Physical strain of work-related and activity-related tasks

Prosthetic arm industries currently appear to be nowhere near providing prosthetic arms that are really up for a physically demanding job or task. So if you are up to do real work, read on. Do not buy a prosthetic arm just yet though, save those dollars for just a moment.

• This must be all new from a manufacturing prosthetic hands view. I would agree to the hypothetical (?) argument that manufacturers are somewhat left alone there, because, no one there to lead the way, because how should they ever know what hazardous work looks like. But they should at least start to consider user loading requirements by taking the CE marking requirements to heart [link]. And if ever they had in mind to build something for real work and real usage, these prosthetic arms would also have to be super comfortable, affordable and really long-lasting. If you ever thought that any manufacturer has all the parts to really get going? Keep dreaming.
• Acting as if, showing off, is not the same as actual exposure for days, weeks, months. It has become clear that manufacturers at least toy with the idea of posing, pretending, acting as if, showcasing or acting out “heavy work” – whereas in fact, the currently sold parts, mounts, or devices may not even be allowed for any or such, or at least we know these parts will not work (such as promoting serious sportive bicycle riding with a myoelectric arm: sweat will in all likelihood cancel the fun that can be had).The industries at least seem to wish they ever had some at least remotely useful devices to offer to the demanding user. It is just that simply adding black rubber or menacing colors, or calling plastic crap “Hero” or “Task” or “Duty” or “Iron Man” or “Bionic Man” or “Beserk” does not make a fragile hand or arm actually robust, it does not make a bad suspension better, and it does not make unreliable controls better.
• A purpose in the context of a prosthetic arm is synonymous to the medically prescribed purpose or similar to the planned use. So when my prosthesis is prescribed to support my work, my line of work, the tasks that I do at work, that is what it is going to be used for. Swiss Medical Product law has one designated goal, and it is not the protection of manufacturers. It is the protection of users, or the patients. There, I am not even free to not use the prosthesis or use it for something else – I am mandated by medical product law to use the device for its (medically prescribed, intended) purpose. Even more so as insurance regulations require me to do the same: they pay for my prosthesis so they do not have to pay for getting me re-trained for another job or, worse, to pay me a monthly pension because I cannot work. So that purpose is against which the prosthesis is measured against. With that, I am not free to do whatever I want. So in order to get a prosthesis to perform for any given possible purpose, all these different medically prescribed purposes that there are or may be, have to be addressed by proper device documentation. That is best started by providing physical aspects of application domains (check this page you’re on, for example). Now, EU regulation 2017/745 goes relatively far in requiring any proper medical device to be documented well, also and particularly with regard to specific capabilities and loads. That is a key requirement to allow any prosthesis prescribing doctor, or prosthesis assembling prosthetist, to match prescribed purpose with correctly chosen prosthetic components. For example, one dimension of any work or task will be, “The person may sweat during any moment where device function may be require or relied upon”. If the answer there is “yes”, then the user either has to sign a form that states that they confirm any damages or problems arising from such goes on their own responsibility, or, the prescribing doctor or assembling prosthetist will pick a body-powered prosthesis setup over a myoelectric device. Conversely, any myoelectric device and any actively powered component will necessarily have to carry a sentence such as “Device or component may be expected to fail to function when user is sweating, depending on circumstances and specific parts used” in their technical documentation, prosthetist manual and user manual. As we currently do not see such declarations, I feel rather confident that the conformance of such devices with CE regulations is not at all a given. It starts, however, by detailing use-specific and build-specific task and job details.
• Conform to CE marking first, understand and conform to real use load numbers second. As a first next step, manufacturers will probably want to up their game, like, with regard to CE marking. There, they will want to declare, for both promotional materials, and documentation (to conform with CE requirements), what the actual loading limits of their parts really are, because if they do not declare these, we … (seriously, are you that lazy? just read the CE requirements yourself will you [link]). Manufacturers sure soon will really want to put total tonnage, total torque, safety limits, application related specifications, etc., on their product sheets, and with that, they will want to avoid any promotion, demonstration, or advertising that shows, showcases hints at, or promotes other/unacceptable use, or that shows unrealistic use scenarios (such as using an iLimb for hammering on nails) [link] (or using an iLimb with defective glove cover) [link].
• The rest of us knew all along that the real figures about hazardous work or other real work like activities were not really a mystery. Having actual numbers, figures, to describe workloads or other activities generally — particularly outside of the domain of prosthetic arms, where R&D has a lot of areas to cover — is not at all new. Sure you may read that here first, but that does not mean any such numbers and figures are new, concept-wise, domain-wise, or application-wise. It just means it never affected you, and depending on where you work that’s fine. If you work for e.g. Fillauuer (check this) it is certainly not fine if/when/that you do not know these things. As we now established that we are not the same, it would be neither unrealistic for me to ask that manufacturers put tech specs on their devices. Medical device regulation would at least in theory require proper testing and documentation anyway – not that our friends at e.g. Fillauer are interested. And it is not impossible to provide proper documentation after hard relevant technical testing, nor is it unknown in the domain of real work to measure and document loads. Medical devices are quite if not insanely expensive, exactly for the reason that proper testing and performance documentation are required. If you read that here first, that sure means something alright, but it probably means something else than this being groundbreaking new information. And even if some of these figures are “old” – heavy or physically demanding work won’t go away – lifting weight today is probably just as strenuous as it was yesterday.
• The whole setup will be just as robust as its weakest link. This certainly must come as the largest surprise to prosthetic arm manufacturers as well as prosthetists – who are selling me a “robust looking prosthetic hook” (robust as in: the metal of the single hook finger), when the hook joint wiggles soon, the wrist starts to wiggle and jam soon as well, the cable dies every 4-10 days, and so on. So really, their statement should have been right from the start, “we do NOT have a prosthesis that is anywhere near robust”.  It is a fact, that a component assemblage may take orders for its repeated failure from a list of parts sorted by weakness with weakest first, and with some interactions to mitigate or exacerbate this [1].  An interesting observation that supports a more convoluted role of “weakest link impact” is that of my iLimb, where fitting a glove that does not decay / tear apart by itself in the cupboard (leave alone when using it) will reduce the function of the precision grip further, reducing overall device function to a degree that the consequence is that instantly, other devices will take over the role of daily prosthetic arm, with a far greater reliability and better grip function altogether – which in itself may be seen as protecting the otherwise admittedly very fragile iLimb from being used too often, with the secondary effect that it fails a lot less over combined use/non-use time. And back to some more mechanical problems: once a prosthetic cable rips / dies every 4-10 days over a period of two years, no particular statistical models are needed to address this, like, at all – what you need is what you always need: good engineering [2].

