 (1.33)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.92 (1.33)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e0.0%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTechnology Affinity (1\u0026ndash;7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5.52 (1.42)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e6.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e5.52 (1.42)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e6.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e0.0%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eUTAUT2 Variables\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003ePerformance Expectancy (α\u0026thinsp;=\u0026thinsp;.86)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWork Relief (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.82 (1.08)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.81 (1.08)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;0.2%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDocuments More (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.62 (1.06)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.58 (1.08)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;1.2%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMore Time for Residents (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.88 (1.06)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.91 (1.03)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e+\u0026thinsp;0.8%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eEffort Expectancy (α\u0026thinsp;=\u0026thinsp;.83)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEasy to Learn (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.93 (0.94)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.95 (0.92)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e+\u0026thinsp;0.4%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEase of Use (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.97 (0.90)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.97 (0.88)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;0.1%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eReliable Speech Recognition (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.58 (1.01)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.54 (1.01)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;1.1%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSystem Reliability (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.80 (1.06)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.76 (1.07)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;1.0%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eFacilitating Conditions (α\u0026thinsp;=\u0026thinsp;.80)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQuick Help Available (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.02 (0.93)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e4.01 (0.93)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;0.4%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEnough Time to Learn (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.84 (1.01)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.83 (0.99)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;0.3%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eHabit (α\u0026thinsp;=\u0026thinsp;.73)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHabit Formation (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.03 (1.03)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e4.04 (1.01)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e+\u0026thinsp;0.2%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIndependent Use (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.15 (0.96)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e4.14 (0.97)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;0.3%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAdditional UTAUT2 Variables\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEnjoyment (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.84 (1.09)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.81 (1.10)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;0.8%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTeam Supports voize (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.62 (0.89)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.59 (0.88)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;0.8%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTrust in Data Security (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.08 (0.90)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e4.07 (0.90)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;0.3%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAuxiliary Variables\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInitial Reaction (1\u0026ndash;4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.69 (1.06)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.69 (1.06)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;0.2%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUsage Frequency (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.19 (1.20)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e5.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e4.18 (1.22)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e5.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;0.3%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUsage Duration (1\u0026ndash;4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.70 (0.99)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e2.72 (1.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e+\u0026thinsp;0.5%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTime Saved per Shift (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.03 (1.37)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.07 (1.30)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e+\u0026thinsp;1.1%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSatisfaction (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.01 (0.95)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e3.98 (0.92)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e4.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;0.7%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRecommendation (0\u0026ndash;10)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7.89 (2.68)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e9.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e7.77 (2.82)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e9.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;1.4%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eBinary Auxiliary Variables (% Yes)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQuality Improved (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e84.3%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e83.5%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u0026minus;0.8%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHandover Improved (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e64.2%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e65.8%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e+\u0026thinsp;1.6%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIntention to Continue (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e90.9%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e91.5%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e+\u0026thinsp;0.6%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"11\"\u003e\u003cem\u003eNote.\u003c/em\u003e Δ\u0026thinsp;=\u0026thinsp;percentage difference between original and pooled, imputed means. Usage duration categories: 1\u0026thinsp;=\u0026thinsp;0\u0026ndash;4 weeks, 2\u0026thinsp;=\u0026thinsp;5\u0026ndash;8 weeks, 3\u0026thinsp;=\u0026thinsp;9\u0026ndash;12 weeks, 4\u0026thinsp;=\u0026thinsp;over 12 weeks. All UTAUT2 items were rated on a 5-point Likert scale (1\u0026thinsp;=\u0026thinsp;strongly disagree, 5\u0026thinsp;=\u0026thinsp;strongly agree). Dashes indicate that median/range are not applicable.