FAST ERGO_X

Work 34 (2009) 133–148 133DOI 10.3233/WOR-2009-0912IOS Press

FAST ERGO X – A tool for ergonomicauditing and work-related musculoskeletaldisorders prevention

Isabel L. NunesUniversidade Nova Lisboa, Faculdade de Ciencias e Tecnologia, Departamento de Engenharia Mecanica eIndustrial; and Centro de Tecnologia e Sistemas, UNINOVA; Campus de Caparica, 2829-516 Caparica, PortugalTel.: +351 212 948 567; E-mail: [email protected]

Received 21 February 2008

Accepted 7 May 2009

Abstract. Work-related musculoskeletal disorders associated with repetitive and strenuous working conditions continue torepresent one of the biggest occupational problems in companies. Despite the variety of efforts to control them, includingengineering design changes, organizational modifications and working methods training programs, work-related musculoskeletaldisorders account for a huge amount of human suffering and economic costs to companies and to healthcare systems. This paperpresents an ergonomic analysis tool, FAST ERGO X, designed to support ergonomic auditing activities related with work-relatedmusculoskeletal disorders. This tool can be used to analyze workplaces regarding potential ergonomic risk factors. The FASTERGO X is a fuzzy expert system designed to help the identification, assessment and control of the risk factors present in the worksystem, due to lack of adequate ergonomics. Based on objective and subjective data, the system evaluates the risk factors that canlead to the development of work-related musculoskeletal disorders, and presents the findings resulting from such evaluation. Thesystem also presents recommendations to eliminate or at least reduce the risk factors present in the work situation under analysis.

Keywords: Ergonomic workstation analysis, prevention of ergonomic risk factors, expert systems, fuzzy logics

1. Introduction

Work-related musculoskeletal disorders (WRMD)are impairments of bodily structures such as muscles,joints, tendons, ligaments, nerves, bones and the lo-calized blood circulation system, caused or aggravatedprimarily by work itself or by the environment in whichwork is implemented.

The WRMD are a central concern in Europe, giv-en the increasingly large number of workers affected.WRMD are the main occupational disease category suf-fered by European workers and they are widespread inall activity sectors. According to the European Founda-tion for the Improvement of Living and Working Con-ditions more than one third of the European workerssuffer from WRMD [5]. Other factors contributing tothe relevance of the subject are the heavy economic

consequences resulting from the high WRMD preva-lence and the suffering they cause, often leading to per-manent, partial or total disability of the worker. Datafrom the Nordic countries and the Netherlands, esti-mate the costs related to WRMD at between 0.5 and2% of Gross Domestic Product. According to the samedata, the WRMD affect women more than men becauseof the type of work they perform [1].

The recognition that the work may adversely affecthealth is not new. Almost 300 years ago (in 1717) theItalian physician Bernardino Ramazzini, father of occu-pational medicine, acknowledged the relationship be-tween work and certain disorders of the musculoskele-tal system due to the performance of sudden and irreg-ular movements and the adoption of awkward postures.In old medical records is also possible to find refer-ences to a variety of injuries related to the execution

1051-9815/09/$17.00 2009 – IOS Press and the authors. All rights reserved

134 I.L. Nunes / FAST ERGO X – a tool for ergonomic auditing and WRMD prevention

of certain work. Such disorders assumed names re-lated with the professions where they mainly occurred(for instance “carpenter’s elbow”, “seamstress’ wrist”or “bricklayer’s shoulder”) [28].

Over the years much has been written about thesedisorders, their incidence and risk factors. See forinstance [2–4,6,8–10,14–16,28,29,33,35–37].

The strong correlation between the incidence ofWRMD and the exertions resulting from the workingconditions is well known, particularly considering thephysical risk factors associated with jobs (e.g., awk-ward postures, high repetition, excessive force, stat-ic work, cold or vibration). Work intensification andstress seem also to be factors that increasingly con-tribute to the onset of those disorders [5].

The Fourth European Working Conditions Surveydata revealed that organisational features such as jobrotation and team working are associated with the inci-dence of WRMD [26]. On the other hand, the same sur-vey states that a good level of job autonomy and controlover work, support from colleagues and superiors, op-portunities to learn new things and worker participationresult in lower levels of exposure to WRMD.

Along the years different ergonomic tools for assess-ing workstations in order to identify WRMD risk fac-tors have been developed by individuals and organiza-tions. Some examples are, for instance, OWAS [7] (andthe associated software WinOWAS [31]), RULA [11],Strain Index [12], NIOSH [13,34] or OCRA [24,25].Despite all the available knowledge there remains someuncertainty about the precise level of exposure to riskfactors that triggers WRMD. In addition there is sig-nificant variability of individual response to the riskfactors exposure. Aware that there was yet room foruse of alternative approaches and the development ofnew features, and recognizing the adequacy of applyingfuzzy expert systems for dealing with the uncertaintyand imprecision inherent in the factors considered in anergonomic analysis, the author developed a fuzzy ex-pert system model for workstation ergonomic analysis,named ERGO X, a first prototype [18,23] and then theFAST ERGO X. The ERGO X method of workstationergonomic analysis was subject to a Portuguese patentapplication [21].

FAST ERGO X is a fuzzy expert system designedto identify, evaluate and control the risk factors due toergonomic inadequacies existing in the work system.Based on objective and subjective data, the system eval-uates the risk factors present in workplaces that canlead to the development of WRMD, and presents thefindings of the evaluation. The system also presents

recommendations that users can follow to eliminate orat least reduce the risk factors present in the work situ-ation.

This paper contains 5 sections. Section 1 is this in-troduction, Section 2 introduces some basic conceptsabout Fuzzy Logics and Expert Systems, Section 3presents FAST ERGO X features, Section 4 demon-strates the use of the system on the analysis of a work-station; and Section 5 presents the Conclusions.