I have oriented myself along some physical aspects and component related build options for a while now, and I think these figures are all interesting and may show relevant areas of current and future work focusing on specifying and documenting actual loads for the benefit of then defining prosthetic parts that sustain these. As there seems to be no literature that matches prosthetic part specifications and activity or workplace aspects, it is time that science, academia, and industrial development looks into this.

• Normal grip strength in adults is 52-55 kg (men) and 32-34 kg (women). Normal forearm rotation torque is 8,9-9,4 Nm (men) and 5,0-5,3 Nm (women). Normal lift strength is 227-249 N (men) and 135-150 N (women) [3].
• Hettinger tables from 1981 suggest upper limits for different sex and age groups for weights for carrying and lifting [link]. Thereby, age/sex/max. load/repetitions are suggested to be as follows for occasional lifting or carrying for women (15-18 years old: 15 kg, up to 2x per hour, up to 3-4x per day; 19-45 years old: 15 kg, up to 2x per hour, up to 3-4x per day; >45 years old: 15 kg, up to 2x per hour, up to 3-4x per day) and for men (15-18 years old: 35 kg, up to 2x per hour, up to 3-4x per day; 19-45 years old: 55 kg, up to 2x per hour, up to 3-4x per day; >45 years old: 45 kg, up to 2x per hour, up to 3-4x per day); they are defined / suggested for frequent lifting and carrying as follows for women (15-18 years old: 10 kg; 19-45 years old: 10 kg; >45 years old: 10 kg) and for men (15-18 years old: 20 kg; 19-45 years old: 30 kg; >45 years old: 25kg).
• Symptoms of physically heavy work appear to focus around lower back pain for manual handling of really heavy weights whereas apparently / according to that study, neck and shoulder pain was more related to repetitive high-speed work with arms and hands; more specifically, a high frequency of symptoms from the neck and shoulder is reported among assembly plant workers, sewing machine operators and workers using video display terminals (original works: see references cited in [4]). Maximum permissible weight (according to Danish guidelines) for hazardous tasks was given as 50 kg, and max torque as 90 Nm, whereas “unfavorable conditions” may be weighed in for all situations where torque is between 20 and 90 Nm. Interestingly, “a total of 375 different single lifts were registered during the observation of work tasks, and the weight of the burdens varied from 0·5 kg to 87·0 kg. Nine burdens (2%) had a weight of more than 50 kg. Among all the lifts 12% caused a torque above 90 Nm, and 39% caused a torque of between 20 and 90 Nm, while 49% of the lifting caused a torque below 20 Nm. Among the lifts that caused a torque of between 20 and 90 Nm, 131 (90%) of the lifts included additional unfavorable conditions“: only 2% were over the legal limit but when considering qualitative aspects (unregulated), over 90% of lifts between 20 and 90 Nm torque were problematic! Total tonnage lifted for an 8 h working day was 9600 kg plus/minus 10900 kg (seemingly not normally distributed: 132-1000 kg for about 12% of the workers, 1001-5000 kg for about 34% of the workers, 5001-10000 kg for about 14% of the workers, 10001 to 20000 kg for about 22% and >20000 kg for about 12%.
  What was measured in the cited study [4]:
  • item weight, once the item was in excess of 0,5 kg, and
  • distance from the body for torque estimation,
  • lifting frequency over a work cycle,
  • occurrence of unfavorable conditions for each lift: lifting from/to below knee level, lifting from/to above shoulder level, rotation or lateral flexion of the body, lifting with one hand only, risks for unexpected loads, and unhandy materials,
  • total tonnage as the added weight of all items weighing over 0,5 kg,
  • working posture, recording flexion angles for neck and lower back.
• Range of up to ~700 N – Lorry drivers that load and deliver gas bottles typically carry / lift / lower cylinders < 25 kg, roll those > 50 kg, while force exertion ranges around  106-367 N (opening an overhead rear truck door, 20-121 times per day), 153 to 274 N (closing door), 63-575 N (opening side curtain, 63-702 N (closing curtain), using tie-down strap ratchet (404-1168 N, 6-32 times per day) [5]. For specific work tasks related to that work, heart rate mean values were 115-129 bpm. Safe to assume that sweating is an everyday experience there.
  Measurements included:
  • Video recordings were analyzed using VEA (video event analysis) (Chappe Software).
  • Task performed, duration, work postures (using manual posture observation methods: e.g. RULA, REBA, OWAS, and coding the postures), characteristics of gas cylinders handled, frequency, and method of handling.
  • Heart rate (HR) profile, average heart rate, percent heart rate reserve (RHR = 100 x (HR work – HR rest) / (HR max – HR rest). Maximum acceptable work time (MAWT) before developing excessive fatigue: MAWT=26.12 × e(- 4.81×RHR) for longer durations of work and MAWT= – 2.67 + e(7.02- 5.72×RHR) for shorter durations of work.
  • Perceived physical demands.