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eParticipants showed generally positive responses, with high recommendation scores (M\u0026thinsp;=\u0026thinsp;7.89, SD\u0026thinsp;=\u0026thinsp;2.68) and strong intentions to continue use (90.9%). Most reported quality improvements (84.3%) and better handovers (64.2%). The majority reported saving time per shift: 22.7% saved 5\u0026ndash;10 minutes, 19.3% saved 10\u0026ndash;20 minutes, 22.7% saved 20\u0026ndash;30 minutes, 18.5% saved 30\u0026ndash;45 minutes, and 16.8% reported saving no time. Participants demonstrated regular usage patterns (usage frequency M\u0026thinsp;=\u0026thinsp;4.19, SD\u0026thinsp;=\u0026thinsp;1.20 on a 5-point scale).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eLatent Class Analysis\u003c/h2\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003eModel Selection\u003c/h2\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e presents model fit indices for the UTAUT2 specification across all class solutions; full results for the sparse and full-item specifications are reported in Appendix C. The three-class solution was consistently supported as optimal across all three indicator specifications. The three-class UTAUT2 solution yielded the lowest BIC (BIC\u0026thinsp;=\u0026thinsp;2,114.55), with the Bayes Factor providing strong evidence for three over two classes (BF\u0026thinsp;=\u0026thinsp;1.38 \u0026times; 10\u0026sup1;\u0026sup2;) and overwhelming evidence against four classes (BF\u0026thinsp;=\u0026thinsp;4.92 \u0026times; 10⁻\u0026sup2;\u0026sup1;).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cem\u003eModel Fit Criteria Across Class Solutions: UTAUT2 Specification\u003c/em\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eClasses\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLL\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAIC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eBIC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSABIC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eBF\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUTAUT2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;1,131.13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2,316.25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2,394.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e2,309.09\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;950.54\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2,011.08\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2,170.46\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1,996.48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e4.46e\u0026thinsp;+\u0026thinsp;48\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;854.02\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1,874.03\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2,114.55\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1,852.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e1.38e\u0026thinsp;+\u0026thinsp;12\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;832.21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1,886.42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2,208.08\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1,856.96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e4.92e\u0026thinsp;\u0026minus;\u0026thinsp;21\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;815.30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1,908.59\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2,311.39\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1,871.70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e3.68e\u0026thinsp;\u0026minus;\u0026thinsp;23\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"7\"\u003e\u003cem\u003eNote.\u003c/em\u003e LL\u0026thinsp;=\u0026thinsp;log-likelihood. AIC\u0026thinsp;=\u0026thinsp;Akaike Information Criterion (lower\u0026thinsp;=\u0026thinsp;better). BIC\u0026thinsp;=\u0026thinsp;Bayesian Information Criterion (lower\u0026thinsp;=\u0026thinsp;better); primary selection criterion. SABIC\u0026thinsp;=\u0026thinsp;sample-size adjusted BIC. BF\u0026thinsp;=\u0026thinsp;BIC-approximated Bayes Factor comparing K vs. K\u0026thinsp;\u0026minus;\u0026thinsp;1 class solutions; values\u0026thinsp;\u0026gt;\u0026thinsp;10 indicate strong evidence for the larger solution, values\u0026thinsp;\u0026lt;\u0026thinsp;0.1 indicate evidence against. * Retained solution. Full results for the full-item and sparse indicator specifications are reported in Appendix C.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eDiagnostic criteria further supported the three-class solution (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Entropy values for the three-class UTAUT2 solution were excellent (.982), indicating near-perfect class separation (with values approaching 1.0 reflecting minimal classification uncertainty). Average latent class posterior probabilities (ALCPP) exceeded .97 in all cases, meaning that on average, individuals were assigned to their most probable class with over 97% certainty. The smallest class in the retained three-class solution contained 27 participants (20.1%), exceeding recommended minimum class sizes. Consistent results across all three indicator specifications are reported in Appendix C.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cem\u003eDiagnostic Criteria Across Class Solutions: UTAUT2 Specification\u003c/em\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eClasses\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSmallest n\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSmallest %\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eEntropy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eALCPP\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUTAUT2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e134\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e100.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e43\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e32.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.951\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.986\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e20.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.982\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.978\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e11.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.952\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.972\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e9.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.956\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.971\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"6\"\u003e\u003cem\u003eNote.\u003c/em\u003e ALCPP\u0026thinsp;=\u0026thinsp;Average Latent Class Posterior Probability; values closer to 1.0 indicate that individuals are assigned to their modal class with high certainty (values \u0026ge; .90 are considered acceptable). Entropy ranges from 0 to 1; values \u0026ge; .80 indicate good class separation, values \u0026ge; .90 indicate excellent separation. * Retained solution. Comparable results for the full-item and sparse specifications are reported in Appendix C.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eMonte Carlo Simulations\u003c/h2\u003e \u003cp\u003eMonte Carlo simulations confirmed the robustness of the three-class solution. Recovery rates exceeded .90 in all conditions for both the UTAUT2 and full-indicator specifications, with the sole exception of the UTAUT2 model at N\u0026thinsp;=\u0026thinsp;100 (recovery\u0026thinsp;=\u0026thinsp;83%; full-indicator model at N\u0026thinsp;=\u0026thinsp;100: 96%). From N\u0026thinsp;=\u0026thinsp;134 onward, both models achieved 100% recovery, near-perfect convergence, and entropy \u0026gt; .90. The optimal three-class solution showed excellent class separation across both specifications (UTAUT2 entropy = .962). These results indicate that the three-class structure is a robust feature of the data and not a statistical artifact of the sample size.