(C-81) Scaling a Generic Upper Limb Musculoskeletal Model to Represent Individuals with Paraplegia

Introduction:: A generic musculoskeletal (MSK) representative of a 50^th percentile adult male can be anthropometrically scaled to represent an individual nonimpaired adult or child using limb length [1,3] and muscle volume [2,3]. Using this approach, a scaled upper limb model predicted maximum isometric shoulder moments that were correlated to moments measured experimentally from the same individuals [2,3]. Accurately representing the strength of individuals is needed to estimate muscle stress during functional activities. It is unknown if this anthropometric scaling approach would represent the strength of individuals with paraplegia that use manual wheelchairs [4]. This is the preliminary analysis to compare the experimentally measured shoulder and elbow moments from those with paraplegia to moments predicted from a model anthropometrically scaled MSK models to represent the same individuals.



Materials and Methods:: The overall approach is to scale a generic model based on reported limb length and muscle volumes [3], and then compare the model predicted and measured maximum isometric moments. Four individuals with paraplegia, including two pediatric (1 female: 16.2 years; 1 male 13.9 years) and two adult subjects (1 female: 19.3 years; 1 male 20.1 years) were selected from a larger cohort. Participants performed maximum isometric voluntary contractions (MVIC) while seated in a Biodex System 3 to measure maximum moments of shoulder abduction, flexion, extension, external rotation, horizontal adduction, and elbow flexion and extension. A scapulothoracic joint model [4] was adapted to include elbow and forearm muscles originating from the MoBL-ARMS model [5] in OpenSim (version 4.3). Linear regression was used to predict segment length [6,7] and total muscle volume [2,7,8] from the individual’s total height and body weight, respectively. The predicted total shoulder muscle volume was portioned into the upper limb muscle groups following the proportions of the generic model. Peak isometric muscle force was calculated using individual muscle volumes, specific tension, and optimal fiber length from length-scaled pediatric models. Assuming full excitation of agonist muscles, the scaled models were used to compute predicted maximum moments in the same arm postures as the experiments. Mean absolute error (MAE) was calculated between the predicted and measured moments for each participant and each moment position.

















Results, Conclusions, and Discussions:: The model predicted and empirically measured maximum isometric moments are reported in Table 1. The highest MAE was found in the adult male participant (MAE=0.46), and the pediatric male participant (MAE=0.40) had the lowest. The moment with the largest MAE was shoulder extension and shoulder internal rotation (0.75 and 0.77). The moment position with the smallest MAE was shoulder abduction (0.14).

Model predictions were most accurate for shoulder abduction, and least accurate for shoulder extension and internal rotation. Unlike a previous study investigating nonimpaired children [3], the model both overestimated and underestimated strength in the different arm postures. This could be due to the use of linear regression models predicting limb length and muscle volumes based on nonimpaired individuals; the results here suggest that this approach is inappropriate.

The high error between model predicted and measured elbow and shoulder moments suggests more direct measures of individual muscle strength and muscle size are needed to represent individuals with paraplegia, however further analysis is required. Work is underway to analyze the relationship between model predicted and measured moments in a larger cohort.



Acknowledgements (Optional): : Funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) of the National Institutes of Health (award number R01HD098698).




References (Optional): : [1] Hicks J. et al. J Biomech Engin 2015;137(2). [2] Vidt M, et al. J Biomech 2012;45(2). [3] Dalman M, et al. J Biomech 2022;141:111170. [4] Seth A, e al. Front Neurorobot 2019;13(90). [5] Saul K, et al. Comput Methods Biomech Biomed Engin. 2015;18(13). [6] Neu CM, et al. American Journal of Phys 2002;283(46). [7] Holzbaur K. et al. J Biomech 2007;40(4). [8] Im HS, et al. J Biomech 2014;47(11).