2. Fuzzy logic and expert systems

The development of ergonomic workstation analysistools is conditioned by the complexity, imprecision andsubjectivity that often characterizes the knowledge anddata used in the ergonomic analysis process. FuzzyLogic (Fuzzy Set Theory) provides appropriate logical-mathematical tools to deal with problems with suchcharacteristics [39]. On the other hand, Expert Systemsoffer support to experts and non-experts in dealing withcomplex and ill structured problems, such as human-centered systems [32].

2.1. Fuzzy Logic

Fuzzy Logic (FL) foundations were laid, in 1965,by Lotfi Zadeh with the formulation of Fuzzy Set The-ory [39]. FL provides a mathematical framework forthe systematic treatment of vagueness and imprecision.The subjective nature of human classification process-es renders classical (dichotomous) approaches almostuseless to deal with imprecise systems. So FL facili-tates the elicitation and encoding of uncertain knowl-edge. It provides a representation mechanism that im-proves the flexibility for dealing with imprecise data.The result is more robust tools that perform better for awider variety of conditions and users. From an encod-ing point of view, fuzzy sets support the representationof knowledge and its uncertainty as a unique entity.The resulting representation is very flexible, since it canbe easily coupled with non-fuzzy forms of knowledgerepresentation, and it can be manipulated by a varietyof evaluation methods [30].

A fuzzy set presents a boundary with a gradual con-tour (see Fig. 1), by contrast with classical sets, whichpresent a discrete border. Let U be the universe of dis-course and u a generic element of U, then U = {u}. Afuzzy set A, defined in U, is one set of the dual pairs:

A = {(u, µA(u))|u ∈ U}

I.L. Nunes / FAST ERGO X – a tool for ergonomic auditing and WRMD prevention 135

distance

Mem

bers

hip

Deg

near

very nearnot near

01 kmdistance

Mem

bers

hip

Deg

ree

near

very nearnot near

1

0 1 km

Fig. 1. Representation of 3 fuzzy sets related to the linguistic variabledistance.

where µA(u) is designated as membership function ormembership grade (also designated as degree of com-patibility or degree of truth) of u in A [39]. The mem-bership function associates to each element u, of U, areal number µA(u), in the interval [0,1]. Further, themembership function of a fuzzy set can be seen as apredicate, since µA(u1) indicates the degree to whichu1 has the property (feature) represented by the fuzzyset A. In this case a 0 (zero) membership degree to afuzzy set A means that it is totally false that u1 be-longs to A (or adheres to the concept represented byA), a membership of 1 (one) to a fuzzy set means thatit is totally true that u1 belongs to A, and intermediatemembership values represent partial belonging to A,i.e., it’s more or less true that u1 belongs to A.

Fuzzy sets admit a set of basic operations such asunion, intersection, complement, product, cartesianproduct, concentration and dilation [41].

Linguistic variables are an important concept in FLused for the approximate characterization of phenom-ena which are too complex or too ill-defined to besusceptible of description in precise terms. Linguisticvariables admit as values words or sentences of naturallanguage, which can be represented as fuzzy sets. Inhuman discourse, variables values are, normally, ex-pressed by words, not by numbers. Thus, one advan-tage of using linguistic variables is that one can dealdirectly with semantic concepts of imprecise nature,with a consistent mathematical formulation [40–43].

Consider an example regarding the distance of aplace. When one states “The school is near”, the wordnear is a linguistic value of the variable distance, i.e.,is the label of the fuzzy set near. The set of terms ofa linguistic variable is the collection of primary terms(e.g., near, far) and the ones obtained by using mod-ifiers [negation (not), qualifiers and quantifiers (few,

very, quite, more or less)] and connectives (and, or).Each modifier has a mathematical function associatedto it. For instance, considering the primary term near,the following terms can be generated:

very near = near2 ⇔ µA(u)2

not near = 1 − near ⇔ 1 − µA(u)

Illustrative membership functions of the three terms(near, very near, not near) are depicted in Fig. 1. Nat-urally the distance scale will depend on the context,for instance, if one considers to walk to the school, thedistance corresponding the point where “near” fuzzyset membership degree reaches 0 can be something like1 km; however if one considers that the travel is doneusing car or public transportations, then the distancecan be something like 10 km. In the example depictedin Fig. 1 it was assumed that the concept of “near” istotally true for a distance smaller than 200 m (member-ship degree of 1) and is false for a distance greater than1 km (membership degree of 0). Distances in between200 m and 1 km will be relatively near, but the adher-ence to the near concept will decrease as the distanceincreases, i.e., the membership degree will tend to 0.

Looking at the 3 concepts represented, it is possibleto observe that as the distance increases the “very near”set becomes false faster than the “near” set, and on theother hand the “not near” set becomes true.

FL allows the formalization of the human abilityto perform approximate reasoning, i.e., reasoning andjudging under uncertainty conditions. FL extends Clas-sical Logic to include a continuum of truth degrees(partially true states) between the true and the falsevalues.

Basically, FL has two principal components. Thefirst is, in effect, a translation system for representingthe meaning of propositions and other semantic enti-ties. The second component is an inferential system forobtaining answers to questions that relate to informa-tion resident in a knowledge base using, for instance,“IF THEN” rules [38]. Both components are used inthe FAST ERGO X.

2.2. Expert Systems

Expert Systems (ES) or Knowledge Based Systemsare computer programs that aim to achieve the samelevel of accuracy as human experts when dealing withcomplex and ill-structured specific domain problems(like the ergonomic analysis), so that they can be usedby non-experts to obtain answers, solve problems orget decision support in the domain [32]. The strength


Fig. 2. Typical architecture of an expert system.

of these systems lies in its ability to put expert knowl-edge to practical use when an expert is not available.ES make knowledge more widely available and helpovercome the problem of translating knowledge intopractical useful results.