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eClass Profiles\u003c/h2\u003e \u003cp\u003eThe three-class solution was consistent across indicator specifications, with highly similar class proportions and within-class response profiles. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e illustrates the mean item responses by class for the UTAUT2 specification.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eChampions (27.3%)\u003c/h2\u003e \u003cp\u003eThese users embrace voice documentation wholeheartedly. Champions demonstrate exceptional trust in data security (M\u0026thinsp;=\u0026thinsp;4.98 out of 5), derive strong pleasure from the system (Hedonic Motivation: 4.91), and perceive substantial performance benefits (Work Relief: 4.87). Their relationship with the technology appears to transcend mere utility: they have thoroughly integrated voice documentation into their daily routines (Habit: 4.85) and experience pronounced workload reduction. Even their most modestly rated constructs\u0026mdash;perceived team support (4.72) and system reliability (4.71)\u0026mdash;remain strongly positive, suggesting comprehensive satisfaction across all facets of the system.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003ePragmatic Adopters (52.6%)\u003c/h2\u003e \u003cp\u003eRepresenting the majority, Pragmatic Adopters display practical acceptance without extraordinary enthusiasm. Their profile emphasizes facilitating conditions (M\u0026thinsp;=\u0026thinsp;4.76 out of 5) and trust (3.98), with effort expectancy moderately positive (Ease of Use: 3.91). Performance benefits (3.73) and enjoyment (3.73) receive overall positive but measured ratings. These users appear to view voice documentation as a functional tool that adequately meets professional needs - valuing organizational support and (technical) reliability without the strong emotional engagement characteristic of Champions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eReluctant Adopters (20.1%)\u003c/h2\u003e \u003cp\u003eAmong the three classes, Reluctant Adopters show the most restrained engagement with the mobile voice documentation. Compared with the other groups, they score lowest on hedonic motivation (M\u0026thinsp;=\u0026thinsp;2.50 out of 5), perceived ease of use (2.63), and performance expectancy (Work Relief: M\u0026thinsp;=\u0026thinsp;2.66), all of which fall below the scale midpoint. Habit and trust, their relatively highest-rated constructs, nonetheless remain substantially below the levels of the other groups (Habit: M\u0026thinsp;=\u0026thinsp;3.15; Trust: M\u0026thinsp;=\u0026thinsp;3.13). Furthermore, this indicates that these users do engage with the system on a regular basis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eClass Validation: Bayesian Multivariate Modelling across Auxiliary Outcomes\u003c/h2\u003e \u003cdiv id=\"Sec21\" class=\"Section3\"\u003e \u003ch2\u003eConvergence and Sampling Quality\u003c/h2\u003e \u003cp\u003eSampling converged cleanly across all 15 imputed datasets. Zero divergent transitions were observed in every imputation, and maximum R̂ did not exceed 1.002 in any run. BFMI ranged from 0.85 to 0.89 across imputations, all within acceptable bounds. Minimum bulk effective sample sizes exceeded 4,200 per imputation, well above the recommended threshold of 400. Stacking the 15 \u0026times; 4 \u0026times; 2,000\u0026thinsp;=\u0026thinsp;120,000 posterior draws yielded a well-characterized posterior suitable for robust inference.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003ePosterior Class Means\u003c/h2\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e presents the stacked posterior means and 95% HDIs for each class across all eight outcomes. The ordering of classes on substantive outcomes was consistent and theoretically coherent: Champions had the highest posterior means across implementation-relevant outcomes, and Reluctant Adopters the lowest. Age-group differences were small and directionally uncertain across classes. Pragmatic Adopters occupied an intermediate position\u0026mdash;closer to Champions on usage frequency, satisfaction, and recommendation, but closer to Reluctant Adopters on time saved.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cem\u003eStacked Posterior Class Means and 95% HDIs for Auxiliary Outcomes\u003c/em\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOutcome\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eChampions\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eReluctant Adopters\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePragmatic Adopters (Reference)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eScale\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUsage Frequency\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.51 [4.17, 4.85]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.61 [3.14, 4.10]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.22 [3.97, 4.48]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1\u0026ndash;5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTime Saved per Shift\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.93 [3.53, 4.33]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.29 [1.83, 2.74]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.90 [2.63, 3.18]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1\u0026ndash;5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSatisfaction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.75 [4.58, 4.92]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.72 [2.47, 2.98]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.96 [3.83, 4.09]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1\u0026ndash;5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRecommendation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9.45 [8.83, 10.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.65 [3.76, 5.52]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e8.14 [7.66, 8.59]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0\u0026ndash;10\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTechnology Affinity\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6.37 [6.00, 6.74]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.00 [3.51, 4.50]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.56 [5.28, 5.84]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1\u0026ndash;7\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge Group\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.13 [3.73, 4.56]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.85 [3.40, 4.29]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.81 [3.50, 4.12]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1\u0026ndash;6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUsage Duration\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.86 [2.57, 3.16]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.40 [1.99, 2.80]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.73 [2.52, 2.94]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1\u0026ndash;4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThree-Item Formative Index\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.77 [2.55, 3.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.53 [1.21, 1.85]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.48 [2.31, 2.65]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0\u0026ndash;3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003e\u003cem\u003eNote.\u003c/em\u003e Posterior means and 95% highest-density intervals (HDIs) from 120,000 stacked draws (m\u0026thinsp;=\u0026thinsp;15 imputations \u0026times; 4 chains \u0026times; 2,000 draws). Three-Item Formative Index\u0026thinsp;=\u0026thinsp;unweighted sum of three binary outcomes (intention to continue use\u0026thinsp;+\u0026thinsp;quality improved\u0026thinsp;+\u0026thinsp;handover improved; range 0\u0026ndash;3). Pragmatic Adopters served as the reference group in contrast models.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003ePosterior Contrasts\u003c/h2\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e presents the posterior contrasts for each class relative to Pragmatic Adopters (reference).