• Repetitions resulting in over 3300 kg of summed up tonnage per day – “Particularly, manual block lifting, which is an integral part of masonry work, requires masons to perform frequent deep bending of their trunk to lift heavy materials—such as concrete masonry units (CMU) (Hess et al. 2010; van der Molen et al. 2004). According to Hess et al. (2010), block masons manually lift at least 200 CMUs per day, and considering the standard CMU size and weight (0.19 × 0.19 × 0.38 m and 16.6 kg) (CCMPA 2013), masons manually typically lift over 3,300 kg per workday. Moreover, masons spend up to 53% of their working time in a bending posture in order to pick up materials at ground or knee level and 38% of working time in aggravating postures (Boschman et al. 2011). Frequent handling of heavy materials with bending postures exposes masons to severe lower back injuries, and in practice, the back injury rate for masons is the second-highest among construction subsectors (CPWR 2013).”[6] – Safe to assume that sweating is an everyday experience there.
  • Own experience with simple IKEA furniture – installing IKEA Pax wardrobe and self-assembly: total of ~550 kg of material, two days of work [link].
  • Own experience cleaning up furniture parts, lift parts up narrow stairs to the attic – up to ~25 kg per part [link].
• Range of repetitions resulting in some 115-195 kg per case and thus some 230-780 kg for a single work shift day, in forensic medicine; torques estimated at 60-100 Nm for yanked shifts/lifts. For death scene work, for a single case, one will/tends to/may have to / normally will (1) lift partial body weight/pull/lift, up to maybe 40 kg or more, then (2) undress one arm/lift/roll body requiring very strong reliable but padded grip (careful, skin!) of around 8-20 kg, (3) undress the other arm, similar 8-20 kg estimate, then (4) roll body to retrieve upper body clothing with some 10-25 kg push preferably with a rounded/ padded shape (hook, TRS Jaws, hand), then (4) lifting legs (15-35 kg each) to remove trousers, etc., so overall summed up weight will amount to around 115 – 195 kg lifted within some 10-30 minutes. A single service workday may cover 2-4 cases easily [source: own estimates / [7]. Considering some lifts etc are performed by yanking, and using an estimate of 0,3 m length for my stump/prosthesis/elbow to tip distance length, a typical torque value range there may be at least a regular ~60 to ~100 Nm for single acts. Overall, since the work is cumulative in terms of effort, one does tends to sweat, also due to wearing long sleeves/trousers/shirts and various layers of protective materials.
• Wind turbine technicians[8] report most single person lift weights to be around 20-25 kg or in excess of 30 kg while lifting weights as a team was items of mostly 40-50 kg or in excess of 60 kg. Peak numbers of reported durations were 1-2 h for working in tight or restricted space, and 1/2 – 1 h for working with arms above shoulders, while other activities of hazardous quality (reaching behind the body, bending/twisting at the wrist, lifting while bending/twisting, bending wrists up/down frequently) were reported to most often last only up to 10 minutes. In the context of injury risk, awkward postures, heavy lifting, and repetition were indicated by over 50% of respondents. This confirms that these types of exposures are the ones to also, when working, proactively look for most.
• Lifting 30-40 kg with the hands may generate torque at the wrist of around 120-160 Nm. That is probably more than an osseointegration into thin osteoporotic/osteopenic forearm stump bones can hold without fracture [link].
• Hand-arm vibration in cycling is a real issue. The study [9] is paper that is concerned with the propagation of vibrations in the hand–arm system of the cyclist in context of the ISO5349-1 standard. These vibrations generated by the road profile can be harmful to the health of the athlete and also, obviously, cause technical problems to the prosthetist. For a unidirectional excitation on the pavement, the vibrations felt by the cyclist are expressed in all three axes, and especially along the normal axis of movement. It is also shown that the vibrations are amplified from a speed of 20 km/h that is 38 Hz, which corresponds to the resonance of the wrist. The radial axis of movement is the most amplified. This excitation is similar to a wavering and oscillation type deformation. We take away that an equivalent for a maximal daily vibration exposure (“A(8)” in ISO 5349-1 according to the cited study) may be already reached, with a weak estimate, after around 7 minutes of cycling at a speed of 35 km/h.
• Handlebar pull forces on a bicycle during riding are estimated to be up to around 421 N (or 0.64 x body weight); horizontal handle bar forces are estimated at 175 N  [10].
• Push-ups may cause wrist forces of up to 283 N, with axial forces of up to 663 N, and moments for the radio-ulnar axis of up to 18,6 Nm [11].
• This points to the relative importance of wearing a well-built rather than an ill-devised body-powered prosthetic arm [link] – and it is to be expected that not every prosthetic technician, possibly only a very few worldwide, are able to list, name, understand, explain, leave alone build, a well-conceived body-powered prosthetic arm, like, at all. So if you want a well-built prosthetic arm, chances are not super you will ever get one because prosthetists lack the know-how.