ES architecture contains four basic components (a)a specialized knowledge base that stores the relevantknowledge about the expertise domain, (b) an inferenceengine, used to reason about some specific problems,for instance using production rules or multiple attributedecision making models, (c) a working memory, whichrecords facts about real world, and (d) an interface thatensures user-system interaction [32] (see Fig. 2).

A Fuzzy Expert System is an ES that uses Fuzzy Log-ic into its reasoning/ inference process and/or knowl-edge representation scheme.

3. FAST ERGO X

FAST ERGO X’s aim is to support and advise occu-pational safety and health professionals in the ergonom-ic analysis of workstations. It also supports, throughadvice, the decision on the corrective or preventive ac-tions to implement in order to minimize the risk factorsresponsible for the ergonomic inadequacies, helping topromote improvement of the overall ergonomic qual-ity of workplaces analyzed. As stated before, FASTERGO X is based on ERGO X’s model [20].

In broad terms, FAST ERGO X is a fuzzy expertsystem with architecture identical to the one shown onFig. 2. The working memory records, among other,

the collected objective and subjective data to evalu-ate. The knowledge base is organized in two domainmodules: posture and work-related musculoskeletaldisorders. The knowledge about those expertise do-mains was obtained using a knowledge elicitation pro-cess involving the available literature (17 referenceswere considered in this process) and subject matter ex-perts’ opinion, namely Occupational Physicians. Theaim and length of this paper is not compatible with amore thorough discussion about the knowledge elicita-tion process, namely the identification of the risk fac-tors contributing to WRMD, and the definition of theirweight for the evaluation process, however a detaileddiscussion about this process can be found in [19].

The inference engine performs the ergonomic work-station analysis based on the model depicted in Fig. 3.This model involves two processes, the evaluation pro-cess and the advisory process. The first, based on afuzzy multiple attribute decision-making methodolo-gy, evaluates the combined effect of the risk factorspresent on the workstation and provides the results inthe form of conclusions, and the corresponding ex-planations. The second selects the recommendations,which provide guidelines, leading to the identificationof corrective measures in order to minimize the risk fac-tors identified on the analyzed workstation. For moreinformation about the model refer to [18,22].

As discussed by Nunes in [17], the use of anergonomic analysis method based on Fuzzy Logicpresents some advantages over the classical ergonomicanalysis methods commonly used (based on discreteassessment criteria). Some of the benefits associated


Fig. 3. FAST ERGO X’s inference model. Objective data (measurements) collected in Record Sheets (recs) are converted, using inadequacymembership function (mfunc), in objective attribute inadequacy degrees (inato) and then aggregated into objective risk factor inadequacy degrees(inrfo). Similarly, subjective data (opinions) collected in Questionnaires (quest) are converted in subjective risk factor inadequacy degrees(inrfs), using linguistic variables (vling). The next step is to combine objective and subjective risk factor inadequacy degrees in risk factorinadequacy degrees (inrf). This risk factor inadequacy degrees are aggregated using weighting factors to evaluate cases inadequacy, which areconverted in conclusions using qualification linguistic variables (vling(qual)). Associated to conclusions are explanations about the results, andrecommendations.

with the use of fuzzy concepts are: (1) the use of con-tinuous membership functions, which are more realis-tic than discrete classifications, (2) enabling the aggre-gation and processing of objective and subjective da-ta in a consistent way, (3) the complexity of a fuzzyrules’ system is approximately constant regardless ofthe number of factors considered as criteria, (4) com-plex fuzzy knowledge bases can be built easily andevaluation of the data can use an extended variety offuzzy operators, allowing the analysis of complex situ-ations, (5) regardless if the results are presented in nu-

merical or linguistic format, the processing using fuzzymethods is numerical, allowing the combination of dif-ferent types of data in the classification and sorting ofdifferent situations under analysis.

FAST ERGO X evaluates the risk factors present inthe workplace based on objective and subjective data.The subjective data result from replies to questionnairesthat collect workers opinions about the perceived ad-equacy of the risk factors. The opinions are capturedusing discrete classifications based on terms of linguis-tic variables. On the other hand the evaluation of each


Fig. 4. Fuzzy set for assessing the elbow flexion posture inadequacy. The graph illustrates the assessment of the inadequacy of an elbow flexionangle of 115◦ , which is 0.25 i.e., the inadequacy is low, meaning that there is no significant risk associated with this posture.

risk factor based on objective data is done considering asubset of relevant attributes that characterizes the workactivity. For instance, when considering the risk factorposture regarding a wrist joint, some of the attributes toconsider are the angles of wrist flexion/extension, andradial/ulnar deviations. The attribute evaluation can bedone based on estimations or measurements performedeither directly (e.g., using goniometers) or indirectly,using the FAST ERGO X virtual goniometer associat-ed with the video analysis tool.

The objective attributes inadequacy evaluation is per-formed based on fuzzy sets. These fuzzy sets, charac-terized by continuous membership functions, create thecorrespondence between numerical values (e.g., pos-ture angles) and membership degrees in the interval[0, 1]. The definition of these membership functionswas based on the available literature and OccupationalPhysicians’ opinion [19]. Consider, for instance, theevaluation of the elbow posture inadequacy, which isperformed based on a value reflecting the elbow flexionangle that the worker assumes during the working ac-tivity. If the angle is 115◦ and the adopted “elbow flex-ion posture inadequacy” fuzzy set is the one presentedin Fig. 4, the resulting membership degree (i.e. the in-adequacy degree concerning such posture) is 0.25. Alow inadequacy degree (close to 0) means that postureis adequate while a high value (close to 1) means thatposture is inadequate.

