Yoganandan Narayan, Arun Mike W J, Pintar Frank A, Szabo Aniko
a Department of Neurosurgery , Medical College of Wisconsin , Milwaukee , Wisconsin.
Traffic Inj Prev. 2014;15 Suppl 1(0 1):S151-6. doi: 10.1080/15389588.2014.935357.
Derive optimum injury probability curves to describe human tolerance of the lower leg using parametric survival analysis.
The study reexamined lower leg postmortem human subjects (PMHS) data from a large group of specimens. Briefly, axial loading experiments were conducted by impacting the plantar surface of the foot. Both injury and noninjury tests were included in the testing process. They were identified by pre- and posttest radiographic images and detailed dissection following the impact test. Fractures included injuries to the calcaneus and distal tibia-fibula complex (including pylon), representing severities at the Abbreviated Injury Score (AIS) level 2+. For the statistical analysis, peak force was chosen as the main explanatory variable and the age was chosen as the covariable. Censoring statuses depended on experimental outcomes. Parameters from the parametric survival analysis were estimated using the maximum likelihood approach and the dfbetas statistic was used to identify overly influential samples. The best fit from the Weibull, log-normal, and log-logistic distributions was based on the Akaike information criterion. Plus and minus 95% confidence intervals were obtained for the optimum injury probability distribution. The relative sizes of the interval were determined at predetermined risk levels. Quality indices were described at each of the selected probability levels.
The mean age, stature, and weight were 58.2±15.1 years, 1.74±0.08 m, and 74.9±13.8 kg, respectively. Excluding all overly influential tests resulted in the tightest confidence intervals. The Weibull distribution was the most optimum function compared to the other 2 distributions. A majority of quality indices were in the good category for this optimum distribution when results were extracted for 25-, 45- and 65-year-olds at 5, 25, and 50% risk levels age groups for lower leg fracture. For 25, 45, and 65 years, peak forces were 8.1, 6.5, and 5.1 kN at 5% risk; 9.6, 7.7, and 6.1 kN at 25% risk; and 10.4, 8.3, and 6.6 kN at 50% risk, respectively.
This study derived axial loading-induced injury risk curves based on survival analysis using peak force and specimen age; adopting different censoring schemes; considering overly influential samples in the analysis; and assessing the quality of the distribution at discrete probability levels. Because procedures used in the present survival analysis are accepted by international automotive communities, current optimum human injury probability distributions can be used at all risk levels with more confidence in future crashworthiness applications for automotive and other disciplines.
使用参数生存分析得出最佳损伤概率曲线，以描述人类小腿的耐受性。
该研究重新审视了来自一大组标本的小腿尸体人类受试者（PMHS）数据。简要来说，通过冲击足底进行轴向加载实验。测试过程中包括损伤和未损伤测试。通过测试前后的X射线图像以及冲击测试后的详细解剖来识别损伤情况。骨折包括跟骨和胫腓骨远端复合体（包括胫距关节）损伤，代表简略损伤评分（AIS）2+级及以上的严重程度。对于统计分析，选择峰值力作为主要解释变量，选择年龄作为协变量。删失状态取决于实验结果。使用最大似然法估计参数生存分析的参数，并使用dfbetas统计量识别影响过大的样本。基于赤池信息准则确定威布尔分布、对数正态分布和对数逻辑分布的最佳拟合。获得最佳损伤概率分布的正负95%置信区间。在预定风险水平确定区间的相对大小。在每个选定的概率水平描述质量指标。
平均年龄、身高和体重分别为58.2±15.1岁、1.74±0.08米和74.9±13.8千克。排除所有影响过大的测试后得到最窄的置信区间。与其他两种分布相比，威布尔分布是最优化的函数。当提取25岁、45岁和65岁年龄组在5%、25%和50%风险水平下小腿骨折的结果时，该最佳分布的大多数质量指标属于良好类别。对于25岁、45岁和65岁年龄组，5%风险水平下的峰值力分别为8.1千牛、6.5千牛和5.1千牛；25%风险水平下为9.6千牛、7.7千牛和6.1千牛；50%风险水平下为10.4千牛、8.3千牛和6.6千牛。
本研究基于生存分析，使用峰值力和标本年龄得出轴向加载引起的损伤风险曲线；采用不同的删失方案；在分析中考虑影响过大的样本；并在离散概率水平评估分布质量。由于本生存分析中使用的程序为国际汽车界所接受，当前的最佳人类损伤概率分布可在所有风险水平上使用，从而在未来汽车及其他学科的耐撞性应用中更有信心。