What are good posture and optimal reach workspaces?

• Posture, in general, is a key issue to handling hazardous work well, and correct posture does not include awkward reach to outer ranges of possible extensions (from [6] – “Awkward body-postures and motions with external forces tend to create excessive musculoskeletal stresses beyond the internal tolerance of tissues (Kumar 2001). Appropriate work posture is also considered an essential factor for improving productivity on the job (Gilbreth and Gilbreth 1917)”). That, however, does not pertain to posture during a non-work-related task such as a clothespin test. Instead, it pertains to specific task-related postures.
• Avoid bending elbows from full extension when lifting relatively heavyweight, as the ideal range for power development of muscle groups appears to be better in angles / joint states that are different from a full extension; for elbows, the best range seems to be between 85-100 degrees of flexion (e.g., from [12] – “The range of motion on elbow joint during this exercise also affects the mechanical loading. If the biceps curl performs starting from 0 degrees of elbow extension to elbow flexion, the length of the effort arm is continuously be increased, and mechanical advantage is going to be greatest at 85-100 degrees of elbow flexion. While the longer effort arm provides the mechanical advantage, as the muscle will be shorter and overlapped myofilaments in a full range of elbow flexion reduces force capacity as it was first described by Gordon et al. (1966), defining the force-length relationship and its direct relation to the myofilament overlap. Therefore, if the full range of elbow flexion is chosen during the biceps curl exercise, the biceps muscle loses the force capacity.“).
• Reach space is not required to be maximal for any real work (read this post too): it is not necessary, particularly for real work (i.e., high quality trained manual work, such as mounting appliances, lifting or repetitive heavy work, etc.), to have a particularly wide range of motion where the task / activity is delivered; a highly reliable control and a robust performance under loads given within a reasonably narrow reach space seems more important.