For the analysis of dynamic activities the objectivedata can be collected using a time sampling of the task.This allows capturing the motions performed by theoperators along their working cycles. The time sam-pled data is processed using the Equivalent Mean Value(VME) function that computes an angle value whoseinadequacy degree is equivalent to the cumulative ef-fect of the different postures observed [22]. The VMEof the data sets collected for each body joint can be usedas an input in a specific record sheet to be processed bythe evaluation process shown in Fig. 3.

The system is modular. Currently two modules areunder use, the Posture module and the WRMD module.The first allows the analysis of workers’ posture, andthe second, the evaluation of the possibility of occur-rence of musculoskeletal disorders in a given workingsituation. The analysis is performed by body joint.

The Posture module restricts the analysis to posturedata. This module is used to produce an expeditiousindicator of the ergonomic quality of a job setup for agiven employee. Posture analysis is also useful sinceposture is one of the most important risk factors for thedevelopment of WRMD.

The WRMD module considers a more complete setof risk factors. The analysis is more time consumingbut the results are more reliable since, besides posture,there are other relevant factors (e.g., force, repetition,vibration) that contribute to the development of occu-pational diseases. Table 1 lists some of the risk fac-tors that the FAST ERGO X assesses in the WRMDevaluation process.

The FAST ERGO X software has the following fea-tures:

– supports the user to collect data – directing thecollection and the filling of the data, according tothe settings of analysis defined by the user andcharacteristics of the workstations and tasks underanalysis;

– performs the assessment of risk factors present onthe workplace – synthesises the elements of anal-ysis and presents the conclusions using differentgraphical or text formats;

– provides explanations about the results obtained inthe ergonomics analysis – allowing an easy identi-fication of individual risk factors that contributedto the result displayed;

– advises corrective or preventive measures to applyto the work situations – the knowledge base in-cludes a set of recommendations in HTML format,


Table 1WRMD assessed Risk Factors

Risk Factors

Physical Psychosocial IndividualPosture Work pace AgeRepetition Autonomy SexForce Monotony Professional ActivitiesDirect pressure Work/rest cycle Sport ActivitiesVibration Task demands Recreation ActivitiesCold Personnel inter relations with management Domestic Activities

Peer personnel inter relations Alcohol/TobaccoJob uncertainty Previous WRMD (same joint)

with hyperlinks that enable the navigation to a setof relevant topics related to the issues addressed(for example, risk factors, potential consequences,preventive measures or good practice references).

The use of FAST ERGO X comprehends three mainphases: analysis configuration, data collection and dataanalysis. These phases are depicted in Fig. 5.

The data collection phase requires the performanceof other related activities, like obtaining workers’ opin-ions using questionnaires or measuring work-relatedfeatures (e.g., durations, postural angles, number ofrepetitions). The acquisition of such measurements canbe supported by video recordings and by equipmentor software that assists the analysis. FAST ERGO Xprovides the Video Analysis Tool, which is one compo-nent that allows the visualization of two synchronizedvideos and the extraction, for instance, of multiple pos-tural angles.

Another characteristic of the FAST ERGO X is theportability of the system. The software can be used inlaptops offering the possibility of making in situ anal-yses, from the data collection phase until the presen-tation of the conclusions about the working situationunder analysis and the discussion of the applicable rec-ommendations.

The evaluation model used by FAST ERGO X wasapplied on the analysis of several working situations,from different activity sectors (e.g. industry and ser-vices) for validation and evaluation purposes. ThusFAST ERGO X results were compared with the resultsproduced by other ergonomic analysis methods, withavailable occupational health records and with the as-sessment of experts. General agreement or correla-tion between FAST ERGO X and other methods’ re-sults was verified. Where discrepancies were found itwas possible to justify them based on the evaluationmethods differences. The results regarding WRMDwere compared with medical records data, in order toverify whether there was correlation between the fore-

seen disorders and those diagnosed. Based on thesecomparisons it was found that all disorders diagnosedby physicians were considered relevant by the expertsystem model. For further details about the validationprocess refer to [20].

The data collection is performed by worker and bytask. Thus, results of the data analysis can reflect aglobal evaluation of all data available or be selected toaddress just a particular worker or a particular task.

The development of the FAST ERGO X softwarewas a project funded by the Portuguese Institute forSafety, Hygiene and Health at Work (currently WorkConditions Authority). The current available versionof the software is in Portuguese, and the English ver-sion is on the final stage of development. The screenspresented in this paper were captured from the Englishversion of FAST ERGO X.

On the next section a case study is presented demon-strating the main features of the FAST ERGO X.

4. Example of FAST ERGO X use

4.1. Characterization of the study

The present case study addresses the ergonomic anal-ysis of checkouts in a large supermarket. The super-market has a checkout line with 80 terminals. Each ter-minal includes one optical bar-code reader placed on afrontal position by the side of a keyboard that includesthe credit cards reader and a printer. The top cover ofthe values drawer is the desktop of the checkout ter-minal. The checkout has two side conveyor belts, oneto feed the terminal and the other to move the articlesaway to the packing zone. The terminals are pairedin sets of two using a layout where the operators areworking back to back. This way, half the terminals areright side fed while the other half are left side fed.

The number of checkout operators is around 250,which is approximately 30% of the supermarket work


Fig. 5. Activities performed on the analysis of a work situation.

force. The number of weekly working hours per opera-tor varies between 40 hours, for full-time workers, and14 to 30 hours, for part-time workers. The majority ofthe checkout operators (67.1%) work 30 or more hoursper week (45.7%–30 h and 21.4%–40 h). The averageworking time is 28.3 hours and the standard deviationis 8.1.

