Abstract

Computational human body models (HBMs) provide the ability to explore numerous candidate injury metrics ranging from local strain based criteria to global combined criteria such as the Tibia Index. Despite these efforts, there have been relatively few studies that focus on determining predicted injury risk from HBMs based on observed postmortem human subjects (PMHS) injury data. Additionally, HBMs provide an opportunity to construct risk curves using measures that are difficult or impossible to obtain experimentally. The Global Human Body Models Consortium (GHBMC) M50-O v 6.0 lower extremity was simulated in 181 different loading conditions based on previous PMHS tests in the underbody blast (UBB) environment and 43 different biomechanical metrics were output. The Brier Metric Score were used to determine the most appropriate metric for injury risk curve development. Using survival analysis, three different injury risk curves (IRC) were developed: “any injury,” “calcaneus injury,” and “tibia injury.” For each injury risk curve, the top three metrics selected using the Brier Metric Score were tested for significant covariates including boot use and posture. The best performing metric for the “any injury,” “calcaneus injury” and “tibia injury” cases were calcaneus strain, calcaneus force, and lower tibia force, respectively. For the six different injury risk curves where covariates were considered, the presence of the boot was found to be a significant covariate reducing injury risk in five out of six cases. Posture was significant for only one curve. The injury risk curves developed from this study can serve as a baseline for model injury prediction, personal protective equipment (PPE) evaluation, and can aid in larger scale testing and experimental protocols.

Introduction

Underbody Blast Lower Extremity Background.

Increased modernization and technological advances have greatly changed the way wars are fought. The antivehicle landmine has become the biggest threat to armored vehicles and the occupants within them [1]. Improvised explosive devices (IEDs) have become the most prevalent cause of fatal injuries [2] and accounted for 78% of injuries in operation Iraqi freedom (OIF) and operation enduring freedom (OEF) [3]. Injuries from IEDs are more severe than typical wounds from previous wars. The high amount of energy transferred from the IED to the vehicle floor causes significant damage to the occupants that often require aggressive surgery [4]. Fortunately, improvements have been made to personal protective equipment (PPE), field care and vehicle design and survival rates are higher than any previous conflict at 90% [2,5]. Danelson et al. queried data from the Joint Trauma Analysis and Prevention of Injury in Combat network on warfighter injuries in mounted underbody blast (UBB) attacks. The goal of the study was to provide an in-depth analysis with specific inclusion criteria and a case review of injuries to determine the corresponding mechanisms for mounted warfighters exposed to a UBB event during OEF and OIF [6]. The most severe injuries were to the pelvis, lumbar and thoracic spine, while the most commonly injured body region was the lower extremity. These finding are in agreement with previous studies [5,79]. Out of the injuries reviewed, 79% had a foot, ankle or leg AIS 2+ fracture. Furthermore, the distal tibia, distal fibula and calcaneus fractures were the most common lower extremity injury.

The most common injury mechanism for foot injuries was compression alone, followed by compression with differential loading. Differential loading was described as the processes in which the forefoot was loading at a different rate compared to the heel and resulted in greater forefoot displacement in the vertical direction. Compression was also the most common injury mechanism for the calcaneus, talus, and tibia. The tibia compression fractures were high energy fractures that resulted from the large floor deformation loading the calcaneus and forefoot bones, which then converged to the talus and into the tibia [6].

Whole Body Testing and Pulse Characterization.

Underbody blast loading conditions are highly variable, short duration events, with high accelerations. The floor pan accelerations can reach over 100 G's and others have reported peak velocities up to 30 m/s in 6–10 milliseconds [10,11]. Therefore, the lower extremity is of prime importance to understand the biomechanics of UBB due to its close proximity to the floor in these conditions.

To study the injury mechanisms and better protect soldiers in the UBB environment, the U.S. Army began the Warrior Injury Assessment Manikin (WIAMan) program in 2011. The goal of the project was to develop an anthropomorphic test device (ATD) primarily for the UBB environment. Previously, ATDs have been used in the automotive environment and were not validated for vertical loading or the extreme nature of the blasts [12,13]. In a study by Peitsch et al., the WIAMan ATD response was compared to matched pair testing of Postmortem Human Subject (PMHS) in a variety of subinjurious loading conditions. Three test conditions were designed with peak seat and floor velocities ranging from 4 to 6 m/s and time to peak ranging from 5 to 20 ms. PPE included boots or boots + helmet + body armor.