Training tips to prepare for hazardous activities

• Listening to older experienced people may be your only chance to avoid or survive the “new mason’s risk hump” – because new workers become old seasoned workers only by adopting safe practices, otherwise, the journey won’t be linear  (from [6] – “The results show that the journeymen with more than 20 years of experience adopt safer and more productive work techniques distinct from those of less experienced workers. The present study contributes to the body of knowledge on masons’ safety and productivity by providing an in-depth understanding of the linkage between body loads, work experience, techniques, and productivity. Additionally, the findings in this study are expected to have a greater impact when they are adopted to apprentice-training methods and applied to other high musculoskeletal-disorders-risk trades.”). In the context of prosthetic arms, a body-powered hook was the tip from older people that provide real work themselves, and I pass it on here. To wear only well engineered parts was another truism from old times, when we still learned that if one wants things to be built right one had to build them oneself. That much other products are just malarkey is a given, you certainly did not come here to have me explain the whys and hows of that again.
• The first and foremost criterium for “prosthetic use training and use proficiency” to come into closer consideration at all, to become a subject or topic in the life of an arm amputee, is a prosthetic arm that – with few exceptions – works, always works, always is comfortable and always functional, that is “extremely” reliable and robust, and stable in construction and build, that does not impose frequent and significant rest/downtime on the user, not by way of component failure, and not by way of skin damages or other user body damages. The reason for that is that any attempt for training, building proficiency, seems like a rather pointless undertaking as such, as long as the device will regularly cease to function early, even after maybe just a week, requiring yet another repair that will delay further use for 1-2 weeks or more, or if it tends to systematically chafe, abrade and blister up the stump, enforcing downtime for prosthetic use of one or several days at least (from [12] – “All beneficial effects of exercise training are reversible if exercise ceases.”). Conversely, a bit of rest between usage/training series is not a bad thing (from [12] – “Optimal adaptation requires sufficient rest periods to be interspersed with training sessions that the adaptations caused by the exercise dose can take place.”). If you add up all use days u and all non-use days nu, over 1, 5, 10 years of time, then obtain a sum of both and check the non-use ration nur = nu/u, you will find typical numbers of 3-10; so we all are 3-10 times more exposed and accustomed to not using the prosthetic arm for any reason – it lasts longer if it can be not used for a day, week or month. It may be necessary to give it a break anyway. It was never that useful to begin with or was it. Financing repairs is a real issue sometimes, to not using it makes it live to a later point in time. Wearing it may cause anger and disappointment, because that is what commercial parts containing prosthetic arms just do – so maybe better not go down that emotional rabbit hole. So the overall process you need to understand is a non-linear incremental model that rolls through time, whereby that train of hope that has you believing these prosthetic arms are the end-all be all, day by day, week by week, is incrementally shattered to tiny sharp glass fragments all the while money, hope, time, material, health, is consumed by that fire of inefficient and tiring prosthetic arm build and use process by people that never are accountable. What they will tell the user at any time is “if anything breaks it is by definition because you violated something – so, admit it, what did you do?”. So that incremental process, over years, is a mix of time toxicity and subterranean product use and “customer service” experience. It is not just bad, as in tolerable. It is a black polished shiny version of how not to do things in a comprehensive way. As user, you enter at your own risk and you do that better on your own terms.
• It seems important to train as close to the application domain as possible, or, if possible, within the application domain (from [12] – “Optimal benefits occur when exercise meet the individuals’ needs and capacities of the patient or client. The exercise prescription should be based on individual needs that will change by person to person.”; “The training stimulus must be specific to the patient or client’s desired outcomes. Therefore, exercise design must be specific to individuals’ goals.”). So you don’t need some funny softwares or grasp stryofoam puppets. You can directly go and move boxes and drill holes. No need to train clothespin tests – just wash, hang, and iron your stuff. Gets the housework done at the same time.