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cem\u003eStacked Posterior Contrasts for Auxiliary Outcomes (Reference: Pragmatic Adopters)\u003c/em\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOutcome (Scale)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eChampions vs. Pragmatic Adopters Mean [95% HDI]\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eReluctant Adopters vs. Pragmatic Adopters Mean [95% HDI]\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eDirection\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eInterpretation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUsage Frequency (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.29 [\u0026minus;\u0026thinsp;0.10, 0.70]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;0.61 [\u0026minus;\u0026thinsp;1.15, \u0026minus;\u0026thinsp;0.07]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e~ / \u0026darr;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eReluctant Adopters use the tool less frequently; Champions show no clear frequency advantage\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTime Saved per Shift (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;1.03 [0.53, 1.52]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;0.62 [\u0026minus;\u0026thinsp;1.13, \u0026minus;\u0026thinsp;0.09]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026uarr; / \u0026darr;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eChampions save meaningfully more time; Reluctant Adopters save less\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSatisfaction (1\u0026ndash;5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.79 [0.57, 1.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;1.24 [\u0026minus;\u0026thinsp;1.52, \u0026minus;\u0026thinsp;0.95]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026uarr; / \u0026darr;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eClear gradient across all three classes\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRecommendation (0\u0026ndash;10)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;1.31 [0.55, 2.09]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;3.49 [\u0026minus;\u0026thinsp;4.49, \u0026minus;\u0026thinsp;2.49]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026uarr; / \u0026darr;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eLargest absolute effect; Reluctant Adopters score 3.5 points below Pragmatic Adopters\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTechnology Affinity (1\u0026ndash;7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.81 [0.34, 1.27]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;1.56 [\u0026minus;\u0026thinsp;2.12, \u0026minus;\u0026thinsp;0.98]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026uarr; / \u0026darr;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eClass membership co-varies with broader dispositional orientation toward technology\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge Group (1\u0026ndash;6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.32 [\u0026minus;\u0026thinsp;0.12, 0.88]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;0.03 [\u0026minus;\u0026thinsp;0.46, 0.54]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e~ / ~\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eNo meaningful age differences; age cannot guide segmentation\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUsage Duration (1\u0026ndash;4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.13 [\u0026minus;\u0026thinsp;0.17, 0.49]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;0.33 [\u0026minus;\u0026thinsp;0.77, 0.08]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e~ / ~\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eClass membership is not explained by length of exposure\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThree-Item Formative Index (0\u0026ndash;3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.30 [0.02, 0.57]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;0.95 [\u0026minus;\u0026thinsp;1.32, \u0026minus;\u0026thinsp;0.59]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026uarr; / \u0026darr;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eChampions endorse nearly all adoption outcomes; Reluctant Adopters endorse fewer than half\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003e\u003cem\u003eNote.\u003c/em\u003e Effects are stacked posterior mean differences on the original scale with 95% highest-density intervals (HDIs). Direction column: \u0026uarr; = Champions robustly higher than Pragmatic Adopters (HDI entirely above zero); \u0026darr; = Reluctant Adopters robustly lower than Pragmatic Adopters (HDI entirely below zero); ~ = inconclusive (HDI crosses zero). Format: Champions / Reluctant Adopters direction. Three-Item Formative Index\u0026thinsp;=\u0026thinsp;unweighted sum of three binary outcomes (range 0\u0026ndash;3).\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eChampions consistently scored higher than Pragmatic Adopters, with clear differences in satisfaction (+\u0026thinsp;0.79 [0.57, 1.00]), time saved per shift (+\u0026thinsp;1.03 [0.53, 1.52]), recommendation (+\u0026thinsp;1.31 [0.55, 2.09]), technology affinity (+\u0026thinsp;0.81 [0.34, 1.27]), and the three-item formative index (+\u0026thinsp;0.30 [0.02, 0.57]). Usage frequency (+\u0026thinsp;0.29 [\u0026minus;\u0026thinsp;0.10, 0.70]) and usage duration (+\u0026thinsp;0.13 [\u0026minus;\u0026thinsp;0.17, 0.49]) were directionally positive but uncertain, suggesting that Champions and Pragmatic Adopters use the system at comparable rates but differ qualitatively in their experience.\u003c/p\u003e \u003cp\u003eReluctant Adopters showed the inverse pattern with clear deficits across most outcomes. Recommendation was markedly lower (\u0026minus;\u0026thinsp;3.49 [\u0026minus;\u0026thinsp;4.49, \u0026minus;\u0026thinsp;2.49]), with a mean of 4.65 representing a 3.5-point gap below Pragmatic Adopters\u0026rsquo; 8.14; satisfaction was also substantially reduced (\u0026minus;\u0026thinsp;1.24 [\u0026minus;\u0026thinsp;1.52, \u0026minus;\u0026thinsp;0.95]), falling below the scale midpoint (M\u0026thinsp;=\u0026thinsp;2.72); and technology affinity showed a robust negative contrast (\u0026minus;\u0026thinsp;1.56 [\u0026minus;\u0026thinsp;2.12, \u0026minus;\u0026thinsp;0.98]). The three-item formative index contrast of \u0026minus;\u0026thinsp;0.95 [\u0026minus;\u0026thinsp;1.32, \u0026minus;\u0026thinsp;0.59] indicates that Reluctant Adopters\u0026mdash;with a mean of 1.53 out of 3.0\u0026mdash;endorse only about half the binary adoption outcomes, compared with Pragmatic Adopters (2.48) and Champions (2.77).\u003c/p\u003e \u003cp\u003eDirectly comparing Champions and Reluctant Adopters, Champions exceeded Reluctant Adopters by 2.03 satisfaction units, 4.80 recommendation points, 2.37 technology affinity units, 1.64 time-saved units, and 1.25 three-item formative index points, underscoring the qualitative distinctiveness of these profiles for implementation planning. Critically, neither age group nor usage duration showed directionally robust contrasts, suggesting that these variables were not strong correlates of class membership in this sample. Interestingly, the identified Champions had the highest descriptive mean age across the classes.\u003c/p\u003e \u003cp\u003ePosterior predictive checks confirmed adequate model fit, with model predictions matching observed data within 3% for all outcomes.