Developing an EMG-Based Distal Upper Extremity Tool for Assessing Risk of Distal Upper Extremity Symptoms: A Proof-of-Concept Comparative Evaluation

Developing an EMG-Based Distal Upper Extremity Tool for Assessing Risk of Distal Upper Extremity Symptoms: A Proof-of-Concept Comparative Evaluation | medRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (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];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-P4HH5NV'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search Developing an EMG-Based Distal Upper Extremity Tool for Assessing Risk of Distal Upper Extremity Symptoms: A Proof-of-Concept Comparative Evaluation View ORCID Profile Xuelong Fan , View ORCID Profile Johan Rydgård , View ORCID Profile Pasan Hettiarachchi , View ORCID Profile Kristina Eliasson , View ORCID Profile Camilla Dahlqvist , View ORCID Profile Peter J. Johansson doi: https://doi.org/10.1101/2025.10.03.25337266 Xuelong Fan 1 Department of Medical Science, Occupational and Environmental Medicine, Uppsala University , Uppsala, Sweden Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Xuelong Fan For correspondence: xuelong.fan{at}uu.se Johan Rydgård 2 Department of Occupational and Environmental Medicine, Uppsala University Hospital , Uppsala, Sweden Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Johan Rydgård Pasan Hettiarachchi 1 Department of Medical Science, Occupational and Environmental Medicine, Uppsala University , Uppsala, Sweden Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Pasan Hettiarachchi Kristina Eliasson 1 Department of Medical Science, Occupational and Environmental Medicine, Uppsala University , Uppsala, Sweden 2 Department of Occupational and Environmental Medicine, Uppsala University Hospital , Uppsala, Sweden Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Kristina Eliasson Camilla Dahlqvist 3 Division of Occupational and Environmental Medicine, Department of Laboratory Medicine, Lund University , Lund, Sweden 4 Department of Occupational and Environmental Medicine, Skåne University Hospital , Lund, Sweden Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Camilla Dahlqvist Peter J. Johansson 1 Department of Medical Science, Occupational and Environmental Medicine, Uppsala University , Uppsala, Sweden 2 Department of Occupational and Environmental Medicine, Uppsala University Hospital , Uppsala, Sweden Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Peter J. Johansson Abstract Full Text Info/History Metrics Data/Code Preview PDF Abstract Work-related distal upper extremity (DUE) disorders are common in hand-intensive occupations due to cumulative strain. Existing assessment tools rely on subjective ratings and often fail to capture dynamic workload over varied work schedules. This proof-of-concept study evaluated a surface electromyography (EMG)-based version of the Distal Upper Extremity Tool (EMG-DUET), designed to provide objective and task-flexible estimates of cumulative strain and risk. Nineteen participants were video recorded during job tasks while forearm EMG was collected, yielding 78 manual tasks for analysis. Risk estimates from EMG-DUET were compared with those from forearm EMG peak values (EMG-Peak), visual observation-based DUET (OBS-DUET), the ACGIH Hand Activity Level Threshold Limit Value (OBS-HATLV), and prevalence estimates of DUE symptoms. Both DUET variants aligned most closely with their input-matched references, and EMG-based tools showed stronger, though modest, correlations with symptom prevalence than observation-based tools. EMG-DUET demonstrates promise, but refinement and validation in full-shift, prospective studies are needed. 1 Introduction Work-related distal upper extremity (DUE) disorders remain a major concern in occupational health ( Govaerts et al., 2021 ). Physical workload—particularly among tasks involving hand-intensive work—plays a key role in the development of DUE disorders such as epicondylitis and carpal tunnel syndrome, primarily through mechanisms involving cumulative tissue damage and chronic inflammation ( Dong et al., 2022 ; Keir et al., 2021 ). Effective workload assessment is critical for predicting and managing the risk of work-related DUE disorders. Traditional workload and risk assessment tools include self-reported surveys ( Grant et al., 1999 ; Robertson et al., 2003 ), observational risk assessment checklists ( Kapellusch et al., 2014 ; McAtamney & Nigel Corlett, 1993 ; Yung et al., 2019 ), and technical measurements ( Balogh et al., 2019 ; Arvidsson et al., 2021 ). Self-reported surveys may be either retrospective or momentary, capturing subjective perceptions of workload and demonstrating reasonable reliability ( Coombes et al., 2021 ; Fanchini et al., 2017 ). However, retrospective surveys are prone to recall bias, especially as the recall period lengthens ( Bell et al., 2019 ). While momentary surveys can reduce recall bias, they may disrupt workflow and have been shown to influence both behavior and perception ( Ortega et al., 2024 ). Observational risk assessment tools—such as on-site task analysis or post-hoc video assessment—minimize participant disruption but introduce inter-observer variability and observer bias, including sex bias ( Dahlgren et al., 2022 ; Graben et al., 2022 ; Nyman et al., 2023 ). The dependence on manual evaluation also makes long-term monitoring across multiple tasks, over a full day, resource-intensive and operationally challenging. Technical measurement approaches, such as surface electromyography (sEMG), offer continuous and passive monitoring based on physiological signals. sEMG provides parameters related to muscle activity that must be translated into binary risk-level estimates using thresholds based on empirical epidemiological data ( Nordander et al., 2013 ). However, these tools often rely on standardized assumptions, such as an 8-hour workday, and consequently struggle to accommodate increasingly common flexible or nonstandard work schedules ( Bolino et al., 2021 ). The reliance on fixed durations and individual-specific analyses complicates cumulative long-term risk assessment and makes group-level risk evaluation more challenging. The Distal Upper Extremity Tool (DUET) was developed specifically to address the limitations inherent in traditional observational assessment methods, particularly their inability to estimate cumulative risk across multiple tasks performed throughout workdays with various schedules. Its framework applies material fatigue failure theory to estimate cumulative tissue damage from perceived or observed exertions and aggregates total damages to predict the probability of DUE symptom occurrence ( Gallagher et al., 2018 ). Unlike conventional assessment tools, DUET calculates cumulative tissue damage by integrating both the intensity and frequency of exertions, allowing for risk predictions over flexible, user-defined time intervals ( Gallagher et al., 2018 ; Mehdizadeh et al., 2020 ). Because DUET uses a logistic regression model grounded in epidemiological data to estimate the probability of DUE symptoms, it can assess risk across multiple tasks for individual workers as well as aggregate exposures for groups performing diverse activities ( Jorgensen et al., 2024 ; Mehdizadeh et al., 2020 ). Despite these strengths, DUET’s current implementation depends primarily on subjective observational input to estimate exertion levels, introducing both inter- and intra-observer variability and limiting its potential for continuous, long-term monitoring in dynamic work settings ( Dahlgren et al., 2022 ; Nyman et al., 2023 ). This reliance on subjective observation is a significant limitation that could be addressed by integrating technical measurements capable of capturing both repetition and load magnitude ( Nyman et al., 2023 ). Recent research has shown that using a forearm-based EMG approach offers a practical and reliable means of estimating exertion during manual tasks ( Fan et al., 2024 ). By integrating EMG-derived exertion estimates into the DUET framework, it becomes possible to create a flexible and promising tool for capturing dynamic patterns of repetitive exertions in the DUEs. This integration enables continuous quantification of cumulative tissue damage and prediction of DUE disorder risk, even under variable and non-standardized work conditions. Despite this potential, the application of EMG data within the DUET framework for assessing DUE disorder risks remains largely unexplored in the current literature. Therefore, as a proof of concept, this study aimed to develop an EMG-based version of the DUET (EMG-DUET) and to evaluate whether it can generate risk estimates comparable to: (1) the previously validated observation-based DUET (OBS-DUET; Gallagher et al., 2018 ); 2) two other established, non-fatigue-failure-based risk assessment benchmarks; and 3) published prevalence of work-related DUE symptoms in relevant occupations ( Statistikdatabasen, 2023 ). The two benchmarks were a) a revised version of the proposed action level of the peak muscular load in the forearm derived from sEMG ( Arvidsson et al., 2021 ), referred to as EMG-Peak in this study, and b) the Revised American Conference of Governmental Industrial Hygienists (ACGIH)® Hand Activity Threshold Limit Value (HATLV)® based on visual observation ( Yung et al., 2019 ), referred to as OBS-HATLV. 