The model response was compared to the PMHS response and quantitatively evaluated using biofidelity response corridors that were developed by methods outlined in Ref. [14]. Overall, the ATD achieved biofidelity scores ranging between 0.59 and 0.63 for the three different test conditions and indicated a fair ATD response compared to the PMHS test subjects [15].

Additional PMHS tests have been conducted to study the full body response in the UBB environment. Bailey et al. tested PMHS using the ODYSSEY blast simulator and found that the pelvis and feet experienced the largest accelerations and loads due to the close proximity to the seat and floor [16]. Danelson et al. also carried out tests using a blast buck (accelerative loading fixture) to compare the response of PMHS to the Hybrid III ATD. The results of the study indicated that the lower extremity is the initial impact point in UBB and therefore experiences the highest loads. Additionally, the Hybrid III ATD was unable to assume the same seated posture as the PMHS and exhibited an overall stiffer response making it less feasible for use in the UBB environment [17]. Most recently, Ott et al. compared the response of the Hybrid III ATD to PMHS in six matched pair testing. The ISO comparison method was used and accelerations were compared at the head, torso, pelvis, and tibia [18]. The PMHS were sensitive to peak velocity and pulse duration. Overall, the tibia of the Hybrid III exhibited “fair” comparison and the pelvis received a “poor” rating [19].

Component Level Testing and Injury Risk Curves.

Previous work and the development of automotive safety regulations have relied on the development of injury risk curves (IRC's) to quantify the probability of injury for a given input measure such as force. This process has also been used with ATD's to use this injury criterion to accurately evaluate the injury risk for a human subjected to the same input loading conditions. Component level PMHS testing has been done on the lower extremity to develop injury risk curves. An overview of the testing done is shown in Table 1. Previous work has primarily been done for the automotive environment to protect against floor pan intrusions [2022], but component level tests were expanded to the UBB environment to develop similar injury criteria [9,2326]. Peak force was used as the input variable for the injury risk curves, but the location that the peak force was collected varied from experiment to experiment and was either collected at the floor plate or plantar surface, or along the midshaft of the tibia. The Medical-College of Wisconsin laboratory has conducted lower extremity component level tests to better understand the injury mechanisms in the UBB environment [79,27]. Chirvi et al. compiled the data from these component level tests to develop IRC's for the foot-ankle complex. Since the dataset was relatively large for biomechanics data (50 specimens), covariates such as age, PPE, and posture were considered. Age and PPE were found to be significant covariates, but ultimately, boot use was dropped from the final risk curves since fracture of a bone should be independent of footwear. However, specific injury risk curves were developed for minor and major calcaneus injuries using the Sanders classification and two severities of tibia risk curves were generated.

Table 1

Overview of previous PMHS work done to develop injury risk curves for the lower extremity

StudyField of studyN PMHSExperimental boundary conditionForce measurementResulting IRC
Yoganandan et al. [28] resampled in 2014 [29]Automotive52Potted at proximal tibia. UnbootedProximal tibia“Any injury”
Crandall et al. [22]Automotive50Potted midfemurProximal tibia“Any injury”
Kuppa et al. [30]Automotive50Whole body data“Tibia” and “calcaneus” injury
Funk et al. [20]Automotive43Potted at proximal tibia. UnbootedMidtibia“Any injury”
Bass et al. [26]UBB20Potted at femur bootedMidtibia“Any injury”
Mckay and Bir [24]UBB18Potted at femur unbootedMidtibia“Any injury”
Henderson et al. [23]UBB18Proximal tibia pottingProximal tibia“Any injury”
Bailey et al. [9]Both82Compiled previous studiesFloor plate“Any injury”
Chirvi et al. [5]UBB50MultipleProximal tibia“Any,” “tibia,” and “calcaneus” injury
StudyField of studyN PMHSExperimental boundary conditionForce measurementResulting IRC
Yoganandan et al. [28] resampled in 2014 [29]Automotive52Potted at proximal tibia. UnbootedProximal tibia“Any injury”
Crandall et al. [22]Automotive50Potted midfemurProximal tibia“Any injury”
Kuppa et al. [30]Automotive50Whole body data“Tibia” and “calcaneus” injury
Funk et al. [20]Automotive43Potted at proximal tibia. UnbootedMidtibia“Any injury”
Bass et al. [26]UBB20Potted at femur bootedMidtibia“Any injury”
Mckay and Bir [24]UBB18Potted at femur unbootedMidtibia“Any injury”
Henderson et al. [23]UBB18Proximal tibia pottingProximal tibia“Any injury”
Bailey et al. [9]Both82Compiled previous studiesFloor plate“Any injury”
Chirvi et al. [5]UBB50MultipleProximal tibia“Any,” “tibia,” and “calcaneus” injury

Human Body Models in the Underbody Blast Environment.