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"DISCUSSION","content":"\u003cdiv id=\"Sec25\" class=\"Section2\"\u003e \u003ch2\u003ePrincipal Findings\u003c/h2\u003e \u003cp\u003eThe latent class analysis identified three acceptance classes among practicing nurses who used an AI-based speech assistant for documentation. Champions, approximately one quarter of participants, combined the highest scores on all UTAUT2-derived constructs\u0026mdash;including performance expectancy, effort expectancy, hedonic motivation, habit, facilitating conditions, social influence, and trust\u0026mdash;with the strongest behavioral and attitudinal outcomes: the highest perceived improvements (as indicated by the formative index) and recommendation rates, greater satisfaction, and greater perceived time savings. Pragmatic Adopters, approximately half of the sample, endorsed facilitating conditions, trust, and ease of use, while reporting moderate enjoyment and perceived performance gains. Their outcomes were solid and characteristic of functional adoption. Reluctant Adopters, approximately one fifth of participants, reported lower performance and effort expectancy and lower enjoyment, alongside clearly weaker outcomes across satisfaction, recommendation, and usage behavior. Taken together, the pattern suggests that core UTAUT2 constructs do not operate in isolation but cluster into coherent user classes that align with meaningful differences in actual and perceived use.\u003c/p\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003eComparison With Previous Work\u003c/h2\u003e \u003cp\u003eThe Champions class conceptually resembles higher-acceptance or innovation-ready classes reported in adjacent literatures. In radiology, latent class analysis distinguished high- versus low-acceptance clinicians, and acceptance interacted with workload and AI use in explaining burnout, indicating that acceptance heterogeneity has clinical consequences.\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e In higher education, a faculty segmentation identified optimistic, critical, critically reflected, and neutral classes. In this context, the optimistic class showed the strongest link between AI self-efficacy and AI usage, echoing the high-engagement orientation similar to the Champions class.\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e Within nursing, related findings have recovered a class characterized by comparatively higher informatics-competence scores associated with greater perceived usefulness of health information systems,\u003csup\u003e24\u003c/sup\u003e an optimal innovator class among U.S. nurses,\u003csup\u003e32\u003c/sup\u003e and a class reflecting higher innovative behavior among nurses in Chinese hospitals.\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e A population-level study in Germany also identified an active user class for digitalized services. Although that analysis concerned citizens rather than professionals, the active class parallels a high-adoption pattern.\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e Across adjacent findings, the prevalence of high-acceptance classes in regard to technology varies substantially by population and context: the active class in the German citizen study comprised 18.3%, and the optimistic faculty profile roughly one third \u0026mdash; broadly consistent with Champions here. In the purposively sampled U.S. innovation-engaged nurse cohort, the 'optimal innovator' class was far larger with about 83%, underscoring that prevalence is highly context-dependent \u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e,\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e,\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003ePragmatic Adopters align with two streams of evidence. At the citizen level, the largest class in the German population study comprised 'Potential' users who had not yet used many digitalized healthcare services but were in principle willing to do so. They were younger, more educated, more frequently employed, and less skeptical than rejecters.\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e This functional profile maps closely onto the present Pragmatic Adopters, who likewise reported moderate but positive perceptions under supportive conditions. Another study among Finnish registered nurses identified groups with comparatively lower, moderate, and higher informatics competence, with higher-competence groups reporting greater perceived usefulness of the health information system.\u0026sup2;⁴ While competence is related but not equivalent to acceptance,\u003csup\u003e30\u003c/sup\u003e the finding that capability groupings align with perceived usefulness helps explain why Pragmatic Adopters may sustain steady use when facilitating conditions are present.\u003c/p\u003e \u003cp\u003eReluctant Adopters mirror lower-acceptance segments described in prior research. In the German citizen study, a \u0026ldquo;Rejecting\u0026rdquo; class was characterized by skepticism and lower readiness to use digitalized services.\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e Hesitancy in adoption is generally accompanied by diminished trust in system reliability, elevated concern over data security, lower enjoyment, and more negative evaluations of the technology such as perceived performance expectancy and effort expectancy.\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e Wang and colleagues\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e found that even participants with high levels of objective AI knowledge could retain strongly negative attitudes and moderate behavioral intentions if trust or risk perception was unfavorable, suggesting that knowledge alone does not alleviate skepticism. Similarly, Leary and colleagues\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e identified four classes among nurses engaged in innovative behaviors and emphasized the importance of organizational support, innovation exposure, and protected time for fostering innovativeness.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section3\"\u003e \u003ch2\u003eTheoretical and Practical Implications\u003c/h2\u003e \u003cp\u003eThe findings extend UTAUT2 by showing that core constructs tend to cluster into stable user groups rather than operating as isolated levers. Champions combined high perceived usefulness, ease, trust, habit, and enjoyment; Pragmatic Adopters exhibited a functional pattern in which facilitating conditions and ease were present but affective engagement and perceived gains were moderate; Reluctant Adopters scored consistently lower across all domains. This suggests that positive perceptions of usefulness, ease of use, trust, habit, and enjoyment tend to co-occur and highlights hedonic motivation as a practically important correlate of higher acceptance in clinical documentation when systems are reliable and are aligned with routine workflows. The speech modality may be experienced as pleasant to use in routine care, which, together with perceived workload relief, can strengthen sustained engagement.\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eImplementation strategies can be tailored accordingly. Champions could serve as peer mentors and early validators who model effective use in realistic workflows.\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e Pragmatic Adopters benefit from robust facilitating conditions\u0026mdash;including rapid support and opportunities for practice\u0026mdash;that maintain perceptions of ease.\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e,\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e Reluctant Adopters require confidence-building that addresses perceived reliability, effort, and risk directly.\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e Staged onboarding, protected time to integrate voice commands, and transparent feedback on recognition accuracy are likely to be more effective than generic training programs. Qualitative fieldwork on AI speech documentation highlights the importance of training, social influence, continuous multimodal technical support, and iterative refinement based on user feedback for successful adoption.\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec28\" class=\"Section2\"\u003e \u003ch2\u003eLimitations\u003c/h2\u003e \u003cp\u003eSeveral limitations should be considered. First, the cross-sectional design precludes causal inference about how classes emerge or evolve with experience. Second, the sampling frame included active users in organizations where the speech assistant was already deployed, so non-users and early refusers were likely not captured, which may bias estimates toward more favorable experiences. Third, outcomes were self-reported and may be affected by social desirability or recall bias; time-motion measures would provide stronger validation.