2 Materials and methods 2.1 Participants Nineteen participants were recruited for the study, of whom 26% were female. The mean age was 37 years, and the average body mass index (BMI) was 26 kg/m². Most participants (84%) were right-handed. Task assessments were performed on each participant’s preferred side, except for one individual who was assessed on both sides. For those evaluated on the left side (N = 3), the average maximal left-hand grip strength was 36 kg, while for right-sided assessments (N = 17), it was 50 kg. ( Table 1 ). All participants provided written informed consent prior to participation, and the study was approved by the Swedish Ethical Review Authority in Gothenburg (Dnr 2022-02531-01). View this table: View inline View popup Download powerpoint Table 1. Participant demographics (N = 19). For variables with missing data, the number of participants with available information is indicated by N. If not specified, data were available for all 19 participants. SD stands for standard deviation. 2.2 Tasks A total of 78 tasks involving DUE activities were recorded between 2022 and 2024. These tasks were selected using convenience sampling to maximize diversity across task types and occupational settings, and the resulting number exceeded the minimum suggested by the power analysis. All tasks reflected typical work activities from 14 different occupations and industries across Sweden. Task selection was guided collaboratively by professional ergonomists, participating companies, and the workers themselves, ensuring both representativeness and practical relevance. Each task was labeled with a combination of occupation abbreviation, occupation number, participant ID, and the hand side used during the task ( Table 2 ). Individual participants could perform multiple tasks; however, all tasks assigned to a given participant came from the same industry or occupational category, except for M04, who performed tasks from two categories. The average task duration was 7.3 minutes, with an interquartile range of 3.9 to 9.5 minutes. View this table: View inline View popup Table 2. The names and descriptions of tasks used in this study (N = 78). The task ID is composed as “sex letter” + “participant ID” + “side letter” + “-”+ “occupation abbreviation” + “task number”. Side letter L stands for the left side, without which it refers to the right side. 2.3 Protocols The measurement protocol consisted of two phases: a calibration phase and a t ask phase . Data collection was performed at the participants’ workplaces. 2.3.1 Calibration phase In the calibration phase, sEMG electrodes were first applied to the participant’s forearm. The participant was seated and instructed to position their working arm at a 90-degree angle from their side, with the wrist in a neutral position. While maintaining this posture, they held a hand dynamometer (G200, Biometrics Ltd, Newport, UK) suspended in the air and rested for at least one minute. After this rest period, participants performed a maximal voluntary contraction (MVC) by squeezing the dynamometer handle as forcefully as possible for at least two seconds, during which they received verbal encouragement. Throughout this effort, muscle activation was recorded using sEMG. ( Fan et al., 2024 ) 2.3.2 Task phase In the task phase, participants carried out one or more predetermined tasks, with the order of tasks being arbitrary. Each participant was instructed to perform their tasks as naturally as possible in their usual workplace environment, verbally reporting their perceived exertion for each self-defined effort. A revised OMNI-Resistance Exercise Scale (OMNI-RES), adapted specifically to forearm exertion and ranging from 0 (no effort) to 10 (maximal effort), was used ( Fan et al., 2024 ; Rydgård, 2025a ). Throughout the tasks, muscle activity was continuously monitored via sEMG, and video footage was recorded from one or two camera angles (GoPro, Inc., California, USA) at 25 frames per second. 2.4 Muscle Activity Forearm muscle activity was recorded using a through-forearm placement of sEMG electrodes. This placement was chosen to simplify the procedure while preserving the accuracy of exertion level estimations ( Fan et al., 2024 ). Two self-adhesive bipolar Ag/AgCl electrodes (N-00-S/25, Ambu, Penang, Malaysia) were applied: one over the belly of the m. extensor digitorum communis and the other over the m. flexor digitorum superficialis. A ground electrode was placed on the olecranon process at the elbow. Raw sEMG signals were captured with a digital data logger (DataLog MWX8, Biometrics Ltd., Newport, UK) and sampled at 1000 Hz per channel via a 24-bit A/D converter embedded in the logger. 2.5 Observation-based exertion After data collection, an experienced ergonomist reviewed the video recordings of each task to identify individual repetitions and assess their intensity. Intensity ratings were assigned using the revised OMNI-RES scale, based on both the ergonomist’s expertise and, when available, participants’ self-reported exertion ratings. For static exertions lasting longer than 5 seconds, the number of repetitions was calculated by assigning one additional repetition for each subsequent 5-second interval (for example, 5–10 seconds = 2 repetitions, 10–15 seconds = 3 repetitions, etc.). The analysis was performed using a custom annotation tool, ErgoAnnotation ( Rydgård, 2025b ), integrated into Blender (version 4.4.3; Blender Foundation, Amsterdam, Netherlands). 2.6 Data processing Raw sEMG signals were processed using an established pipeline that included filtering, smoothing, denoising, and normalization in preparation for exertion prediction ( Fan et al., 2024 ). Initially, raw sEMG data were filtered with a digital bandpass filter (20–400 Hz) and an additional 50 Hz notch comb filter (Q factor = 35) to eliminate powerline interference and other high-frequency artifacts, resulting in the base sEMG data. The base sEMG signals were then smoothed using a 0.125-second root mean square (RMS) moving window. To account for background activity, the smoothed signals were power-subtracted by the resting noise level. This resting noise level was estimated as the minimum value of the base sEMG during the resting period, further smoothed by a 2.375-second RMS moving window. The power-subtracted signal was calculated as follows: , where xᵢ is the base sEMG value at time point i, and x₀ is the resting noise level. Finally, the denoised and smoothed signals were normalized to the participant’s maximal voluntary electrical activation (MVE), expressing muscle activity as a percentage of MVE (%MVE). MVE was determined by the maximum value of the base sEMG signal during the MVC period, smoothed using a 0.5-second RMS window and power-subtracted using the same resting noise level. The resulting normalized signal (%MVE) was then used in all subsequent analyses. 2.7 Estimation of exertion Observation-based exertion ratings assigned by the ergonomist were incorporated directly into subsequent analyses as observational input. Since real-time muscle exertion during tasks could not be measured directly, exertion was instead estimated from normalized muscle activity using a previously established model ( Fan et al., 2024 ). This estimation relied on a mathematical function that converted normalized sEMG (%MVE) into estimated exertion (%MVC), as described in Equation 1 ( Eq. 1 ): where 𝑥 ! is the normalized muscle activity (%MVE) at the time point 𝑖, and 𝐸𝑀𝐺_𝐸𝑥𝑒𝑟𝑡𝑖𝑜𝑛 ! is the estimated exertion (%MVC) from EMG at the same time point. 2.8 Risk assessment outcomes In total, four risk assessment tools were applied to all tasks to estimate risk outcomes for the DUE symptoms, including 1) EMG-DUET and 2) OBS-DUET, and two other non-DUET-framework benchmarks, 3) EMG-Peak and 4) OBS-HATLV. 