While whole body and component level tests have provided insight into the UBB injury mechanisms, they are limited in number due to the complex nature of the UBB environment, the cost of physical testing, and the destructive nature of the blast itself. Human body models (HBMs) have emerged as a strong compliment to physical testing and have been widely used in the automotive environment for development of safety regulations [31,32]. Furthermore, HBMs offer an advantage because they are developed from clinical scan data, making them anatomically accurate [33], constitutive material behavior is derived from localized biomechanical testing on PMHS or animal model samples [34,35] and do not need to be ruggedized to withstand the destructive nature of the blast environment unlike the ATDs that are tested. With the decreasing cost of computational resources, HBM's can be used to replicate previously conducted PMHS experiments.

There are several HBMs that have been tested in the UBB environment. The CAVEMAN lower extremity model was developed based on the Zygote 50th percentile male CAD model. The model's lower extremity biofidelity was evaluated against 14 test conditions used in the WIAMan program. Overall the model exhibited good axial force agreement but slightly underestimated axial force in some cases [36]. Hampton et al. developed a lower extremity model [37] using Zygote 50th percentile CAD and compared against pendulum impacts conducted by Gallenberger et al. [27] to determine the force mitigated by a boot. The model achieved good agreement with the experimental data and the combat boot reduced force by 32% in the simulations. Furthermore, the model showed that the boot shifted stress concentrations from the calcaneus in the unbooted to the talus and tibia in the booted cases. This is in agreement with previous findings by Pintar et al. [7] An additional study conducted by Rebelo et al. developed a lower extremity FE model based on a 35-year-old male cadaver. The model response was compared with the PMHS experiments and indicated good agreement with CORA scores of 0.84 for the hind foot forces. The model was then used to compare the force mitigation by a boot and found that the combat boot reduced the injury severity for the lower severity loading cases with lower forces and longer times to peak velocity [38]. While these models have been validated and provide useful insight into evaluation of PPE, to the authors knowledge, none of the models have been used for the development of injury risk curves.

A more widely available HBM is the Global Human Body Models Consortium (GHBMC) average male occupant (M50-O) model. On a component level, the lower extremity has been validated in isolated cases as well as knee-thigh- hip setups. The model has been tested in various loading conditions including lateral bending, axial rotation, and axial compressive loading. The ankle has been specifically studied in impacts for axial loading, ankle inversion, eversion, dorsiflexion, and rotation [20,3941]. Furthermore, the tibia has been validated in a three-point bending setup as well as axial loading for the entire leg [42,43]. These studies are primarily aimed at validation and model assessment for the automotive environment but recently, the model has been used to study injury in the UBB environment. Gabler et al. investigated the GHBMC (v 3.5) response in a single UBB loading condition. The study found that the heel pad material properties were highly sensitive to the rate of loading and affected the peak force response of the model [44]. A study by Weaver et al. investigated the pelvic model response of the GHBMC (v 4.3). The study found that the model response was well correlated with peak forces and strains used to develop injury risk curves from PMHS subjects [45]. A prior study investigated the feasibility of the GHBMC (v 4.5) lower extremity in the UBB environment. The study found that the model was stable and reported peak force and acceleration values in range of previously reported PMHS studies [46]. The GHBMC (v 6.0) lower extremity was validated against UBB PMHS test data from Pintar et al. [7] and Gallenberger et al. [27], and the axial forces and displacements showed good agreement [47].

Despite the increasing use and validity of HBMs in the UBB environment, there are a lack of HBM specific IRC for this environment. An IRC specific to the validated HBM would allow for larger scale design of experiments testing, PPE studies to reduce injury risk, and alternative seating postures to be studied in-silico without the need to conduct additional physical PMHS or ATD tests. Therefore, the goal of this study is to use the GHBMC M50-O lower extremity to develop injury risk curves for the commonly reported UBB injurious outcomes in the lower extremity (“any injury,” “calcaneus injury,” and “tibia injury”) in the UBB environment reported by Chirvi et al. [5] (note that “injury” in this paper refers to bony fracture in the leg or foot).

Methods

Brier Metric Score for Best Predictive Injury Measure.