• Training is best performed in the upper range of what one can already do. Pushing the envelope is usually and regularly perceived as “hard”. Just reading a newspaper or making coffee the way one always did it may not constitute a type of use exposure that helps the prosthetic arm user get better (from [bibcite key=turgut2020biomechanical – “It is important to consider that for adaptation to occur the volume of exercise must overload the body in some way in line with the capacity of the individual to cope with that overload.”; “For continual adaptation, overload must be progressive, that is the dose of loading or stress must increase.”).
• As training may blend into a real-life application, a particular stratification or grouping of exercises seems not necessary; much rather, letting training occur within a seemingly unplanned or even chaotic everyday exposure seems to be a sensible and also cost-effective as well as training-effective approach (from [12] – “It should be kept in mind that for optimal adaptation and to avoid stagnation, overuse, and injury the exercise stimulus must be varied. Therefore, variety allows recovery and can reduce injury risk.”).
• Using clearly dedicated “zones” or “application domains” may help at first. Because at first one may feel lost, not knowing where to start or where to apply oneself. Thus clearly defining positive use boundaries (“here you must use prosthesis”) versus non-use boundaries (“here it is acceptable to stop any use and take time out for relaxing”) may be a sensible approach to creating an observable, applicable, and even measurable scenario within which use cases and training efforts may be defined and instanced.
• Performance that is objectified/measured may remain lower/less when performed by an arm amputee regardless of optimal or perfect prosthetic setup. For a golf sport prosthetic extension, despite perfect prosthetic equipment, lower club speeds resulted always in one study [13].
• Know specific injury risks and take precautions.
  • Bicycle riding, for example, appears to have its own dangers in terms of long-term exposure to possibly hazardous levels of activity. From [14]: “Overuse injuries may occur in bicycle riders who regularly ride their bicycle [..]. Ensuring that the bicycle seat (saddle), handlebars, and pedals are correctly adjusted and that the bicycle is the appropriate size can be key in preventing overuse syndromes. Neck aches and backaches are common complaints resulting from the cyclist’s upper body position with hyperextension of the neck and flexion of the lower back. In addition to advice on rest, stretching exercise, and anti-inflammatory medications, the physician may suggest shortening the handle-bar reach, creating a slight upward tip of the saddle angle of 10 to 15 degrees, or regularly changing hand and arm position on the handlebars and keeping the elbows slightly flexed while riding. Prolonged pressure on the handlebars and the position of the wrists may cause compression neuropathies in the hands.” — In terms of riding position, also steep uphill riding, varying position – standing, sitting, etc. – has been shown to change also upper extremity strain and activation, so changing sitting position etc. is a proven method to vary or change muscle and tendon strain while riding bicycle [15], which in essence seems to be just what the doctor recommended (read the previous paragraph).

@article{tol2007weakest,
  title={The weakest link hypothesis for adaptive capacity: an empirical test},
  author={Tol, Richard SJ and Yohe, Gary W},
  journal={Global Environmental Change},
  volume={17},
  number={2},
  pages={218--227},
  year={2007},
  publisher={Elsevier}
}

@article{hristopulos2015weakest,
  title={Weakest-link scaling and extreme events in finite-sized systems},
  author={Hristopulos, Dionissios T and Petrakis, Manolis P and Kaniadakis, Giorgio},
  journal={Entropy},
  volume={17},
  number={3},
  pages={1103--1122},
  year={2015},
  publisher={Multidisciplinary Digital Publishing Institute}
}

@article{axelsson2018forearm,
  title={Forearm torque and lifting strength: normative data},
  author={Axelsson, Peter and Fredrikson, Per and Nilsson, Anders and Andersson, Jonny K and K{\"a}rrholm, Johan},
  journal={The Journal of Hand Surgery},
  volume={43},
  number={7},
  pages={677--e1},
  year={2018},
  publisher={Elsevier}
}