2.8.1 EMG-based DUET (EMG-DUET) EMG-DUET integrated EMG data into the DUET framework by using EMG-based estimates of exertion as input to predict the probability of developing DUE symptoms ( Gallagher et al., 2018 ). Continuous exertion estimates from EMG readings were processed with a Rainflow algorithm to classify the intensity and frequency of distinct exertion cycles for each task ( Nail-Ulloa et al., 2025 ). The fatigue-failure model within the DUET framework estimated tendon strain from exertion intensity, combining this with cycle frequency to calculate cumulative tissue damage based on theoretical cycles to failure, as derived from cadaveric research ( Schechtman & Bader, 2002 ). Cumulative damage was then normalized to a cumulative damage density (per unit time) according to task duration and projected to a 6-hour continuous work exposure, enabling comparison across different tools. Finally, the calculated cumulative damage was mapped to the probability of DUE symptom occurrence using the DUET framework’s epidemiological model. To ensure consistency between tools, EMG-based exertion values below 5 %MVC—the minimum exertion considered in OBS-DUET (corresponding to OMNI-RES = 0)—and exertion duration less than 0.25 seconds were excluded, as 0-rated exertions and exertions less than 0.25 seconds were not annotated in OBS-DUET. 2.8.2 Original DUET (OBS-DUET) OBS-DUET followed the original DUET and shares the same framework and most analytical pipeline as EMG-DUET. The key distinction lies in the source of input data. Rather than using EMG-based estimates of exertion, OBS-DUET used the observation-based exertions for each task from the experienced ergonomists to estimate the risk outcomes. All subsequent steps in the DUET framework—strain estimation, cumulative damage calculation, normalization, and risk probability estimation—were conducted identically to those in the EMG-DUET. 2.8.3 EMG-Peak EMG-Peak was determined as the 90th percentile of muscle activity (%MVE) recorded from sEMG signals with a through-forearm electrode placement ( Fan et al., 2024 ). This differs from the original proposed action level for the peak muscular load of the forearm ( Arvidsson et al., 2021 ), which used an extensor-specific EMG placement ( Nordander et al., 2004 ). Previous research has indicated that extensor-focused recordings may miss significant flexor activity ( Dahlqvist et al., 2024 ), and our findings support that the through-forearm configuration more accurately reflects overall forearm exertion ( Fan et al., 2024 ). Despite these anatomical differences, we adopted the same risk threshold for EMG-Peak (>30 %MVE) as proposed for extensor-based measurements ( Arvidsson et al., 2021 ), for its convenience and consistency under this proof-of-concept evaluation. This decision was further supported by a strong correlation (r=0.9) between the two placements, as demonstrated in prior work ( Fan et al., 2024 ). For the purposes of this study, EMG-Peak refers specifically to the through-forearm measurement unless otherwise noted. 2.8.4 OBS-HATLV OBS-HATLV refers to a modified version of the ACGIH® HATLV® ( Yung et al., 2019 ), which estimates the risk of carpal tunnel syndrome by combining measures of applied hand force and hand exertion repetition. To determine applied hand force, OMNI-RES scores for each exertion were first converted to Borg CR-10 using a matching table developed from empirical research (currently unpublished; Table 3 ). These Borg CR-10 scores were subsequently used to calculate the normalized peak force (NPF; Yung et al., 2019 ). HAL values were obtained by assessing repetition data from the annotation tool and applying a validated method ( Radwin et al., 2015 ), then averaging across the task. Finally, a continuous risk score was calculated using Equation 2 ( Eq. 2 ): View this table: View inline View popup Download powerpoint Table 3. The matching table between OMNI-RES and Borg CR-10. where 𝐻𝐴𝐿 ! and 𝑁𝑃𝐹 ! represent the estimated hand activity level and normalized peak force, respectively, for exertion 𝑖. Tasks were categorized as follows ( Yung et al., 2019 ): Score > 0.56: High risk (above the Threshold Limit Value, TLV) 0.36 < Score ≤ 0.56: Moderate risk (between the Action Limit, AL, and TLV) Score ≤ 0.36: Low risk (below AL) 2.9 Prevalence of Work-related DUE Symptoms The clinical relevance of the risk assessment results was explored by comparing task-level risk scores with real-world prevalence data for work-related DUE symptoms. These prevalence rates served as proxies for DUE disorders across occupational titles. Data were sourced from national surveys conducted by Statistics Sweden ( Statistikdatabasen, 2023 ), representing the average percentage of workers—regardless of sex—who reported work-related symptoms in the fingers, hands, or wrists during 2018, 2020, and 2022. Each occupation in the study was matched to its official title and code following the national classification system ( Statistikdatabasen, 2014 ). Only occupations with at least one available prevalence data point from the surveyed years were included. In certain cases, multiple occupations shared the same prevalence value due to data being available only at a higher level in the occupational classification hierarchy. 2.10 Statistics To evaluate the EMG-DUET tool, which integrates EMG-based exertion estimates into the DUET framework, analyses were structured along three main dimensions. First, to isolate the contribution of the EMG input, EMG-DUET was compared to OBS-DUET, with the analogous comparison between EMG-Peak and OBS-HATLV serving as a reference. Second, to assess the added value of the DUET framework itself, EMG-DUET was compared to EMG-Peak, with the OBS-DUET versus OBS-HATLV comparison as a parallel. Third, the risk outcomes from all four tools were compared to the real-world prevalence of DUE symptoms on an occupational level. Specifically, three analytical approaches were used: (1) risk-rank comparisons, (2) risk-level classification comparisons, and (3) outcome-rank comparisons. All statistical analyses were conducted using MATLAB 2023a (The MathWorks, Inc., Massachusetts, USA). Risk-rank comparisons were performed both visually and quantitatively. For visual assessment, horizontal bar plots illustrated task rankings based on each tool’s risk scores. To aid interpretation, tasks were sorted according to their ranks from the two common reference tools (OBS-HATLV and EMG-Peak), enabling a clear comparison of how all four tools classified risk across tasks. For quantitative analysis, a repeated-measures correlation (r rm ) was calculated between the ranks of risk outcomes for each pair of tools to account for unbalanced repetition (tasks) per subject ( Bakdash & Marusich, 2017 ). Specifically, the ranks were subject-centered (value minus that subject’s mean), and Pearson’s correlation was computed on these centered values for each tool pair. The 95% confidence intervals (CIs) were obtained via Fisher’s z transform and back-transformed to r rm to account for the skewness of averaged correlations ( Silver & Dunlap, 1987 ). Statistical significance was set at p < 0.05. Correlation strengths were interpreted using established thresholds ( Schober et al., 2018 ): negligible (<0.10), weak (0.10–0.39), moderate (0.40–0.69), strong (0.70–0.89), and very strong (≥0.90). Because no established risk classification thresholds exist for DUET tools, receiver operating characteristic (ROC) analysis was employed to evaluate the ability of EMG-DUET and OBS-DUET to predict risk classifications from established benchmark tools—EMG-Peak and OBS-HATLV ( Fawcett, 2006 ). For EMG-Peak, 30 %MVE was used as the criterium, and for OBS-HATLV, Threshold Limit Value (TLV) and Action Level (AL) were used as the criteria. Continuous risk scores from these benchmark tools were converted into binary high- and low-risk categories using these criteria. ROC curves for EMG-DUET and OBS-DUET were generated by plotting sensitivity (true positive rate) against 1 – specificity (false positive rate) across a range of possible thresholds. The area under the ROC curve (AUC) was used as the primary measure of classification accuracy, with higher values indicating better discrimination. Optimal cutoffs for each ROC curve were selected by maximizing Youden’s J statistic, i.e., sensitivity + specificity – 1 ( Youden, 1950 ). Sensitivity and specificity at these optimal thresholds were then compared to directly evaluate how the DUET tools performed relative to reference standards. To assess clinical relevance, outcome-rank comparisons were conducted. Specifically, task-level risk outcomes from each tool were first averaged within each participant and then further aggregated at the occupational level to align with corresponding prevalence data of DUE symptoms. A Spearman’s correlation (ρ) was then calculated between the aggregated risk outcomes and occupational-level prevalence of DUE symptoms. Correlation strengths were interpreted according to the same rule-of-thumb thresholds ( Schober et al., 2018 ), with statistical significance defined as p < 0.05. 