Often times in injury biomechanics, the number of outputs from an ATD or PMHS test is limited due to the number of load cells that can be instrumented on the test specimen. In the case of PMHS, instrumentation in regions of the anatomy that are complex and highly articulated such as the ankle can disrupt the biomechanical response. Experimentally observed data often consists of global measurements (force or moments) but can also include data from strain gauges that show localized strain. However, depending on where the axial force is measured (knee load cell, midtibia shaft, impacting plate), the magnitude will vary [9]. Strain values can vary depending on where and how the strain gauge is oriented and bonded to the bone [37]. HBMs provide the ability to explore numerous candidate injury metrics ranging from local strain based criteria to global combined criteria such as the Tibia Index [30] without disrupting the biomechanical response. HBMs have the advantage of testing repeatability and do not need to be manufactured to withstand the destructive nature of the UBB environment. However just because various candidate injury metrics can be calculated, it is not necessarily clear which among them would be most suitable for the development of injury risk curves.

To answer the question, the Brier Score Metric (BSM) was used to provide guidance on identifying the most appropriate metric from an experiment that predicts injury outcomes. De-Vogel et al. showed the BSM is a feasible approach to use for collecting multiple outcome metrics and developing injury risk curves for them from the lower extremity [48]. The process of the BSM is outlined in a previously published paper [49] but briefly, the injury risk curve development follows a multistep process that relies on the Akaike information criteria (AIC), dmax distributions, and several other factors when deciding which metric best predicts the true injury outcome of the experiment. The biomechanical metric producing the lowest BMS score is the most appropriate for further analysis and injury risk curve development. A simplified diagram outlining the decision process and methods for metric selection is shown in Fig. 1.

Fig. 1
Simplified diagram outlining the criteria (AIC) and Brier metric score and factors the R studio library package uses to select the most appropriate injury metric
Fig. 1
Simplified diagram outlining the criteria (AIC) and Brier metric score and factors the R studio library package uses to select the most appropriate injury metric
Close modal

The process outlined in Fig. 2 was used as a means to rank a number of candidate metrics for injury risk curve development. A custom developed library package consisting of several open source R studio packages from the CRAN repository including: “Icens,” “survival,” “interval,” and “pROC” was developed for the metric selection process. When evaluating the normalized confidence intervals size (NCIS) values < 0.5 were considered “good,” between 0.5 and 1 were “fair” and 1 to 1.5 were “marginal.” This provides and objective way to evaluate the confidence intervals (CI) on the derived injury risk curves [50]. For covariate consideration, the likelihood ratio test was used to identify significant covariates in the survival model. The model was fit with and without covariates, and the likelihood ratio was compared. If the likelihood ratio test was significant (p < 0.05), the covariates are associated with injury risk and should be included in the model.

Fig. 2
Instrumentation overview of the GHMBC M50-O lower extremity. Force measurements are shown in middle and were used to derive the impulse. Solid and shell strains are shown on the right. Energy was calculated for the tibia, fibula, talus, and calcaneus bones.
Fig. 2
Instrumentation overview of the GHMBC M50-O lower extremity. Force measurements are shown in middle and were used to derive the impulse. Solid and shell strains are shown on the right. Energy was calculated for the tibia, fibula, talus, and calcaneus bones.
Close modal

Global Human Body Models Consortium Lower Extremity Instrumentation.

The HBM instrumentation details are outlined in Fig. 2 and Table 2. Force measurements (middle) were taken at the knee load cell, upper, middle, and lower tibia and fibula, talus, calcaneus, and interface between the plate and the boot/foot depending on if the combat boot was present or not. The force time history was then integrated to derive the impulse at the respective locations. Maximum principal strain was calculated from solid elements at the upper, middle, and lower tibia and fibula on the diaphysis of the bone. Additionally, max principal shell strains were output from the distal tibia and fibula, talus and calcaneus (right). Energy for the tibia, fibula, talus and calcaneus was calculated using LS-DYNA's “matsum” feature. To calculate the energy, LS-DYNA considers the six components of stress and strain and is done incrementally over each element at each time-step output and then summed for a total part energy. The tibia index was calculated at the upper, middle, and lower tibia and is based on previous tibia index definitions [30]. For each force, moment, energy and impulse metric, the peak value was extracted from the simulation then assigned the appropriate censoring based on PMHS injury status for the given test. For the strain measurements, the maximum principal strain was calculated for every solid or shell element. To avoid outlier strain values due to mesh deformation or poor element quality during the simulation, the top 10% of elements were then averaged and the maximum value was taken.