The age distribution is mainly concentrated in therange 20–29 years (79%). The average age is 23.3 years(minimum = 18 years; maximum = 43; standard de-viation = 4.1). The majority (84%) of operators arefemales.

Considering the operators’ time on job accordingto the number of years of service, the majority of theworkers (78.2%) worked less than three years in thecompany.

Despite the youth of workers and the reduced timeperforming the task, the Occupational Physician fre-quently receives complaints about wrist, shoulder andback pain on the checkout sector. Based on the occu-pational health records about 15% of the checkout op-

erators were affected by WRMD, mostly on the shoul-der (Shoulder Tendonitis) and wrist (Carpal TunnelSyndrome and De Quervain Disease), but also on theneck (Tension Neck Syndrome), back (Low Back Pain)and elbow (Epicondylitis). The information about theworkers’ complaints and the existence of already diag-nosed WRMD were the trigger for action.

The evaluation was performed using the FAST ER-GO X system, based on objective and subjective da-ta. The objective data was gathered by means of videorecordings and the subjective data was obtained fromthe workers’ replies to the FAST ERGO X question-naire.

The videos were used, for instance, to evaluate pos-tural angles, number of repetitions and cycle times.Due to operational reasons, the recordings were takenon one single checkout, but were repeated on differ-ent days and involved several workers. Two cameraswere used to record different views of the same activi-ty. Surveillance videos from the supermarket securityrecording system were also used in order to have accessto the top view of the checkout workstation.


Fig. 6. User interface for the configuration of a new analysis session.

4.2. Performing the analysis using FAST ERGO X

Using FAST ERGO X a new analysis session wasconfigured by introducing data about Company, work-places, tasks and workers involved on the study. It wasalso selected the type of analysis to perform. In thecurrent study both postural and WRMD analyses wereactivated. Figure 6 presents a view of the configurationenvironment interface where the analysis type and thebody joints to assess are selected.

After analysis configuration the next step was to col-lect the data. Postural data sets were collected fromthe videos using the Video Analysis Tool. The userinterface shown in Fig. 7 allows the analysis of videoframes and the collection of multiple postural data us-ing a virtual goniometer. The dual video mode illus-trated in the figure allows the analysis of synchronizedvideos, making possible, for instance, the collection ofdata from opposite side body parts in one single step.

The objective data used for the checkout analysis wasimported from the video analysis tool (the VME valuesthat synthesized the data sets) or inputted manually.Figure 8 illustrates the objective data collection andedition interface. This interface lists the complete setof attributes to fill according to the configuration ofthe analysis. It is possible to observe that software

immediately offers a preliminary evaluation of the dataof the selected risk factor (e.g., posture, repetition) forthe different body joints. Such assessment is presentedas colour coded areas over the joints of the mannequin.

Subjective data was collected from the operators’replies to the Questionnaires produced by the FASTERGO X according to the analysis configuration. Thereplies are inserted on the system using a user interfaceequivalent to the one shown in Fig. 8.

Once the input of data is finished the system is readyto start the evaluation and to present results. These re-sults can assume different formats, like lists or graphs.On the other hand it is possible to select if the evalua-tion is global (considering the data from all tasks andworkers analysed on the workstation) or if is focusedon a specific task or worker.

As shown on Fig. 9, one of the conclusions’ outputsis a list of Conclusions in the form of sentences like“The possibility for development of a WRMD on theRight Wrist is extreme (0,92)”. “Extreme” is a linguis-tic qualifier for the computed inadequacy degree thatis shown inside the parenthesis. The fuzzy inadequacyscale is the interval [0,1] where 0 means no inadequacy(i.e., absence of risk) and 1 means an extremely highpossibility of occurrence of WRMD on the body jointunder consideration (if the conclusions refer to the WR-


Fig. 7. FAST ERGO X Analysis Video Tool user interface. The tool allows the independent or synchronised analysis of up to two videos. Thedots connected by dotted lines over the left movie are used to capture the joint posture angles.

Fig. 8. User interface for objective data collection. The input data are the objective attributes (measurements), which are organized by RiskFactor. Using a colour code on the mannequin a preliminary evaluation of the selected risk factor data is presented for the different body joints.


Fig. 9. Presentation of the conclusions and their explanations in a text style.

MD Module). Considering the Posture module Con-clusions an inadequacy evaluation of 0 means a neutralposture, while 1 means an extremely bad posture.

The Conclusions can be explained by presenting thecomputed risk factors inadequacy degrees that con-tributed to the overall result. As for the Conclusionsthe Explanations are presented as sentences like “Theweighted inadequacy of the number of Repetitions per-formed by the Right Wrist is very high”. “Very high”is the linguistic qualifier for the computed inadequa-cy degree regarding the repetition risk factor. BothConclusions and Explanations have associated a colourcoded bar, on the left, that highlights the severity of thesituation.

Note that the fuzzy aggregation operators used onthe evaluation process reflect the synergistic effect ob-served when several risk factors are combined. This isthe reason why most of the conclusions present a levelof inadequacy higher than the individual risk factorsthat have been aggregated.

Another way of presenting the results is as a hori-zontal bar graph. The same colour codes referred tobefore are used here to code the level of possibility fordevelopment of WRMD. Figure 10 shows this type ofinformation output.

Finally, the results can be presented as a bar graph,but this time depicting the risk factors’ inadequacy de-grees for each upper body joint (i.e., the Explanation ofa Conclusion). The different graphs can be selected byclicking on the mannequin at the left side of the graph.As for the other graphical representations, colour codesare used. In this case the colours reflect the inade-quacy degree of the individual risk factors. Figure 11illustrates this type of output.