3 Results 3.1 Risk-rank comparisons Overall, the two DUETs displayed distinct characteristics in task-level risk assessment. EMG-DUET yielded a narrower range of risk probabilities, spanning from 58% to 80% ( Figure 1a ), while OBS-DUET displayed a broader distribution, with probabilities ranging from 18% to 90% ( Figure 1b ). Download figure Open in new tab Figure 1. Comparison of risk assessment outcomes of tasks across (a) EMG-DUET, (b) OBS-DUET, (c) EMG-Peak, and (d) OBS-HATLV. Blacklines indicate the thresholds for categorizing risky tasks in the corresponding reference tools. Colors encode each task’s risk outcomes, ranging from low (cool hues) to high (warm hues). Comparing the two non-DUET benchmark tools, differences emerged in their task classification. OBS-HATLV identified 8 tasks (10% of all tasks) as exceeding the TLV threshold and 39 tasks (50%) as falling between the AL and TLV thresholds ( Figure 1d ). By contrast, EMG-Peak, with values ranging from 6 %MVE to 59 %MVE, classified 24 tasks (31%) as having peak muscular loads above the 30 %MVE threshold ( Figure 1c ). Significant correlations were observed between the DUET tools and their respective reference risk assessment instruments. EMG-DUET showed a significant, positive, and moderate-to-strong correlation with EMG-Peak %MVE (r rm [CIs] = 0.63 [0.45, 0.76], p < 0.001; Figure 2c ), while the correlations with OBS-DUET (r rm = 0.42 [0.18, 0.60], p < 0.001; Figure 2a ) and OBS-HATLV (r rm = 0.32 [0.07, 0.53], p = 0.014; Figure 2d ) were weaker but still statistically significant. In comparison, the correlation between OBS-DUET and OBS-HATLV was strong (r rm = 0.72 [0.57, 0.82], p < 0.001; Figure 2f ), whereas the correlation between OBS-DUET and EMG-Peak, though also moderate, was more dispersed (r rm = 0.45 [0.22, 0.63], p < 0.001; Figure 2e ). As a reference, the correlation between EMG-Peak and OBS-HATLV was significant but moderate-to-weak (r rm = 0.41 [0.18, 0.60], p = 0.001; Figure 2b ). Download figure Open in new tab Figure 2. Rank–rank comparisons among DUET and reference risk assessment tools. Each blue circle represents a task’s rank in paired comparisons of EMG-DUET (y-axis) versus (a) OBS-DUET, (c) EMG-Peak, and (d) OBS-HATLV (x-axis); OBS-DUET (y-axis) versus (e) EMG-Peak and (f) OBS-HATLV (x-axis); and EMG-Peak (y-axis) versus OBS-HATLV (x-axis) in (b). The black dotted line indicates identity (y = x). Each panel reports the repeated-measures correlation coefficient (rrm) and corresponding significance: *, p<0.05; **, p<0.01; ***, p<0.001. 3.2 Comparisons of risk-level classification When evaluated against the EMG-Peak reference ( Figure 3a ), the EMG-DUET exhibited high predictive accuracy, achieving an area under the curve (AUC) of 0.94. In contrast, the OBS-DUET showed moderate performance with an AUC of 0.68. Download figure Open in new tab Figure 3. Receiver-operating characteristic (ROC) curves for EMG-DUET (blue) and OBS-DUET (red) risk scores against the two reference tools: (a) EMG-Peak, and (b1) the Threshold Limit Value (TLV) and (b2) the Action Level (AL) in OBS-HATLV. The black dashed line in each panel denotes chance performance (AUC = 0.50), and the inset AUC values report each DUET’s ability to distinguish between risky and non-risky tasks against the corresponding reference tool. When compared to the TLV threshold in OBS-HATLV ( Figure 3b1 ), OBS-DUET demonstrated a good differentiation with an AUC of 0.92, while EMG-DUET performed worse than random chance, with an AUC of 0.42. Similarly, against the AL threshold in OBS-HATLV ( Figure 3b2 ), OBS-DUET continued to show strong predictive power with an AUC of 0.91, whereas EMG-DUET’s performance was comparatively weaker, with an AUC of 0.68. EMG-DUET and OBS-DUET exhibited distinct performance patterns when classifying task risk levels against EMG-Peak and OBS-HAL ( Table 4 ). EMG-DUET achieved optimal performance at a threshold of 70%, yielding a high sensitivity of 88% and specificity of 87% in comparison to EMG-Peak. However, its performance declined when evaluated against the TLV threshold in OBS-HATLV at an optimal threshold of 60%, where specificity dropped to 17%, albeit with perfect sensitivity (100%). Against the AL threshold in OBS-HATLV, EMG-DUET reached a sensitivity of 83% and specificity of 48% at its optimal threshold of 63%. View this table: View inline View popup Download powerpoint Table 4. Sensitivity and specificity of EMG-DUET and OBS-DUET at their optimal decision thresholds (expressed as %) against the two reference tools, EMG-Peak and the TLV and AL in OBS-HATLV. In comparison, OBS-DUET demonstrated superior performance relative to the TLV and AL thresholds in OBS-HATLV. At an optimal threshold of 52%, it achieved a perfect sensitivity of 100% and specificity of 79% against the TLV threshold, and at 45%, it showed 79% sensitivity and 94% specificity against the AL threshold. However, OBS-DUET’s sensitivity and specificity were less robust against EMG-Peak, peaking at 75% and 63%, respectively, at an optimal threshold of 46%. 3.3 Correlations with real-world prevalence of DUE symptoms When comparing the ranks of risk assessment outcomes to the ranks of real-world work-related symptom prevalence in the fingers, hands, and wrists across occupations, the tools demonstrated varying degrees of alignment. Although none of the correlations reached statistical significance, EMG-DUET showed a weak correlation (ρ [CIs] = 0.35 [-0.23, 0.78], p = 0.215; Figure 4a ), and EMG-Peak exhibited a moderate correlation (ρ = 0.46 [-0.16, 0.80], p = 0.100; Figure 4c ). In contrast, OBS-DUET (ρ = 0.00 [-0.61, 0.59], p = 0.994; Figure 4b ) and OBS-HATLV (ρ = 0.09 [-0.56, 0.67], p = 0.764; Figure 4d ) showed no meaningful correlation with prevalence data. Download figure Open in new tab Figure 4. Rank comparisons between aggregated occupation-level risk assessment outcomes from (a) EMG-DUET, (b) OBS-DUET, (c) EMG-Peak, and (d) OBS-HATLV, and the corresponding real-world prevalence of DUE symptoms in the fingers, hands, and wrists. Each point represents an occupation. Spearman’s correlation coefficients (ρ) are shown in each panel. 4 Discussion As a proof of concept, this study evaluated EMG-DUET by examining how its two core components—EMG-based inputs and the DUET framework—aligned with their respective reference tools (EMG-Peak and OBS-DUET), and how their combined implementation compared with a tool lacking both components (OBS-HATLV). Both DUET variants demonstrated the strongest agreement with non-DUET reference tools that use the same data type: EMG-DUET showed a moderate-to-strong correlation with EMG-Peak (r rm = 0.63, AUC = 0.94, sensitivity = 88%, specificity = 87%), while OBS-DUET was most consistent with OBS-HATLV (r rm = 0.72, AUC = 0.92, sensitivity = 100%, specificity = 79%). However, the correlation between EMG-based and observation-based tools was only moderate-to-weak (r rm s < 0.45), indicating that they capture overlapping yet distinct aspects of exposure. When considering clinical relevance, none of the tools demonstrated a statistically significant correlation with the occupational prevalence of DUE symptoms. Nonetheless, EMG-based measures showed stronger, though not significant, correlations (EMG-DUET: ρ = 0.35; EMG-Peak: ρ = 0.46) compared to observation-based tools (OBS-DUET: ρ = 0.00; OBS-HATLV: ρ = 0.09), further highlighting the unique information each tool provides. 4.1 Concordance of DUETs with Their Reference Tools EMG-DUET risk scores showed strong concordance with their EMG-based reference (EMG-Peak), demonstrated by a moderate-to-strong rank correlation (r rm = 0.63; Figure 2c ) and excellent classification performance (AUC = 0.94; sensitivity = 88%, specificity = 87% at optimal threshold; Figure 3a , Table 4 ). This close alignment is expected since both tools used the same primary input—normalized muscle activity (%MVE)—and applied similar monotonic transformations throughout their analytical pipelines. In parallel, OBS-DUET showed robust agreement with its observational benchmark, OBS-HATLV, with a strong rank correlation (r rm = 0.72; Figure 2f ) and high discrimination for both the TLV (AUC = 0.92) and AL (AUC = 0.91) thresholds ( Figure 3b 1,b2 ). This finding aligns well with prior studies, such as Jorgensen et al. (2024) , who reported a strong correlation (ρ = 0.92) between OBS-DUET and the Revised Strain Index, another established observational risk assessment tool. Despite these promising consistencies, both DUET variants exhibited weaker correlations with occupation-level prevalence of DUE symptoms compared to their respective reference tools ( Figure 4 ). This result contrasts to certain extent with prior reports, such as a study involving 293 workers across 216 jobs, which found that cumulative damage estimates from a similarly structured shoulder risk tool significantly predicted shoulder symptoms (Bani Hani et al., 2021 ). This suggests that while the DUET framework can capture clinically meaningful risk, its technical tuning and model calibration are crucial for optimal predictive performance. Examining observational tools more closely, both OBS-DUET and OBS-HATLV used exertion intensity and repetition frequency as inputs but integrated them differently. OBS-HATLV applied a straightforward linear combination, while OBS-DUET uses a non-linear framework to translate perceived exertion into cumulative tendon strain, assigning different weights to exertions across the entire intensity range. Although this non-linear approach theoretically provides a more nuanced and interpretable estimate of cumulative risk, the observed weaker correlation with occupational prevalence of DUE symptoms suggests that important predictive information may be obscured or distorted in the process. Overall, these findings indicate that the DUET framework is conceptually robust but demands further technical refinement, particularly regarding the EMG-based pipeline. The probabilistic structure of DUET, which allows for the integration of risk across varying task durations, rotations, and individual workloads, remains a notable advantage for practical and scalable ergonomic risk assessment ( Mehdizadeh et al., 2020 ). Future research should prioritize fine-tuning modeling parameters and performing rigorous validation in real-world occupational settings to fully realize DUET’s potential as an occupational