Reply to "Letter to the editor: Consistency among musculoskeletal models: caveat utilitor".
Reply to "Letter to the editor: Consistency among musculoskeletal models: caveat utilitor".
复制标题
回复“给编辑的信：肌肉骨骼模型之间的一致性：警告实用程序”。
DOI：
10.1007/s10439-014-1152-z
复制
发表时间：
2015
影响因子：
3.8
通讯作者：
Beaupre,GaryS
中科院分区：
文献类型：
--
作者：
Wagner,DavidW;Stepanyan,Vahagn;Shippen,JamesM;DeMers,MatthewS;Gibbons,RobinS;Andrews,BrianJ;Creasey,GrahamH;Beaupre,GaryS
我们要感谢Modenese博士和他的同事花时间讨论我们的文章“肌肉骨骼模型的一致性：警告效用”。我们也要感谢这些研究人员将他们的团队开发的模型，伦敦下肢模型（LLLM），提供给肌肉骨骼建模界。在回顾了他们的分析之后，我们确实同意LLLM研究人员的观点，即他们的模型与我们分析的其他模型相比产生的股四头肌肌肉恢复的巨大差异（在我们的原始文章中见图4）主要与他们的模型中“肌肉执行器既没有执行收缩动力学也没有执行力-长度-速度关系”这一事实有关。我们最初分析的目的并不是要歪曲LLLM模型。其目的是在各自软件共享的规范分析中使用模型，并为未来的模型用户提供建议。由于原始研究的目的是“比较来自三个软件包的八种模型，并评估股四头肌力臂，预测肌肉力量和预测胫股接触力的差异……”我们为所有模型和所有软件包选择了标准化的方法和分析技术。更具体地说，我们选择了最有可能被模型和软件的未来用户采用的方法。例如，在OpenSim中，标准化方法默认考虑肌肉力-长度关系，今天生物力学家选择模型最有可能采用这些标准化的默认方法。此外，值得注意的是，LLLM模型并没有规定相关的肌纤维长度和肌腱长度参数，而不管这些参数是否有效。我们感到遗憾的是，LLLM的肌肉恢复力结果是由一项分析产生的，该分析最初并非打算与LLLM一起使用，并且结果被解释为潜在的建模问题。在同一OpemSim建模软件2中实现的其余模型没有与LLLM模型相同的问题，在基于相同Klein-Horsman解剖数据集的任何人建模系统1中构建的模型也没有同样的问题。我们同意法学硕士研究人员的观点，“试图联系模型的开发人员来澄清不确定性……是一种好做法”。然而，这样做的必要性对用户来说可能并不总是显而易见的，就像我们使用他们的模型的情况一样；当时，我们相信我们正在按照开发人员的意图使用模型。llm研究人员提出的差异表明，肌肉骨骼建模要被广泛接受和采用为日常工程工具，仍然存在一些实质性的障碍。我们感谢llm研究人员澄清了他们模型的局限性，并为如何使用他们的模型更紧密地匹配其他现有的肌肉骨骼模型提供了指导。我们相信，这种健康的讨论对于肌肉骨骼建模作为一门工程学科的发展和接受是至关重要的，我们希望工程界正在朝着这个方向前进。我们认为是这样的
To the Editor, We would like to thank Dr. Modenese and colleagues for taking the time to discuss our article ‘‘Consistency Among Musculoskeletal Models: Caveat Utilitor’’. 6 We would also like to thank those same researchers for making the model developed by their group, the London Lower Limb Model (LLLM), 5 available to the musculoskeletal modeling community. After reviewing their analysis, we do agree with the LLLM researchers that the large discrepancy in quadriceps muscle recruitment (Fig. 4 in our original article) produced from their model as compared with the other models that were analyzed was primarily related to the fact that ‘‘neither contraction dynamics nor force–length–velocity relationships were implemented for the muscle actuators’’5 in their model. The intent of our original analysis was not to misrepresent the LLLM model. The intent was to employ the models in canonical analyses shared by their respective software and advise future model users. As the purpose of the original study was to ‘‘compare eight models from three software packages and evaluate differences in quadriceps moment arms, predicted muscle forces, and predicted tibiofemoral contact forces…’’we selected standardized methods and analysis techniques for all models and all software packages. More specifically, we selected the methods most likely to be adopted by future users of the models and software. For example, in OpenSim, the standardized methods account for muscle force–length relationships by default and a biomechanist selecting a model today is most likely to adopt these standardized, default methods. Additionally, it should be noted that the LLLM model does specify the relevant muscle fiber length and tendon length parameters, regardless of whether those parameters are valid. We regret that the muscle recruitment force results from the LLLM were produced from an analysis that was not originally intended to be used with the LLLM and that the results were interpreted as a potential modeling issue. The remaining models implemented in the same OpemSim modeling software 2 did not have the same issue as the LLLM model, nor did the model constructed in the AnyBody Modeling System 1 that was based on the same Klein-Horsman anatomical data set. 3 We agree with the LLLM researchers that,‘‘attempting to contact the developers of a model to clarify uncertainties is… good practice’’. The necessity to do so, however, may not always be obvious to the users, as was the situation with our use of their model; at the time we believed we were using the model as intended by the developers.The discrepancy raised by the LLLM researchers suggests that there remain several substantial hurdles for musculoskeletal modeling to be widely accepted and adopted as an everyday engineering tool. We thank the LLLM researchers for clarifying the limitations of their model and for providing guidance on how to use their model to more closely match other currently available musculoskeletal models. We believe that this healthy discussion is critical for the growth and acceptance of musculoskeletal modeling as an engineering discipline, a direction we hope the engineering community is heading toward. We believe it is