From the list of Conclusions it is possible to invokethe Recommendations regarding specific risk factors.This will open a browser like interface where the usercan look for further information on the issue and foradvice on typical preventive and corrective actions andbest practice references. Figure 12 shows the interfaceused in this feature.

The FAST ERGO X also produces reports and othertype of outputs that include, for instance, different typesof graphs. For instance, postural data collected withthe Video Analysis Tool can be presented as joint anglevs. time graphs. Figure 13 shows the interface usedto present such graphs. On the mannequin, at the rightside, the user selects a joint and the corresponding graphis depicted, at the left side. The graph can then be savedas a picture for future use.


Fig. 10. Presentation of the conclusions in a graphical format.

Fig. 11. Presentation of the explanation of a conclusion in a graphical style.


Fig. 12. Recommendations presented in a browser like environment.

Fig. 13. Left elbow flexion angle vs. time.


Table 2FAST ERGO X results vs. reported WRMD

ResultsGlobal assessment

Main Risk FactorsReported WRMD

Joint Posture Repetition Force Contact pressure

Neck Very high (0.74) High (0.66) High (0.56) − − Tension Neck SyndromeTrunk Very high (0.72) Medium (0.21) High (0.61) High (0.63) − Low Back Pain

ShoulderRight Very high (0.71) Medium (0.23) High (0.61) High (0.47) −

Shoulder TendonitisLeft High (0.62) Medium (0.23) High (0.59) High (0.47) −Elbow

Right Very high (0.68) Medium (0.28) High (0.61) Medium (0.31) High (0.50) Lateral and Medial Epic-Left Very high (0.70) Medium (0.28) High (0.61) Medium (0.31) High (0.50) ondylitis

Wrist/HandRight Extreme (0.92) Medium (0.46) Very High (0.88) Medium (0.56) Medium (0.17) Carpal Tunnel SyndromeLeft Very high (0.86) Medium (0.28) Very High (0.81) High (0.47) Medium (0.25) & De Quervain Disease

4.3. Results discussion

As shown above, FAST ERGO X evaluation pro-duced results regarding the degree of possibility of de-velopment of WRMD on the upper body joints andabout the main contributing risk factors. These resultsare compiled in Table 2. The results are presented bothquantitatively (as membership degrees to “inadequacy”fuzzy set, defined in the interval [0, 1]) and qualitative-ly (as terms of a linguistic variable “intensity”). Thetable also specifies the WRMD already identified in theOccupational Health records.

It is possible to verify that the results produced byFAST ERGO X are consistent with the OccupationalHealth records. According to the company’s Depart-ment of Occupational Medicine estimation, about 15%of the approximately 250 checkout operators are af-fected by musculoskeletal disorders at some point oftheir activity. The most frequent injuries, which occurin equal proportion, are the shoulder (Shoulder Ten-donitis) and wrist (Carpal Tunnel Syndrome and DeQuervain Disease). With lower incidence, and sortedby decreasing number of occurrences, there are cas-es of Tension Neck Syndrome (neck), Low Back Pain(trunk) and Lateral and Medial Epicondylitis (elbow).They are also in line with the ergonomic experts’ ob-servations, regarding the occupational risks present onthe checkout workstations.

The data presented in Table 2 demonstrate that theexpert system positively identified the body parts thathave potential to develop disorders. In fact FAST ER-GO X concluded there was an extreme to very highdegree of possibility of development of WRMD on allthe body joints where there is already some history ofexistence of occupational disorders.

The activity has a high level of repetition, and thejoint movements are combined with forceful exertions,since the checkout operators have to lift the productsand slide or suspend them over the scanner. This ma-terial handling imposes high solicitations to the hands,

mainly the one that is on the incoming conveyor side,since the operators perform very frequent grip and graspmovements. The effort to handle the loads is also prop-agated to the shoulders and trunk, which are also sub-ject to frequent motions, since the operators have toreach the products and pass them to the opposite con-veyor. The neck is flexed most of the time causingdiscomfort and ultimately tension and disorders.

The operators alternate the work between terminalsthat are right side fed and left side fed. This organi-zational measure reduces the incidence of injuries onone specific body side. Probably this also is why thereis no significant difference on the results obtained foreach side of the operator’s upper limbs.

It is not demonstrated in this paper but, as was men-tioned, FAST ERGO X can collect data per worker andper task. In this specific case study 3 workers wereanalyzed, representing the distribution of characteris-tics of the checkout operators’ population, in terms ofage, time on activity, gender, and anthropometric char-acteristics. All were young individuals (aged around23 years), working for about 2 years on the job; 2 werefemale and 1 was male. The workers were selectedalso considering their body size, so that they were dis-tributed among different percentiles (one woman wassmall, the other had an average size, and the man wastaller than normal). The data presented above reflectsthe global analysis. However, considering the analysisof the data performed regarding individual workers itis possible to confirm that the system is sensitive tothe variation of workers’ characteristics, and workingpostures and habits of the operators. For instance, thetaller operator presented more severe possibilities ofdeveloping WRMD on low back and neck, which canbe explained by the fact that he works with the trunkand the neck more flexed that the other workers ana-lyzed. On the hand the smaller worker had higher pos-sibility of developing WRMD on shoulders, which canbe explained by the fact that she works with the armsmore raised than the other workers.