health risk assessment tool. 4.2 Divergence among Exertion Exposures This study utilized two accepted reference risk assessment tools, EMG-Peak and OBS-HATLV, to benchmark task-level exposures of the wrist and hand. Although both tools are designed to estimate risk factors for DUE symptoms, this study showed that they produced only a near-weak rank correlation of the same tasks (r rm = 0.41; Figure 2b ). This discrepancy mirrors previous findings by Dahlqvist et al. (2024) , who reported only 26% absolute agreement between EMG- and observation-based risk assessment tools, underscoring that these modalities assess overlapping but distinct features of biomechanical exposure. Compared to observation-based tools, sEMG provides a direct measurement of electrical activity in the forearm muscles, capturing not only electrical signals from force-generating contractions but also information regarding low-level stabilizing activity and co-contraction, which are overlooked by observational methods ( Batista et al., 2024 ; Kimura et al., 2007 ). However, EMG recordings are susceptible to technical issues such as electrode placement, perspiration, and motion artefacts, which can introduce both physiological and measurement noise ( Abdoli-Eramaki et al., 2012 ; Clancy et al., 2002 ). In contrast, OBS-HATLV relies on visual estimates of exertion intensity and duty cycle, which introduces issues from human factors, including inter- and intra-rater variability and systematic bias influenced by factors such as rater experience and demographic characteristics ( Dahlgren et al., 2022 ; Nyman et al., 2023 ). These methodological differences likely contribute to the near-weak correlation observed between the DUET variants (EMG-DUET vs. OBS-DUET: r rm = 0.42; Figure 2a ). To assess practical relevance, the study compared aggregated occupation-level risk rankings from each metric to occupation-level prevalence rankings for DUE symptoms in the fingers, hands, and wrists. Despite a lack of statistically significant correlations, EMG-based measures generally outperformed observational tools: EMG-DUET (ρ = 0.35, p = 0.215) and EMG-Peak (ρ = 0.46, p = 0.100) showed weak-to-moderate correlation to the health endpoint, while OBS-DUET (ρ = 0.00, p = 0.994) and OBS-HATLV (ρ = 0.09, p = 0.764) exhibited negligible correlations ( Figure 4 ). While these findings align with certain previous study linking peak EMG load to discomfort and diagnosed DUE symptoms ( Nordander et al., 2013 ), other studies have raised important concerns. While OBS-HATLV scores have been strongly correlated with wrist velocity—a known kinematic risk factor for carpal tunnel syndrome—EMG-Peak has often shown limited correlation with DUE prevalence ( Dahlqvist et al., 2024 ; Nordander et al., 2013 ). Moreover, contrasting to the results from this study, OBS-HATLV has previously demonstrated reliability in predicting outcomes such as carpal tunnel syndrome ( Bonfiglioli et al., 2013 ; Yung et al., 2019 ). One explanation could be that EMG-based and observational risk assessment tools appear to capture complementary, though non-interchangeable, aspects of physical exposure. While OBS-HATLV is good at capturing certain key risk factors in hand intensive work such as fast hand movement, such factors may be overlooked by EMG-based methods. Alternatively, it could also be linked to this study’s approach—averaging and aggregating task-level risk estimates to approximate occupation-level outcomes—which may have inherent limitations. As such, definitive claims regarding the superiority of any single tool are premature, and integrating both approaches depending on the stage and scope of projects could theoretically provide a more comprehensive exposure profile. Future longitudinal studies employing continuous EMG and automated observational tracking, combined with prospective DUE symptom surveillance, are crucial for determining the optimal use and integration of these complementary assessment paradigms within ergonomic risk evaluation. One possible explanation is that EMG-based and observational risk assessment tools capture different aspects of physical exposure. For instance, OBS-HATLV could excel at identifying certain risk factors in hand-intensive work, such as rapid hand movements, which may escape detection by EMG-based approaches. On the other hand, the study’s approach—averaging and aggregating task-level risk estimates to represent occupation-level outcomes—may have also contributed to the observed discrepancies, as it potentially blurred distinctions among the variety of tasks within the same occupation and overlooked differences in rest patterns throughout the workday. Given these considerations, it is premature to assert the superiority of any single tool. Ultimately, longitudinal studies that integrate continuous EMG monitoring and observational tracking in parallel with prospective DUE symptom surveillance will be essential to establish best practices for combining, weighting, or selecting among these paradigms in ergonomic risk evaluation. 4.3 Practical Implications Modern workplaces consist of varied tasks, fluctuating workloads, and a diverse workforce. To effectively assess ergonomic risks in such environments, tools must be both flexible—capable of addressing exposures across different jobs, shifts, and individuals—and interpretable for practical decision-making. The DUET framework, in theory, offers practicality because it calculates cumulative damage at the task level, which can be integrated or weighted across rotations and teams depending on application ( Mehdizadeh et al., 2020 ). Importantly, DUET produces a probability of adverse outcomes, facilitating clear interpretation for stakeholders ( Gallagher et al., 2018 ; Mehdizadeh et al., 2020 ). However, our results highlight that DUET’s predictive value is highly dependent on the specification of each step within its analytical pipeline. A central issue is the definition and acquisition of “exertion”. OBS-DUET and OBS-HATLV rely on visual ratings of force and repetition—methods that, while cost-effective and unobtrusive, are inherently labor-intensive and susceptible to rater variability ( Dahlgren et al., 2022 ). Emerging technologies, such as computer-vision systems estimating exertion from posture or facial cues ( Asadi et al., 2020 ), promise to automate and standardize this process, though these approaches remain under development and are not yet widely adopted. In contrast, sEMG provides an objective, continuous measure of muscular loading. Advances in textile and high-density EMG sensors have improved wearability and signal quality ( Lam et al., 2022 ; Martinez et al., 2020 ), and even single-channel or through-forearm placements can yield actionable data for repetitive, hand-intensive tasks ( Fan et al., 2024 ). However, EMG is not without limitations. Commonly used electrodes can capture stabilizing muscle activity or artefacts (e.g., perspiration, motion, electrode displacement) that do not directly correlate to exerted force ( Abdoli-Eramaki et al., 2012 ; Kimura et al., 2007 ), while late technology is costly to implement. Given these complementary strengths and weaknesses, a hybrid assessment strategy— integrating physiological metrics from EMG with contextual observational data at different stages or levels in a project—may provide the most efficient, comprehensive and interpretable evaluation of ergonomic risk. Lastly, the observed modest and non-significant correlation between EMG-DUET and occupational prevalence of distal upper extremity (DUE) symptoms suggests that while the tool is conceptually promising, wide adoption of this tool in clinical settings is still early. Targeted improvements such as optimizing EMG signal processing, refining repetition identification, and enhancing the mapping of muscle activity to cumulative tendon damage are necessary. Furthermore, rigorous field studies that implement full-shift or longitudinal real-world assessments using EMG-DUET, paired with prospective monitoring of musculoskeletal disorder symptoms, are essential for robust validation. With these refinements, DUET—particularly in an EMG-augmented, hybrid format—could evolve into a scalable, evidence-based platform for proactive ergonomic risk assessment. 