\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e Fourth, trust and hedonic motivation were assessed with single-item indicators, limiting measurement depth. Fifth, generalizability is bounded by the focus on German long-term care facilities and a single AI speech assistant.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec29\" class=\"Section2\"\u003e \u003ch2\u003eOutlook and Future Directions\u003c/h2\u003e \u003cp\u003eLongitudinal designs would add considerable depth by testing whether users transition across classes as accuracy, training, and workflow fit evolve over time. Latent transition or growth mixture models can quantify movement between profiles and identify predictors of shifts from Reluctant Adopter to Pragmatic Adopter or Champion status.\u003c/p\u003e \u003cp\u003eWith larger samples, Bayesian network or structural equation approaches could test hypothesized directional relationships among UTAUT2 constructs, moving beyond descriptive classification toward models that specify which constructs actually drive acceptance and represent bridges between constructs. Complementarily, once classes have been identified in larger samples, network analysis methods could be applied within and across classes to identify central (hub) variables that sustain positive acceptance and bridge variables whose strengthening could facilitate movement between acceptance profiles. Such analyses would help prioritize intervention targets within each segment.\u003c/p\u003e \u003cp\u003eClass-tailored interventions could also be developed and trialed: peer-mentor networks and co-design workshops with Champions to identify core drivers for adoption, reinforced support and micro-learning exercises for Pragmatic Adopters to strengthen existing capabilities, and enhanced support and requirements analysis for Reluctant Adopters, with class-specific success criteria defined in advance. Finally, integrating objective system analytics alongside self-report measures would strengthen the validity of future evaluations.\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusions","content":"\u003cp\u003e The proposed three-class taxonomy suggests how nurses in long-term care relate to AI-based speech documentation in routine practice. Champions combined high perceived usefulness, ease of use, trust, habit, and enjoyment with strong adoption and perceived benefits. Pragmatic Adopters adopted the system functionally and emphasized supportive conditions. The smallest group, Reluctant Adopters, remained a measured profile despite regular usage. The taxonomy offers a practical basis for developing and tailoring implementation strategies in real care settings: pairing functional reliability with visible benefits, sustaining fast social and organizational support, providing practice opportunities, and addressing concerns directly for hesitant users. When adoption strategies are aligned with user classes\u0026rsquo; specific needs and characteristics, AI speech documentation is more likely to translate from technical promise to perceptible relief in documentation and handovers, without assuming that technology is a benefit by default. Class-aware implementation provides a path toward inclusive, evidence-based digital transformation in nursing.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003e\u003cem\u003eAI-ASSISTED SPEECH DOCUMENTATION USER GROUPS\u003c/em\u003e\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003ch2\u003eCompeting Interests\u003c/h2\u003e\u003cp\u003eK.S. was employed by voize GmbH, the developer of the mobile AI speech documentation system evaluated in this study, at the time of data collection and manuscript preparation. The study was conducted within the joint research project PYSA. All other authors declare that they have no competing interests.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e \u003ch2\u003eEthics Declaration\u003c/h2\u003e \u003cp\u003eThis study involved a voluntary and anonymous online survey of adult nursing staff in German senior care facilities. Prior to participation, all respondents received study information and provided electronic informed consent. No patient data was collected. To protect participants\u0026rsquo; privacy and reduce identifiability, no information on the specific care facility was collected and only a minimal set of sociodemographic variables was recorded. The questionnaire was reviewed and pretested by experienced nurses, and the study procedure was approved by the employee representative committees of the participating care facilities. The study was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eThis work received funding through the project PYSA (\u0026ldquo;Pflegedokumentation mit Hybridem Sprachassistenten\u0026rdquo;), a joint research project funded by the German Federal Ministry of Education and Research (now the Federal Ministry of Research, Technology and Space).\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eConceptualization: D.F., S.N., C.B.-H., A.R., K.S., N.L. and S.S.-L.; methodology and study design: D.F. and S.N.; investigation and data acquisition, D.F., S.N., C.B.-H., A.R. and K.S.; formal analysis: D.F.; data interpretation: D.F., S.N., and S.S.-L.; writing-original draft preparation: D.F.; writing-review and editing: S.N., L.P., N.R., C.B.-H., A.R., K.S., N.L. and S.S.-L.; visualization: D.F.; supervision: N.L. and S.S.-L. All authors reviewed the manuscript, approved the final submitted version, and agree to be accountable for their own contributions.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eWe sincerely thank all participants and the care facilities that took part in this study for their valuable feedback, support, and engagement throughout the project.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe datasets generated and analyzed during the study are not publicly available. The datasets may be available from the corresponding author on reasonable request, subject to institutional, legal, and ethical requirements.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eFederal Statistical Office. Number of people working in hospital nursing care up 18% in ten years. (2022). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.destatis.de/EN/Press/2022/05/PE22_N026_2313.html\u003c/span\u003e\u003cspan address=\"https://www.destatis.de/EN/Press/2022/05/PE22_N026_2313.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcCarthy, B. et al. Electronic nursing documentation interventions to promote or improve patient safety and quality care: A systematic review. \u003cem\u003eJ. Nurs. Adm. Manag.\u003c/em\u003e \u003cb\u003e27\u003c/b\u003e (3), 491\u0026ndash;501. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/jonm.12727\u003c/span\u003e\u003cspan address=\"10.1111/jonm.12727\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHeislbetz, C. Die Dokumentation in der Altenpflege. Universit\u0026auml;t Bamberg. (2008). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://fis.uni-bamberg.de/handle/uniba/190\u003c/span\u003e\u003cspan address=\"https://fis.uni-bamberg.de/handle/uniba/190\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchenk, E. et al. Time motion analysis of nursing work in ICU, telemetry and medical-surgical units. \u003cem\u003eJ. Nurs. Adm. Manag.\u003c/em\u003e \u003cb\u003e25\u003c/b\u003e (8), 640\u0026ndash;646. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/jonm.12502\u003c/span\u003e\u003cspan address=\"10.1111/jonm.12502\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYen, P. Y. et al. Nurses\u0026rsquo; time allocation and multitasking of nursing activities: A time motion study. AMIA Annual Symposium Proceedings, 2018, 1137\u0026ndash;1146. (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGesner, E., Dykes, P. C., Zhang, L. \u0026amp; Gazarian, P. Documentation burden in nursing and its role in clinician burnout syndrome. \u003cem\u003eAppl. Clin. Inf.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e (5), 983\u0026ndash;990. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1055/s-0042-1757157\u003c/span\u003e\u003cspan address=\"10.1055/s-0042-1757157\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoy, A. J. et al. Measurement of clinical documentation burden among physicians and nurses using electronic health records: A scoping review. \u003cem\u003eJ. Am. Med. Inform. Assoc.\u003c/em\u003e \u003cb\u003e28\u003c/b\u003e (5), 998\u0026ndash;1008. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/jamia/ocaa325\u003c/span\u003e\u003cspan address=\"10.1093/jamia/ocaa325\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFerizaj, D. \u0026amp; Neumann, S. Assessing perceptions and experiences of an AI-driven speech assistant for nursing documentation: A qualitative study in German nursing homes. In M. Kurosu \u0026amp; A. Hashizume (Eds.), Human-Computer Interaction (pp. 17\u0026ndash;34). Springer Nature Switzerland. (2024). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-031-60449-2_2\u003c/span\u003e\u003cspan address=\"10.1007/978-3-031-60449-2_2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJoseph, J., Moore, Z. E. H., Patton, D., O\u0026rsquo;Connor, T. \u0026amp; Nugent, L. E. The impact of implementing speech recognition technology on the accuracy and efficiency of clinical documentation by nurses: A systematic review. \u003cem\u003eJ. Clin. Nurs.\u003c/em\u003e \u003cb\u003e29\u003c/b\u003e (13\u0026ndash;14), 2125\u0026ndash;2137. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/jocn.15261\u003c/span\u003e\u003cspan address=\"10.1111/jocn.15261\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEhrler, F., Wu, D. T. Y., Ducloux, P. \u0026amp; Blondon, K. A mobile application to support bedside nurse documentation and care: A time and motion study. \u003cem\u003eJAMIA Open.\u003c/em\u003e \u003cb\u003e4\u003c/b\u003e (3), ooab046. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/jamiaopen/ooab046\u003c/span\u003e\u003cspan address=\"10.1093/jamiaopen/ooab046\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchwabe, K., Ferizaj, D., Neumann, S., Strube-Lahmann, S. \u0026amp; Lahmann, N. Time Savings Through an AI Speech Assistant for Nursing Documentation: A Pre-Post Time-Motion Study in German Long-Term Care. \u003cem\u003eJ. Med. Internet Res.\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2196/86078\u003c/span\u003e\u003cspan address=\"10.2196/86078\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAmmenwerth, E., Mansmann, U., Iller, C. \u0026amp; Eichst\u0026auml;dter, R. Factors affecting and affected by user acceptance of computer-based nursing documentation: Results of a two-year study. \u003cem\u003eJ. Am. Med. Inform. Assoc.\u003c/em\u003e \u003cb\u003e10\u003c/b\u003e (1), 69\u0026ndash;84. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1197/jamia.m1118\u003c/span\u003e\u003cspan address=\"10.1197/jamia.m1118\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2003).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVenkatesh, V., Thong, J. Y. L. \u0026amp; Xu, X. Consumer acceptance and use of information technology: Extending the unified theory of acceptance and use of technology. \u003cem\u003eMIS Q.\u003c/em\u003e \u003cb\u003e36\u003c/b\u003e (1), 157\u0026ndash;178. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2307/41410412\u003c/span\u003e\u003cspan address=\"10.2307/41410412\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee, A. T., Ramasamy, R. K. \u0026amp; Subbarao, A. Understanding psychosocial barriers to healthcare technology adoption: A review of TAM and UTAUT frameworks. \u003cem\u003eHealthcare\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e (3), 250. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/healthcare13030250\u003c/span\u003e\u003cspan address=\"10.3390/healthcare13030250\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVenkatesh, V., Morris, M. G., Davis, G. B. \u0026amp; Davis, F. D. User acceptance of information technology: Toward a unified view. \u003cem\u003eMIS Q.\u003c/em\u003e \u003cb\u003e27\u003c/b\u003e (3), 425\u0026ndash;478. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2307/30036540\u003c/span\u003e\u003cspan address=\"10.2307/30036540\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2003).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHussain, A. et al. The mediating effects of perceived usefulness and perceived ease of use on nurses\u0026rsquo; intentions to adopt advanced technology. \u003cem\u003eBMC Nurs.\u003c/em\u003e \u003cb\u003e24\u003c/b\u003e (1), 33. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12912-024-02648-8\u003c/span\u003e\u003cspan address=\"10.1186/s12912-024-02648-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee, W. I., Fu, H. P., Mendoza, N. \u0026amp; Liu, T. Y. Determinants impacting user behavior towards emergency use intentions of m-health services in Taiwan. \u003cem\u003eHealthcare\u003c/em\u003e \u003cb\u003e9\u003c/b\u003e (5), 535. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/healthcare9050535\u003c/span\u003e\u003cspan address=\"10.3390/healthcare9050535\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKn\u0026ouml;chelmann, A. et al. User profiles in digitalized healthcare: Active, potential, and rejecting\u0026mdash;A cross-sectional study using latent class analysis. \u003cem\u003eBMC Health Serv. Res.\u003c/em\u003e \u003cb\u003e24\u003c/b\u003e (1), 1083. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s12913-024-11523-w\u003c/span\u003e\u003cspan address=\"10.1186/s12913-024-11523-w\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLanza, S. T. \u0026amp; Rhoades, B. L. Latent class analysis: An alternative perspective on subgroup analysis in prevention and treatment. \u003cem\u003ePrev. Sci.\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e (2), 157\u0026ndash;168. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11121-011-0201-1\u003c/span\u003e\u003cspan address=\"10.1007/s11121-011-0201-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSinha, P., Calfee, C. S. \u0026amp; Delucchi, K. L. Practitioner\u0026rsquo;s guide to latent class analysis: Methodological considerations and common pitfalls. \u003cem\u003eCrit. Care Med.\u003c/em\u003e \u003cb\u003e49\u003c/b\u003e (1), e63\u0026ndash;e79. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1097/CCM.0000000000004710\u003c/span\u003e\u003cspan address=\"10.1097/CCM.0000000000004710\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLinzer, D. A. \u0026amp; Lewis, J. B. poLCA: An R package for polytomous variable latent class analysis. \u003cem\u003eJ. Stat. Softw.\u003c/em\u003e \u003cb\u003e42\u003c/b\u003e, 1\u0026ndash;29. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18637/jss.v042.i10\u003c/span\u003e\u003cspan address=\"10.18637/jss.v042.i10\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAboueid, S., Meyer, S. B., Wallace, J. \u0026amp; Chaurasia, A. Latent classes associated with the intention to use a symptom checker for self-triage. \u003cem\u003ePLOS ONE\u003c/em\u003e. \u003cb\u003e16\u003c/b\u003e (11), e0259547. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0259547\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0259547\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStaeck, R., St\u0026uuml;ble, M. \u0026amp; Dr\u0026uuml;ge, M. Exploring subgroups of acceptance prediction for e-mental health among psychotherapists-in-training: A latent class analysis. \u003cem\u003eFront. Psychiatry\u003c/em\u003e. \u003cb\u003e15\u003c/b\u003e, 1296449. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyt.2024.1296449\u003c/span\u003e\u003cspan address=\"10.3389/fpsyt.2024.1296449\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaihlanen, A. M. et al. Nursing informatics competence profiles and perceptions of health information system usefulness among registered nurses: A latent profile analysis. \u003cem\u003eJ. Adv. Nurs.\u003c/em\u003e \u003cb\u003e79\u003c/b\u003e (10), 4022\u0026ndash;4033. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/jan.15718\u003c/span\u003e\u003cspan address=\"10.1111/jan.15718\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFu, L., Xie, Y., Zhu, Y., Zhang, C. \u0026amp; Ge, Y. Innovative behavior profile and its associated factors among nurses in China: A cross-sectional study based on latent profile analysis. \u003cem\u003eBMJ Open.\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e (6), e084932. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1136/bmjopen-2024-084932\u003c/span\u003e\u003cspan address=\"10.1136/bmjopen-2024-084932\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBundesministerium f\u0026uuml;r Forschung, Technologie und Raumfahrt. PYSA: Pflege entlasten - Pflegedokumentation mit hybridem Sprachassistenten. (2025). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.interaktive-technologien.de/projekte/pysa\u003c/span\u003e\u003cspan address=\"https://www.interaktive-technologien.de/projekte/pysa\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWurpts, I. C. \u0026amp; Geiser, C. Is adding more indicators to a latent class analysis beneficial or detrimental? Results of a Monte-Carlo study. \u003cem\u003eFront. Psychol.\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpsyg.2014.00920\u003c/span\u003e\u003cspan address=\"10.3389/fpsyg.2014.00920\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2014). 5.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu, H. et al. Artificial intelligence and radiologist burnout. \u003cem\u003eJAMA Netw. Open.\u003c/em\u003e \u003cb\u003e7\u003c/b\u003e (11), e2448714. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1001/jamanetworkopen.2024.48714\u003c/span\u003e\u003cspan address=\"10.1001/jamanetworkopen.2024.48714\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMah, D. K. \u0026amp; Gro\u0026szlig;, N. Artificial intelligence in higher education: Exploring faculty use, self-efficacy, distinct profiles, and professional development needs. \u003cem\u003eInt. J. Educational Technol. High. Educ.\u003c/em\u003e \u003cb\u003e21\u003c/b\u003e (1), 58. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s41239-024-00490-1\u003c/span\u003e\u003cspan address=\"10.1186/s41239-024-00490-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFerizaj, D., Perotti, L., Dahms, R. \u0026amp; Heimann-Steinert, A. Use of technology in old age: Associations between acceptance, competence, control, interest and social indicators in individuals over 60 years old. \u003cem\u003eZ. f\u0026uuml;r Gerontologie und Geriatrie\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s00391-023-02225-9\u003c/span\u003e\u003cspan address=\"10.1007/s00391-023-02225-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, H., Sun, Q., Gu, L., Lai, K. \u0026amp; He, L. Diversity in people\u0026rsquo;s reluctance to use medical artificial intelligence: Identifying subgroups through latent profile analysis. \u003cem\u003eFront. Artif. Intell.\u003c/em\u003e \u003cb\u003e5\u003c/b\u003e, 1006173. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/frai.2022.1006173\u003c/span\u003e\u003cspan address=\"10.3389/frai.2022.1006173\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeary, M. et al. Creating a phenotype and taxonomy of nurses engaging in innovative behaviors. \u003cem\u003eOnline. J. Issues. Nurs.\u003c/em\u003e \u003cb\u003e30\u003c/b\u003e (1). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3912/OJIN.Vol30No01Man02\u003c/span\u003e\u003cspan address=\"10.3912/OJIN.Vol30No01Man02\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchwabe, K., Ferizaj, D. \u0026amp; Neumann, S. Reducing Nurses' Workload with an AI Speech Assistant for Documentation (2025). In: \u003cem\u003eProceedings of the First International Conference on AI in Medicine and Healthcare (AiMH' 2025)\u003c/em\u003e, pp. 119\u0026ndash;122.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAbril-Pla, O. et al. PyMC: A modern and comprehensive probabilistic programming framework in Python. \u003cem\u003ePeerJ Comput. Sci.\u003c/em\u003e \u003cb\u003e9\u003c/b\u003e, e1516. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.7717/peerj-cs.1516\u003c/span\u003e\u003cspan address=\"10.7717/peerj-cs.1516\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumar, R., Carroll, C., Hartikainen, A. \u0026amp; Martin, O. ArviZ: A unified library for exploratory analysis of Bayesian models in Python. \u003cem\u003eJ. Open. Source Softw.\u003c/em\u003e \u003cb\u003e4\u003c/b\u003e (33), 1143. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.21105/joss.01143\u003c/span\u003e\u003cspan address=\"10.21105/joss.01143\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"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":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"latent class analysis, nursing, automatic speech recognition, mobile documentation","lastPublishedDoi":"10.21203/rs.3.rs-9557939/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9557939/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis study applied latent class analysis (LCA) to identify distinct technology acceptance groups among nursing staff using an AI speech assistant for nursing documentation (voize) in German long-term care facilities. Using a cross-sectional survey design (\u003cem\u003eN\u003c/em\u003e\u0026thinsp;=\u0026thinsp;134) and three complementary indicator specifications derived from Unified Theory of Acceptance and Use of Technology (UTAUT2) constructs, we identified a consistent three-class solution with excellent entropy (.982): Champions (27.3%), Pragmatic Adopters (52.6%), and Reluctant Adopters (20.1%). Monte Carlo parameter-recovery simulations confirmed the robustness of this structure across simulated sample sizes from 100 to 1,000, with most recovery rates above 90%. Class profiles were validated through a Bayesian multivariate modelling approach with stacked posterior inference across 15 multiply imputed datasets, totaling 120,000 posterior draws. Compared with Pragmatic Adopters, Champions were more satisfied, perceived greater time savings, and were more willing to recommend the system, whereas Reluctant Adopters showed the opposite pattern across all practice-related outcomes. The clearest practical contrast was in likelihood to recommend: Champions rated the system 9.5 out of 10 on average, compared with 8.1 among Pragmatic Adopters and 4.7 among Reluctant Adopters. Champions endorsed nearly all three positive implementation outcomes on average\u0026mdash;intention to continue use, better documentation, and better handovers\u0026mdash;whereas Reluctant Adopters endorsed only about half. Age and duration of use did not meaningfully differentiate classes. These findings reveal distinct acceptance groups in AI-assisted nursing documentation and suggest that class-specific, targeted implementation strategies, rather than a uniform rollout, should be considered to support skeptical users while sustaining enthusiastic adoption.\u003c/p\u003e","manuscriptTitle":"AI-Assisted Speech Documentation in Nursing: Identifying User Groups via Latent Class Analysis","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-06 06:09:06","doi":"10.21203/rs.3.rs-9557939/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-05-04T08:17:45+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-30T12:17:57+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2026-04-30T10:47:22+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"41823c80-f92e-49cf-9486-d76f7557bcbe","owner":[],"postedDate":"May 6th, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Revision requested","date":"2026-05-04T08:17:45+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-30T12:17:57+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2026-04-30T10:47:22+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"in-revision","subjectAreas":[{"id":67464869,"name":"Health sciences/Health care"},{"id":67464870,"name":"Physical sciences/Mathematics and computing"}],"tags":[],"updatedAt":"2026-05-06T06:09:07+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-06 06:09:06","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9557939","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9557939","identity":"rs-9557939","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}