The experience of using FAST ERGO X shows thatthe systems signals the possibility of occurrence of dis-orders on body joints where there is yet no history ofidentified disorders. This is normal because WRMDhave different latency times, meaning that the time re-quired for occupational diseases to reveal their symp-toms varies. The results presented by the FAST ER-GO X suggest that, if the operators work under thesame conditions for a significant period of time, someWRMD may appear. This was already confirmed bysubsequent queries to occupational health departmentsregarding workplaces that were evaluated using FASTERGO X but where no ergonomic intervention wasperformed. Notice that, in fact, this forecast capabilityis one of the main merits of the evaluation model, sinceit allows the system to evaluate workstations in an earlyphase prior to the development of WRMD. This capa-bility is very important for WRMD prevention, since itcreates the opportunity to act on the identified risk fac-tors in a workstation, avoiding the WRMD associatedcosts and suffering.

5. Conclusions

FAST ERGO X is a fuzzy expert system whose aimis to assist Occupational Health and Safety profession-als in the identification, assessment and control of er-gonomic risks related with the development of WRMD.

FAST ERGO X application was based on the ER-GO X model. This model was developed by the au-thor using Fuzzy Logics. This is an innovative ap-proach that uses Artificial Intelligence concepts. Thisapproach presents some advantages over the classicalmethods commonly used.

This paper presents the use of the FAST ERGO X onthe process of ergonomic analysis of checkouts. Themain features and characteristics of the software wereintroduced while the required analysis steps were de-scribed. In the discussion it was shown that the re-sults produced by the system were valid and consistentwith the existent Occupational Health records regard-ing WRMD and with experts’ analysis.

The use of the software is very flexible. On one hand,because it allows the use of objective and subjectivedata, separately or combined;on the other hand becauseit can be used on portable computers, which makesits utilization possible in situ either to collect data, topresent the results and to support any decision-makingthat may be required, for instance due to the need ofcorrective interventions.

The forecast capability of the evaluation model al-lows the use of the system as a WRMD prevention toolcreating the opportunity to act on identified risk factors,avoiding the WRMD associated costs and pains.

Finally, FAST ERGO X can also be used as a toolto promote participatory ergonomics. For instance, thesoftware and the media used for the analysis of the worksituations (e.g., video recordings) can be used to sup-port the training of workers in the field of OccupationalSafety and Health. This can be achieved either by us-ing the knowledge repository compiled on the knowl-edge base, by discussing the results of analyses carriedout, or by proceeding to critical reviews of the videoscollected for the analysis of work situations. Workers’awareness is a key success factor for the reduction ofpotentially risky behaviors, the identification of inade-quate situations, and the development of solutions thathelp the prevention of WRMD.

Acknowledgments

The FAST ERGO X was developed at the Er-gonomics Laboratory, Faculty of Science and Technol-ogy of the New University of Lisbon, within a Projectfunded by the Portuguese Institute for Safety, Hygieneand Health at Work, currently Work Conditions Au-thority.

References

[1] P. Buckle and J. Devereux, Work-related Neck and UpperLimb Musculoskeletal Disorders, European Agency for Safetyand Health at Work-2000, 1999.

[2] M. Christmansson, Repetitive and Manual Jobs – Content andeffects in terms of physical stress and work-related muscu-loskeletal disorders, The International Journal of Human Fac-tors in Manufacturing 4(3) (1994), 281–292.

[3] D. Colombini and E. Occhipinti, Preventing upper limb work-related musculoskeletal disorders (UL-WMSDS): New ap-proaches in job (re)design and current trends in standardiza-tion, Applied Ergonomics 37(4) (2006), 441–450.

[4] EASHW, Work-related musculoskeletal disorders: Preventionreport, European Agency for Safety and Health at Work, 2008

[5] EFILWC, Musculoskeletal disorders and organisationalchange, European Foundation for the Improvement of Livingand Working Conditions, Lisbon, 2007.

[6] M. Hagberg, B. Silverstein, R. Wells et al., Les lesions at-tribuables au travail repetitif: ouvrage de reference sur leslesions musculo-squelettiques liees au travail, Editions Multi-Mondes, 1995.

[7] O. Karhu, P. Kansi and I. Kuorinka, Correcting working pos-tures in industry: a practical method for analysis, AppliedErgonomics 8 (1977), 199–201.


[8] A. Kilbom, Repetitive work of the upper extremity: Part I –Guidelines for the practitioner, International Journal of In-dustrial Ergonomics 14 (1994), 51–57.

[9] A. Kilbom, Repetitive work of the upper extremity: Part II -The scientific basis (Knowledge base) for the guide, Interna-tional Journal of Industrial Ergonomics 14 (1994), 59–86.

[10] R. Marklin, Biomechanical Aspects of CTDs, in: The Occupa-tional Ergonomics Handbook, W. Karwowski and W. Marras,eds, CRC Press, 1999, pp. 795–832.

[11] L. McAtamney and E.N. Corlett, RULA: a survey method forthe investigation of work-related upper limb disorders, AppliedErgonomics 24 (2) (1993), 91–99.

[12] J.S. Moore and A. Garg, The Strain Index: A proposed methodto analyze jobs for risk of distal upper extremity disorders,American Industrial Hygiene Association 56 (1995), 443–458.

[13] NIOSH, Applications Manual for the Revised NIOSH LiftingEquation. NIOSH Publication No. 94–110. (http://www.cdc.gov/niosh/docs/94-110/), 1994.

[14] M. Nordin, M. Pope and G. Andersson, Musculoskeletal Dis-orders in the Workplace: Principles and Practice, (2nd ed.),Elsevier Health Sciences, 9780323026222, 2006.

[15] NRC and IOM, Musculoskeletal Disorders and the Workplace.Low Back and Upper Extremities, National Research Coun-cil/Institute of Medicine//National Academy Press, 2001.

[16] NRC, National Research Council/National Academy Press,1999.

[17] I.L. Nunes and R.A. Ribeiro, The ERGO X workstation anal-ysis compared with other ergonomics tools, XIV TriennialCongress of the International Ergonomics Association: Er-gonomics for the New Millennium, San Diego-USA, (2000),533–536.