4.4 Technical consideration This study served as an exploratory evaluation of the EMG-DUET ergonomic risk assessment tool, which integrates EMG-based exertion estimates with the DUET framework. To systematically assess the contribution of each component, we explicitly separated two factors: (i) the impact of using EMG versus observational methods for estimating exertion, and (ii) the influence of the DUET framework compared to non-DUET approaches. This design enabled us to disentangle the effects of measurement modality and computational modeling on the resulting risk estimates. The selected set of tasks encompassed a wide range of occupational exposures, accounting for variation in physical demand, repetition, and posture. The sample size, task diversity, and range of job types are similar to or exceed those seen in prior risk-assessment studies ( Jorgensen et al., 2024 ; Mehdizadeh et al., 2020 ; Nail-Ulloa et al., 2025 ), allowing for robust observation of how distinct pipeline components perform under varied conditions and supporting identification of points in need of further refinement. Additionally, we benchmarked the outcomes of the four risk assessment tools against actual occupational prevalence data for DUE symptoms to evaluate their clinical relevance. By aggregating task-level risk estimates to the occupational level and correlating these with prevalence rankings, we obtained preliminary insights into the potential clinical utility of each tool. Yet, it is important to acknowledge that such aggregation relied on several simplifications—notably, extrapolating brief task recordings to represent daily workloads and oversimplifying the complexity and variety of real-world occupations. While these approximations are suitable for early-phase validation, comprehensive evaluation will ultimately require full-shift exposure data and prospective DUE symptom tracking. In summary, this structured evaluation highlights the DUET framework’s modularity and flexibility, as well as the essential need for targeted refinement and rigorous validation at each stage of the analytical pipeline. The findings provide a foundation for ongoing methodological improvements and offer a roadmap for progressing from exploratory validation to practical ergonomic risk assessment. 4.5 Limitations As an exploratory study, this work is subject to several limitations—both methodological and technical—that should be considered when interpreting the findings and planning future research. A primary limitation concerns the duration and scope of task sampling. Muscle activity (measured by EMG) and observed exertion levels were recorded during brief task segments, averaging 7.3 minutes each. Because full-day work schedules were not available—and the non-DUET benchmark tools require a complete workday for assessment—we projected each task’s risk estimate over a standardized six-hour work period to ensure comparability across all tools. While this approach does oversimplify the variability and temporal dynamics of real-world jobs, it preserves the rank ordering of task-level risks and allows for meaningful comparisons across job roles. Importantly, the selected tasks were representative of typical job activities for participants within their occupation, so the occupational risk profiles remain relevant and informative. For more accurate cumulative risk estimates and robust exposure-outcome validation, future studies should use full-workday recordings and incorporate real-time work-rest cycles. Second, the EMG-DUET’s technical parameters—such as EMG signal processing settings, rain-flow cycle definition, and the method for converting muscle contraction to tendon strain—were taken from previously published models rather than optimized for this integrated application ( Fan et al., 2024 ; Gallagher et al., 2018 ). These parameter choices have a direct and sometimes disproportionate impact on the final risk scores, especially within the DUET framework’s non-linear cumulative model. While using parameters from earlier studies maintains consistency with the previous studies and the other three reference tools, the lack of tuning for the current context may prevent the EMG-DUET from reaching its full potential. Moreover, certain approximations taken in this study, while convenient, may limit the generalization of the findings. For example, EMG-Peak was measured using through-forearm placement rather than the traditional extensor site ( Arvidsson et al., 2021 ), which may result in lower absolute thresholds and the identification of more high-risk cases, potentially altering observed correlations. All in all, future validation studies should therefore include sensitivity analyses for each parameter and empirical calibration against established biomechanical or epidemiological benchmarks. Third, the DUET framework itself relies on several assumptions that currently limit its generalizability. The fatigue-failure model within the framework uses a unidirectional mapping from exertion to damage, applying an exponential tendon-strain function derived from cadaveric data ( Gallagher et al., 2018 ; Schechtman & Bader, 2002 ). This approach simplifies the complexity of the musculoskeletal system, as it does not account for inter-muscle coordination or antagonistic co-activation—factors that can significantly influence the distribution of tissue loading and adaptation in real-world settings. Furthermore, recovery periods and rest breaks, which are known to impact injury risk, have not yet been incorporated into the model’s estimation of cumulative damage ( Koch et al., 2024 ). The DUET framework also appears to compress the variation of risk outcomes, especially in EMG-based applications (for example, reducing the EMG-Peak value range from 53 %MVE to approximately 22% in DUET). This compression suggests a tendency toward overestimation and information saturation, which may reduce score interpretability, tool sensitivity, and the ability to distinguish between tasks. Addressing these issues is crucial for the DUET framework’s future application and requires targeted refinement. Nevertheless, by strategically comparing the four tools—EMG-DUET, OBS-DUET, EMG-Peak, and OBS-HATLV—we were able to isolate the effects of EMG-based input and the DUET framework on risk outcomes, providing actionable insights for future model development. Future work should focus on refining model assumptions, optimizing signal processing pipelines, and validating the exposure-risk relationships in field settings with prospective health outcome tracking. 5 Conclusions This study assessed the EMG-DUET ergonomic risk assessment tool across three primary dimensions: (i) using EMG-based versus observation-based exertion inputs, (ii) comparing the DUET framework with non-DUET framework, and (iii) evaluating how each tool’s rank correlation with a clinical endpoint—the prevalence of DUE symptoms at the occupational level. Both DUET variants closely aligned with their corresponding reference tools with the same exertion inputs in the outcomes of risk assessments (EMG-DUET with EMG-Peak, and OBS-DUET with OBS-HATLV), but each showed a slightly weaker rank correlation with DUE symptom prevalence than their respective references. Of the four tools, EMG-based risk assessment methods demonstrated the strongest, though still moderate or weak, correlations with the clinical outcome, highlighting a potential benefit of EMG-based exertion estimation. The DUET framework remains advantageous for its probabilistic, task-flexible design. However, simplifications in the present EMG-DUET implementation appear to limit its predictive capacity. Targeted improvements—such as refining signal processing, cycle counting, and damage mapping—and thorough validation with full-shift exposure data and prospective health outcomes are essential to realize the full potential of EMG-enhanced DUET for occupational risk assessment. Data Availability All data produced in the present study are available upon reasonable request to the authors Acknowledgement We thank Eric Martini Linger for assistance with data collection and Professor Emeritus Sean Gallagher from Samuel Ginn College of Engineering at Auburn University for valuable scientific and technical guidance. We are also grateful to the participating companies and organizations for their support with recruitment and for providing measurement locations. Finally, we extend our sincere appreciation to all participants for their time and effort. Footnotes CRediT authorship contribution statement: Xuelong Fan: Conceptualization, Methodology, Formal analysis, Software, Data Curation, Investigation, Writing - Original Draft, Writing - Review & Editing, Visualization, Project administration; Johan Rydgård: Software, Data collection, Data Curation, Writing - Review & Editing; Pasan Hettiarachchi: Methodology, Writing - Review & Editing; Kristina Eliasson: Data collection, Writing - Review & Editing; Camilla Dahlqvist: Data collection, Writing - Review & Editing; Peter J. Johansson: Conceptualization, Methodology, Data collection, Resources, Project administration, Writing - Review & Editing, Supervision, Funding acquisition. Funding sources: This work was supported by AFA Insurance [Ref. No. 200070]. Declaration of competing interest: There are no conflicts of interest to declare. Declaration of generative AI and AI-assisted technologies in the writing process: During the preparation of this work the author(s) used ChatGPT and Copilot in order to improve the language and readability. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the publication. References ↵ Abdoli-Eramaki , M. , Damecour , C. , Christenson , J. , & Stevenson , J . ( 2012 ). The effect of perspiration on the sEMG amplitude and power spectrum . Journal of Electromyography and Kinesiology , 22 ( 6 ), 908 – 913 . doi: 10.1016/j.jelekin.2012.04.009 OpenUrl CrossRef PubMed ↵ Arvidsson , I. , Dahlqvist , C. , Enquist , H. , & Nordander , C . ( 2021 ). Action Levels for the Prevention of Work-Related Musculoskeletal Disorders in the Neck and Upper Extremities: A Proposal . Annals of Work Exposures and Health , 65 ( 7 ), 741 – 747 . doi: 10.1093/annweh/wxab012 OpenUrl CrossRef ↵ Asadi , H. , Zhou , G. , Lee , J. J. , Aggarwal , V. , & Yu , D . ( 2020 ). A computer vision approach for classifying isometric grip force exertion levels . Ergonomics , 63 ( 8 ), 1010 – 1026 . doi: 10.1080/00140139.2020.1745898 OpenUrl CrossRef PubMed ↵ Bakdash , J. Z. , & Marusich , L. R . ( 2017 ). Repeated Measures Correlation . Frontiers in Psychology , 8 , 456 . doi: 10.3389/fpsyg.2017.00456 OpenUrl CrossRef PubMed ↵ Balogh , I. , Arvidsson , I. , Björk , J. , Hansson , G.-Å. , Ohlsson , K. , Skerfving , S. , & Nordander , C . ( 2019 ). Work-related neck and upper limb disorders – quantitative exposure–response relationships adjusted for personal characteristics and psychosocial conditions . BMC Musculoskeletal Disorders , 20 ( 1 ), 139 . doi: 10.1186/s12891-019-2491-6 OpenUrl CrossRef PubMed ↵ Bani Hani , D. , Huangfu , R. , Sesek , R. , Schall , M. C. , Davis , G. A. , & Gallagher , S. ( 2021 ). Development and validation of a cumulative exposure shoulder risk assessment tool based on fatigue failure theory . Ergonomics , 64 ( 1 ), 39 – 54 . doi: 10.1080/00140139.2020.1811399 OpenUrl CrossRef PubMed ↵ Batista , G. D. A. , Beltrán , S. P. , Passos , M. H. P. D. , Calixtre , L. B. , Santos , L. R. D. H. , & De Araújo , R. C. ( 2024 ). Comparison of the Electromyography Activity during Exercises with Stable and Unstable Surfaces: A Systematic Review and Meta-Analysis . Sports , 12 ( 4 ), 111 . doi: 10.3390/sports12040111 OpenUrl CrossRef PubMed ↵ Bell , A. , Ward , P. , Tamal , Md . E. H. , & Killilea , M. ( 2019 ). Assessing recall bias and measurement error in high-frequency social data collection for human-environment research . Population and Environment , 40 ( 3 ), 325 – 345 . doi: 10.1007/s11111-019-0314-1 OpenUrl CrossRef PubMed ↵ Bolino , M. C. , Kelemen , T. K. , & Matthews , S. H . ( 2021 ). Working 9-to-5? A review of research on nonstandard work schedules . Journal of Organizational Behavior , 42 ( 2 ), 188 – 211 . doi: 10.1002/job.2440 OpenUrl CrossRef ↵ Bonfiglioli , R. , Mattioli , S. , Armstrong , T. J. , Graziosi , F. , Marinelli , F. , Farioli , A. , & Violante , F. S . ( 2013 ). Validation of the ACGIH TLV for hand activity level in the OCTOPUS cohort: A two-year longitudinal study of carpal tunnel syndrome . Scandinavian Journal of Work, Environment & Health , 39 ( 2 ), 155 – 163 . doi: 10.5271/sjweh.3312 OpenUrl CrossRef PubMed Web of Science ↵ Clancy , E. A. , Morin , E. L. , & Merletti , R . ( 2002 ). Sampling, noise-reduction and amplitude estimation issues in surface electromyography . Journal of Electromyography and Kinesiology , 12 ( 1 ), 1 – 16 . doi: 10.1016/S1050-6411(01)00033-5 OpenUrl CrossRef PubMed Web of Science ↵ Coombes , L. , Bristowe , K. , Ellis-Smith , C. , Aworinde , J. , Fraser , L. K. , Downing , J. , Bluebond-Langner , M. , Chambers , L. , Murtagh , F. E. M. , & Harding , R . ( 2021 ). Enhancing validity, reliability and participation in self-reported health outcome measurement for children and young people: A systematic review of recall period, response scale format, and administration modality . Quality of Life Research , 30 ( 7 ), 1803 – 1832 . doi: 10.1007/s11136-021-02814-4 OpenUrl CrossRef PubMed ↵ Dahlgren , G. , Liv , P. , Öhberg , F. , Slunga Järvholm , L. , Forsman , M. , & Rehn , B . ( 2022 ). Ratings of Hand Activity and Force Levels among Women and Men Who Perform Identical Hand-Intensive Work Tasks . International Journal of Environmental Research and Public Health , 19 ( 24 ), 16706 . doi: 10.3390/ijerph192416706 OpenUrl CrossRef ↵ Dahlqvist , C. , Arvidsson , I. , Löfqvist , L. , & Gremark Simonsen , J . ( 2024 ). Consistency between the ACGIH TLV for hand activity and proposed action levels for wrist velocity and forearm muscular load based on objective measurements: An example from the assembly industry . International Journal of Occupational Safety and Ergonomics , 30 ( 3 ), 927 – 935 . doi: 10.1080/10803548.2024.2367367 OpenUrl CrossRef ↵ Dong , Y. , Jiang , P. , Jin , X. , Jiang , N. , Huang , W. , Peng , Y. , Shen , Y. , He , L. , Forsman , M. , & Yang , L . ( 2022 ). Association between long-term static postures exposure and musculoskeletal disorders among university employees: A viewpoint of inflammatory pathways . Frontiers in Public Health , 10 , 1055374 . doi: 10.3389/fpubh.2022.1055374 OpenUrl CrossRef PubMed ↵ Fan , X. , Rydgård , J. , Yang , L. , & Johansson , P. J . ( 2024 ). Assessing Forearm Exertion in Manual Tasks with Surface EMG: A Comparative Analysis of Through-Forearm vs . Muscle Specific EMG placements. Bioengineering . doi: 10.1101/2024.12.10.627589 OpenUrl Abstract / FREE Full Text ↵ Fanchini , M. , Ferraresi , I. , Petruolo , A. , Azzalin , A. , Ghielmetti , R. , Schena , F. , & Impellizzeri , F. M . ( 2017 ). Is a retrospective RPE appropriate in soccer? Response shift and recall bias. Science and Medicine in Football , 1 ( 1 ), 53 – 59 . doi: 10.1080/02640414.2016.1231411 OpenUrl CrossRef ↵ Fawcett , T . ( 2006 ). An introduction to ROC analysis . Pattern Recognition Letters , 27 ( 8 ), 861 – 874 . doi: 10.1016/j.patrec.2005.10.010 OpenUrl CrossRef Web of Science ↵ Gallagher , S. , Schall , M. C. , Sesek , R. F. , & Huangfu , R . ( 2018 ). An Upper Extremity Risk Assessment Tool Based on Material Fatigue Failure Theory: The Distal Upper Extremity Tool (DUET) . Human Factors: The Journal of the Human Factors and Ergonomics Society , 60 ( 8 ), 1146 – 1162 . doi: 10.1177/0018720818789319 OpenUrl CrossRef ↵ Govaerts , R. , Tassignon , B. , Ghillebert , J. , Serrien , B. , Bock , S. D. , Ampe , T. , Makrini , I. E. , Vanderborght , B. , Meeusen , R. , & De Pauw , K. ( 2021 ). Prevalence and incidence of work-related musculoskeletal disorders in secondary industries of 21st century Europe: A systematic review and meta-analysis . BMC Musculoskeletal Disorders , 22 ( 1 ), 751 . doi: 10.1186/s12891-021-04615-9 OpenUrl CrossRef PubMed ↵ Graben , P. R. , Schall , M. C. , Gallagher , S. , Sesek , R. , & Acosta-Sojo , Y . ( 2022 ). Reliability Analysis of Observation-Based Exposure Assessment Tools for the Upper Extremities: A Systematic Review . International Journal of Environmental Research and Public Health , 19 ( 17 ), 10595 . doi: 10.3390/ijerph191710595 OpenUrl CrossRef ↵ Grant , S. , Aitchison , T. , Henderson , E. , Christie , J. , Zare , S. , Mc Murray , J. , & Dargie , H . ( 1999 ). A Comparison of the Reproducibility and the Sensitivity to Change of Visual Analogue Scales, Borg Scales, and Likert Scales in Normal Subjects During Submaximal Exercise . Chest , 116 ( 5 ), 1208 – 1217 . doi: 10.1378/chest.116.5.1208 OpenUrl CrossRef PubMed Web of Science ↵ Jorgensen , M. J. , Martinez , A. , & Hakansson , N. A . ( 2024 ). Comparison of multi-task ergonomic assessment methods for risk of upper extremity and low back musculoskeletal disorders . Applied Ergonomics , 119 , 104313 . doi: 10.1016/j.apergo.2024.104313 OpenUrl CrossRef PubMed ↵ Kapellusch , J. M. , Garg , A. , Hegmann , K. T. , Thiese , M. S. , & Malloy , E. J . ( 2014 ). The Strain Index and ACGIH TLV for HAL: Risk of Trigger Digit in the WISTAH Prospective Cohort . Human Factors: The Journal of the Human Factors and Ergonomics Society , 56 ( 1 ), 98 – 111 . doi: 10.1177/0018720813493115 OpenUrl CrossRef PubMed ↵ Keir , P. J. , Farias Zuniga , A. , Mulla , D. M. , & Somasundram , K. G . ( 2021 ). Relationships and Mechanisms Between Occupational Risk Factors and Distal Upper Extremity Disorders . Human Factors: The Journal of the Human Factors and Ergonomics Society , 63 ( 1 ), 5 – 31 . doi: 10.1177/0018720819860683 OpenUrl CrossRef ↵ Kimura , K. , Murakami , H. , Yamamoto , M. , Yokoyama , T. , Morita , T. , Ito , Y. , & Hiraba , K . ( 2007 ). Velocity-dependent EMG activity of masseter and sternocleidomastoideus muscles during a ballistic arm thrusting movement . Human Movement Science , 26 ( 1 ), 48 – 67 . doi: 10.1016/j.humov.2006.08.001 OpenUrl CrossRef PubMed ↵ Koch , M. , Forsman , M. , Enquist , H. , Baare Olsen , H. , Søgaard , K. , Sjøgaard , G. , Østensvik , T. , Nilsen , P. , Andersen , L. L. , Due Jacobsen , M. , Brandt , M. , Westgaard , R. , Mork , P. J. , Fan , X. , Wærsted , M. , & Veiersted , K. B . ( 2024 ). Frequency of breaks, amount of muscular rest, and sustained muscle activity related to neck pain in a pooled dataset . PLOS ONE , 19 ( 6 ), e0297859 . doi: 10.1371/journal.pone.0297859 OpenUrl CrossRef PubMed ↵ Lam , E. , Alizadeh-Meghrazi , M. , Schlums , A. , Eskandarian , L. , Mahnam , A. , Moineau , B. , & Popovic , M. R . ( 2022 ). Exploring textile-based electrode materials for electromyography smart garments . Journal of Rehabilitation and Assistive Technologies Engineering , 9 , 20556683211061995 . doi: 10.1177/20556683211061995 OpenUrl CrossRef ↵ Martinez , I. J. R. , Mannini , A. , Clemente , F. , Sabatini , A. M. , & Cipriani , C . ( 2020 ). Grasp force estimation from the transient EMG using high-density surface recordings . Journal of Neural Engineering , 17 ( 1 ), 016052 . doi: 10.1088/1741-2552/ab673f OpenUrl CrossRef PubMed ↵ McAtamney , L. , & Nigel Corlett , E . ( 1993 ). RULA: A survey method for the investigation of work-related upper limb disorders . Applied Ergonomics , 24 ( 2 ), 91 – 99 . doi: 10.1016/0003-6870(93)90080-S OpenUrl CrossRef PubMed Web of Science ↵ Mehdizadeh , A. , Vinel , A. , Hu , Q. , Schall , M. C. , Gallagher , S. , & Sesek , R. F . ( 2020 ). Job rotation and work-related musculoskeletal disorders: A fatigue-failure perspective . Ergonomics , 63 ( 4 ), 461 – 476 . doi: 10.1080/00140139.2020.1717644 OpenUrl CrossRef PubMed ↵ Nail-Ulloa , I. , Zabala , M. , Pool , N. , Sesek , R. , Thiese , M. , Sesek , R. , Schall , M. C. , & Gallagher , S . ( 2025 ). A fatigue failure framework for the assessment of highly variable low back loading using inertial motion capture – a case study . Ergonomics , 1–17. doi: 10.1080/00140139.2025.2460695 OpenUrl CrossRef ↵ Nordander , C. , Balogh , I. , Mathiassen , S. E. , Ohlsson , K. , Unge , J. , Skerfving , S. , & Hansson , G.