@article{christensen1995national,
  title={A national cross-sectional study in the Danish wood and furniture industry on working postures and manual materials handling},
  author={CHRISTENSEN*, HANNE and PEDERSEN, MIKKEL B and SJ{\o}GAARD, GISELA},
  journal={Ergonomics},
  volume={38},
  number={4},
  pages={793--805},
  year={1995},
  publisher={Taylor \& Francis}
}

@article{sekkay2020assessment,
  title={Assessment of physical work demand of short distance industrial gas delivery truck drivers},
  author={Sekkay, Firdaous and Imbeau, Daniel and Dub{\'e}, Philippe-Antoine and Chinniah, Yuvin and de Marcellis-Warin, Nathalie and Beauregard, Nancy and Tr{\'e}panier, Martin},
  journal={Applied Ergonomics},
  volume={89},
  pages={103222},
  year={2020}, 
  publisher={Elsevier}
}

@article{ryu2020analysis,
  title={Analysis of relationships between body load and training, work methods, and work rate: Overcoming the novice mason’s risk hump},
  author={Ryu, JuHyeong and Alwasel, Abdullatif and Haas, Carl T and Abdel-Rahman, Eihab},
  journal={Journal of Construction Engineering and Management},
  volume={146},
  number={8},
  pages={04020097},
  year={2020},
  publisher={American Society of Civil Engineers}
}

@article{fischer2021cross,
  title={A cross-sectional survey of musculoskeletal disorder hazard exposures and self-reported discomfort among on-shore wind turbine service technicians},
  author={Fischer, Steven L and Koltun, Sarah and Lee, Jessie},
  journal={Ergonomics},
  volume={64},
  number={3},
  pages={383--395},
  year={2021},
  publisher={Taylor \& Francis}
}

@article{chiementin2013hand,
  title={Hand--arm vibration in cycling},
  author={Chiementin, X and Rigaut, M and Crequy, S and Bolaers, F and Bertucci, W},
  journal={Journal of Vibration and Control},
  volume={19},
  number={16},
  pages={2551--2560},
  year={2013},
  publisher={Sage Publications Sage UK: London, England}
}

@article{soden1979forces,
  title={Forces applied to a bicycle during normal cycling},
  author={Soden, PD and Adeyefa, BA},
  journal={Journal of Biomechanics},
  volume={12},
  number={7},
  pages={527--541},
  year={1979},
  publisher={Elsevier}
}

@article{smith2018forces,
  title={Forces in the distal radius during a pushup or active wrist motions},
  author={Smith, Jared M and Werner, Frederick W and Harley, Brian J},
  journal={The Journal of Hand Surgery},
  volume={43},
  number={9},
  pages={806--811},
  year={2018},
  publisher={Elsevier}
}

@incollection{turgut2020biomechanical,
  title={Biomechanical principles of the exercise design},
  author={Turgut, Elif and Baltaci, Gul},
  booktitle={Comparative Kinesiology of the Human Body},
  pages={527--538},
  year={2020},
  publisher={Elsevier}
}

@article{carey2015golf,
  title={Golf hand prosthesis performance of transradial amputees},
  author={Carey, Stephanie L and Wernke, Matthew M and Lura, Derek J and Kahle, Jason T and Dubey, Rajiv V and Highsmith, M Jason},
  journal={Prosthetics and Orthotics International},
  volume={39},
  number={3},
  pages={244--249},
  year={2015},
  publisher={SAGE Publications Sage UK: London, England}
}

@article{thompson2001bicycle,
  title={Bicycle-related injuries},
  author={Thompson, Matthew J and Rivara, Frederick P},
  journal={American Family Physician},
  volume={63},
  number={10},
  pages={2007},
  year={2001}
}

@article{duc2008muscular,
  title={Muscular activity during uphill cycling: effect of slope, posture, hand grip position and constrained bicycle lateral sways},
  author={Duc, Sebastien and Bertucci, W and Pernin, JN and Grappe, Fr{\'e}d{\'e}ric},
  journal={Journal of Electromyography and Kinesiology},
  volume={18},
  number={1},
  pages={116--127},
  year={2008},
  publisher={Elsevier}
}