[18] I.L. Nunes, ERGO X – The Model of a Fuzzy Expert Systemfor Workstation Ergonomic Analysis, in: International Ency-clopedia of Ergonomics and Human Factors, W. Karwowski,ed., CRC Press, 2006, pp. 3114–3121.

[19] I.L. Nunes, Knowledge Acquisition for the Development of anUpper – Body Work-Related Musculoskeletal Disorders Anal-ysis Tool, Human Factors and Ergonomics in Manufacturing17(2) (2007), 149–162.

[20] I.L. Nunes, Modelo de Sistema Pericial Difuso para Apoio?Analise Ergonomica de Postos de Trabalho [Fuzzy ExpertSystem Model to Support Workstation Ergonomic Analysis],PhD Dissertation, Universidade Nova de Lisboa (Lisbon, Por-tugal), 2003.

[21] I.L. Nunes, National patent n◦103446: Workstation Ergonom-ic Analysis Method – pending, 2006.

[22] I.L. Nunes, Quantitative Method for Processing Objective Da-ta from Posture Analysis, in: International Encyclopedia ofErgonomics and Human Factors, W. Karwowski, ed., CRCPress, 2006, pp. 3306–3309.

[23] I.L. Nunes, R.A. Ribeiro and V.A. Machado, ERGO X – Afuzzy expert system for the design of workstations, First WorldCongress on Ergonomics for Global Quality and Productivity(ERGON-AXIA ’98), Hong Kong – China, (1998), 232–235.

[24] E. Occhipinti and D. Colombini, Updating reference valuesand predictive models of the OCRA method in the risk assess-ment of work-related musculoskeletal disorders of the upperlimbs, Ergonomics 50(11) (2007), 1727–1739.

[25] E. Occhipinti, OCRA: a concise index for the assessmentof exposure to repetitive movements of the upper limbs, Er-gonomics 41(9) (1998), 1290–1311.

[26] A. Parent-Thirion, H.F. Macıas, J. Hurley et al., Fourth Eu-ropean Working Conditions Survey, European Foundation forthe Improvement of Living and Working Conditions, Luxem-bourg, 2007.

[27] A. Pethick, M. Mabey and R. Graves, Development of a prac-tical method for workplace redesign to reduce upper limbstrain injury, Musculoskeletal Disorders at Work, Univ Surrey,Guildford (1987), 239–246.

[28] V. Putz-Anderson, Cumulative Trauma Disorders: A Manualfor Musculoskeletal Diseases of the Upper Limbs, Taylor &Francis, 1988.

[29] M.J. Sanders, Ergonomics and the Management of Muscu-loskeletal Disorders, Elsevier Health Sciences, 0750674091,2004.

[30] C. Santamarina and G. Salvendy, Fuzzy sets based knowledgesystems and knowledge elicitation, Behaviour and Informa-tion Technology 10(1) (1991), 23–40.

[31] I. Tiilikainen, WinOWAS – software for OWAS analysis. Tam-pere University of Technology, Occupational Safety Engineer-ing. (http://turva1.me.tut.fi/owas/), 1996.

[32] E. Turban, J. Aronson and T.-P. Liang, Decision Support Sys-tems and Intelligent Systems, 7 ed., 2005.

[33] F. Violante, T.J. Armstrong and A. Kilbom, Occupational Er-gonomics: Work Related Musculoskeletal Disorders of theUpper Limb and Back, CRC Press, 978-0-7484-0933-4, 2000.

[34] T. Waters, V. Putz-Anderson, A. Garg et al., Revised NIOSHequation for the design and evaluation of manual lifting tasks,Ergonomics 36 (1993), 749–776.

[35] R. Wells, S.E. Mathiassen, L. Medbo et al., Time – A keyissue for musculoskeletal health and manufacturing, AppliedErgonomics 38 (6) (2007), 733–744.

[36] J. Winkel and R.H. Westgaard, Occupational and individu-al risk factors for shoulder-neck complaints: Part I – Guide-lines for the practitioner, International Journal of IndustrialErgonomics 10 (1992), 79–83.

[37] J. Winkel and R.H. Westgaard, Occupational and individu-al risk factors for shoulder-neck complaints: Part II – Thescientific basis (literature review) for the guide, InternationalJournal of Industrial Ergonomics 10 (1992), 85–104.

[38] L.A. Zadeh, Commonsense knowledge representation basedon fuzzy logic, Computer (1983), Out: 61–65.

[39] L.A. Zadeh, Fuzzy sets, Information and Control 8 (1965),338–353.

[40] L.A. Zadeh, Linguistic variables, approximate reasoning anddispositions, Med Inform 8(3) (1983), 173–186.

[41] L.A. Zadeh, The concept of a linguistic variable and its appli-cation to approximate reasoning-part I, Information Sciences8 (1975), 199–249.

[42] L.A. Zadeh, The concept of a linguistic variable and its appli-cation to approximate reasoning-part II, Information Sciences8 (1975), 301–357.

[43] L.A. Zadeh, The concept of a linguistic variable and its appli-cation to approximate reasoning-part III, Information Sciences9 (1975), 43–80.

Copyright of Work is the property of IOS Press and its content may not be copied or emailed to multiple sites or

posted to a listserv without the copyright holder's express written permission. However, users may print,

download, or email articles for individual use.

Copyright of Work is the property of IOS Press and its content may not be copied or emailed to multiple sites or

posted to a listserv without the copyright holder's express written permission. However, users may print,

download, or email articles for individual use.

FAST ERGO_X

Documents

carpal tunnel

de quervain

express written

continuous

elsevier health

identied risk

risk factors

occupational