- Å . ( 2004 ). Precision of measurements of physical workload during standardised manual handling . Part I: Surface electromyography of m. trapezius, m. infraspinatus and the forearm extensors. Journal of Electromyography and Kinesiology , 14 ( 4 ), 443 – 454 . doi: 10.1016/j.jelekin.2003.12.003 OpenUrl CrossRef PubMed Web of Science ↵ Nordander , C. , Ohlsson , K. , Åkesson , I. , Arvidsson , I. , Balogh , I. , Hansson , G.-Å. , Strömberg , U. , Rittner , R. , & Skerfving , S . ( 2013 ). Exposure–response relationships in work-related musculoskeletal disorders in elbows and hands – A synthesis of group-level data on exposure and response obtained using uniform methods of data collection . Applied Ergonomics , 44 ( 2 ), 241 – 253 . doi: 10.1016/j.apergo.2012.07.009 OpenUrl CrossRef PubMed ↵ Nyman , T. , Rhén , I.-M. , Johansson , P. J. , Eliasson , K. , Kjellberg , K. , Lindberg , P. , Fan , X. , & Forsman , M . ( 2023 ). Reliability and Validity of Six Selected Observational Methods for Risk Assessment of Hand Intensive and Repetitive Work . International Journal of Environmental Research and Public Health , 20 ( 8 ), 5505 . doi: 10.3390/ijerph20085505 OpenUrl CrossRef ↵ Ortega , D. G. , Housh , T. J. , Smith , R. W. , Arnett , J. E. , Neltner , T. J. , Schmidt , R. J. , & Johnson , G. O . ( 2024 ). The Effects of Anchoring a Fatiguing Forearm Flexion Task to a High vs . Low Rating of Perceived Exertion on Torque and Neuromuscular Responses. Journal of Strength & Conditioning Research , 38 ( 5 ), e219 – e225 . doi: 10.1519/JSC.0000000000004730 OpenUrl CrossRef PubMed Power , J. D. , Perruccio , A. V. , Paterson , J. M. , Canizares , M. , Veillette , C. , Coyte , P. C. , Badley , E. M. , Mahomed , N. N. , & Rampersaud , Y. R . ( 2022 ). Healthcare Utilization and Costs for Musculoskeletal Disorders in Ontario, Canada . The Journal of Rheumatology , 49 ( 7 ), 740 – 747 . doi: 10.3899/jrheum.210938 OpenUrl Abstract / FREE Full Text ↵ Radwin , R. G. , Azari , D. P. , Lindstrom , M. J. , Ulin , S. S. , Armstrong , T. J. , & Rempel , D . ( 2015 ). A frequency–duty cycle equation for the ACGIH hand activity level . Ergonomics , 58 ( 2 ), 173 – 183 . doi: 10.1080/00140139.2014.966154 OpenUrl CrossRef PubMed Ritsuno , Y. , Kawado , M. , Morita , M. , Yamada , H. , Kanaji , A. , Nakamura , M. , Matsumoto , M. , Hashimoto , S. , & Fujita , N . ( 2021 ). Impact of Musculoskeletal Disorders on Healthy Life Expectancy in Japan [Preprint] . In Review . doi: 10.21203/rs.3.rs-153067/v1 OpenUrl CrossRef ↵ Robertson , R. J. , Goss , F. L. , Rutkowski , J. , Lenz , B. , Dixon , C. , Timmer , J. , Frazee , K. , Dube , J. , & Andreacci , J . ( 2003 ). Concurrent Validation of the OMNI Perceived Exertion Scale for Resistance Exercise: Medicine & Science in Sports & Exercise , 35 ( 2 ), 333 – 341 . doi: 10.1249/01.MSS.0000048831.15016.2A OpenUrl CrossRef PubMed Web of Science ↵ Rydgård , J. ( 2025a ). Construct validity of OMNI Hand Effort Scale and Borg CR-10 for hand intensive work. Examining relationships between subjective ratings and objective hand force measures in adults [Student thesis] . In TRITA-CBH-GRU: Vol. Independent thesis Advanced level (degree of Master (Two Years)) . DiVA. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-365079 ↵ Rydgård , J. ( 2025b ). ErgoAnnotation (Version 4.4) [Blender 4.4.3]. Uppsala University . https://github.com/Rockfella/ErgoAnnotation ↵ Schechtman , H. , & Bader , D. L . ( 2002 ). Fatigue damage of human tendons . Journal of Biomechanics , 35 ( 3 ), 347 – 353 . doi: 10.1016/S0021-9290(01)00177-4 OpenUrl CrossRef PubMed Web of Science ↵ Schober , P. , Boer , C. , & Schwarte , L. A . ( 2018 ). Correlation Coefficients: Appropriate Use and Interpretation . Anesthesia & Analgesia , 126 ( 5 ), 1763 – 1768 . doi: 10.1213/ANE.0000000000002864 OpenUrl CrossRef PubMed ↵ Silver , N. C. , & Dunlap , W. P . ( 1987 ). Averaging Correlation Coefficients: Should Fisher’s z Transformation Be Used? Journal of Applied Psychology , 72 ( 1 ), 146 – 148 . OpenUrl CrossRef Web of Science ↵ Statistikdatabasen . ( 2014 ). Standard för svensk yrkesklassificering (SSYK 2012) [Dataset]. https://www.h5.scb.se/yreg/ssyk2012.asp ↵ Statistikdatabasen . ( 2023 ). Andel sysselsatta personer enligt undersökningen Arbetsorsakade besvär per vald besvärsfråga efter kön och yrke. Vartannat år 2018—2022 [Dataset]. https://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__AM__AM0502__AM0502A/BesvarSSYK/?rxid=b161242a-954b-4a2f-b54b-d176c5381269 Tuček , M. , & Vaněček , V . ( 2020 ). Musculoskeletal disorders and working risk factors . Central European Journal of Public Health , 28 ( Supplement ), S06 – S11 . doi: 10.21101/cejph.a6183 OpenUrl CrossRef PubMed ↵ Youden , W. J . ( 1950 ). Index for rating diagnostic tests . Cancer , 3 ( 1 ), 32 – 35 . doi: 10.1002/1097-0142(1950)3:1%253C32::AID-CNCR2820030106%253E3.0.CO;2-3 OpenUrl CrossRef PubMed Web of Science ↵ Yung , M. , Dale , A. M. , Kapellusch , J. , Bao , S. , Harris-Adamson , C. , Meyers , A. R. , Hegmann , K. T. , Rempel , D. , & Evanoff , B. A . ( 2019 ). Modeling the Effect of the 2018 Revised ACGIH ® Hand Activity Threshold Limit Value ® (TLV) at Reducing Risk for Carpal Tunnel Syndrome . Journal of Occupational and Environmental Hygiene , 16 ( 9 ), 628 – 633 . doi: 10.1080/15459624.2019.1640366 OpenUrl CrossRef PubMed View the discussion thread. Back to top Previous Next Posted October 07, 2025. Download PDF Data/Code Email Thank you for your interest in spreading the word about medRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Developing an EMG-Based Distal Upper Extremity Tool for Assessing Risk of Distal Upper Extremity Symptoms: A Proof-of-Concept Comparative Evaluation Message Subject (Your Name) has forwarded a page to you from medRxiv Message Body (Your Name) thought you would like to see this page from the medRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Developing an EMG-Based Distal Upper Extremity Tool for Assessing Risk of Distal Upper Extremity Symptoms: A Proof-of-Concept Comparative Evaluation Xuelong Fan , Johan Rydgård , Pasan Hettiarachchi , Kristina Eliasson , Camilla Dahlqvist , Peter J. Johansson medRxiv 2025.10.03.25337266; doi: https://doi.org/10.1101/2025.10.03.25337266 Share This Article: Copy Citation Tools Developing an EMG-Based Distal Upper Extremity Tool for Assessing Risk of Distal Upper Extremity Symptoms: A Proof-of-Concept Comparative Evaluation Xuelong Fan , Johan Rydgård , Pasan Hettiarachchi , Kristina Eliasson , Camilla Dahlqvist , Peter J. Johansson medRxiv 2025.10.03.25337266; doi: https://doi.org/10.1101/2025.10.03.25337266 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Occupational and Environmental Health Subject Areas All Articles Addiction Medicine (570) Allergy and Immunology (864) Anesthesia (302) Cardiovascular Medicine (4445) Dentistry and Oral Medicine (444) Dermatology (383) Emergency Medicine (609) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1515) Epidemiology (15236) Forensic Medicine (30) Gastroenterology (1127) Genetic and Genomic Medicine (6610) Geriatric Medicine (669) Health Economics (1000) Health Informatics (4549) Health Policy (1370) Health Systems and Quality Improvement (1613) Hematology (543) HIV/AIDS (1266) Infectious Diseases (except HIV/AIDS) (15926) Intensive Care and Critical Care Medicine (1104) Medical Education (623) Medical Ethics (147) Nephrology (668) Neurology (6613) Nursing (346) Nutrition (999) Obstetrics and Gynecology (1147) Occupational and Environmental Health (957) Oncology (3341) Ophthalmology (975) Orthopedics (369) Otolaryngology (420) Pain Medicine (436) Palliative Medicine (130) Pathology (665) Pediatrics (1694) Pharmacology and Therapeutics (693) Primary Care Research (714) Psychiatry and Clinical Psychology (5458) Public and Global Health (9244) Radiology and Imaging (2205) Rehabilitation Medicine and Physical Therapy (1370) Respiratory Medicine (1197) Rheumatology (596) Sexual and Reproductive Health (715) Sports Medicine (530) Surgery (713) Toxicology (99) Transplantation (289) Urology (265) (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'a02674890ee2e72e',t:'MTc3OTg5OTk2MQ=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();