Simulation of Postmortem Human Subjects Tests for Injury Risk Curve Development.

A series of 181 previously conducted PMHS studies were modeled in a matched pair sense to collect biomechanical outcome data based on the GHBMC M50-O v. 6.0. Each simulation was based on a test conducted by either Pintar et al. [7] or Gallenberger et al. [27]. Details of the simulation environment can be found in previous work validating the GHBMC M50-O v. 6.0 model in 33 tests that spanned the range of loading conditions presented here (neutral, dorsi-flexed and plantar-flexed) [51]. Briefly, 77 cases were run in a neutral posture, 56 in the dorsi-flexed posture, and 48 in the plantar-flexed posture. The individual acceleration traces from the experimental tests were integrated and applied as velocity-time curve to the floor plate in the respective simulations. The boot model was based on a Belleville desert combat boot and has been previously validated [52] and fit onto the GHBMC model for use in the UBB environment [53]. Element deletion was turned off for all simulations and the cortical strain of the talus, calcaneus, tibia, and fibula were quantified. Further details of the breakdown of cases by boot status, and injury (single or multiple bony fracture) are given in Fig. 3. Censoring status for the PMHS tests consisted of either right censored for specimens that were noninjured or exact censored for specimens that included repeat testing. Exact censoring was determined by the experimentalists by using acoustic sensors and ability to predict force at fracture [5]. To best match experimental testing censoring, the model data were considered right censored for specimen tests that were noninjured or left censored for specimen tests that were injured. Thus, on a test-by-test basis, the outcome of the test was assigned based on the PMHS test outcome. Interval censoring was tested but not used for two reasons; first because the GHBMC model was not adjusted to match specific specimen mass or age and second because the computational modeling approach cannot (at this time) include the subinjurious test history any other way. Exact censoring was not used for the injury cases since element deletion was not utilized in the model. Accurate modeling of explicit model fracture is complex and highly specimen specific, was not validated for the GHBMC lower extremity model, and was considered beyond the scope of the study. The correlative approach to injury model was instead taken. All cases were run on the DEAC computational cluster using LS-Dyna R 10.2 using 20 processors. The typical simulation run time was 20 min for the shorter pendulum and 2 h for the vertical impacts.

Fig. 3
Overview of simulations conducted by posture (neutral (N) dorsi-flexed (DF) or plantar-flexed (PF)), boot status, and injury status. Injury refers to bony fracture (single or multiple bony fractures).
Fig. 3
Overview of simulations conducted by posture (neutral (N) dorsi-flexed (DF) or plantar-flexed (PF)), boot status, and injury status. Injury refers to bony fracture (single or multiple bony fractures).
Close modal

In total, there were 39 cases that were used for the “any injury” outcome, 30 cases that were used for the “calcaneus injury” outcome and 9 cases that were used for the “tibia injury” outcome. For each injury outcome (“any,” “calcaneus,” and “tibia”), the risk curves of the top three metrics ranked by the BMS were considered for covariates. For each metric the covariate of boot status (booted versus nonbooted) and posture (neutral versus alternate posture) were considered. Due to the smaller sample size and lack of injury data for some off axis tests, posture covariates were grouped into neutral or alternate posture (includes plantar and dorsi-flexed cases). Significance for covariates was set at p < 0.05 using the likelihood ratio test. For each IRC developed the NCIS, mean, upper, and lower bounds were reported for the 5, 10, 25, 50, 75, 90, and 95th percentiles.

Results

Average Response Values for the Different Metrics.

The GHBMC M50-O lower extremity was stable in all 181 simulations. A check of the total mass and energy balance was conducted to ensure that nonphysical mass or energy would not alter the results. For all simulations the percent of added mass to the model was under 1% and total energy was conserved. The peak force values, measured in kN, from the model locations instrumented were extracted and reported for each injury type (“any injury,” “calcaneus injury,” and “tibia injury”) in Table 3. The peak forces for all measured model locations were highest for the “tibia injury” risk curve and lowest for the “calcaneus injury” risk curve. This is hypothesized due to most of the tibia injuries resulting from the booted test cases where the peak velocity and forces were higher compared to the calcaneus injuries, which came primarily from the unbooted, lower velocity test cases.

Table 2

Table for instrumentation and calculation of the 43 metrics that were considered for injury risk curve development

No.LocationMetricMethodNo.MetricMethod
1KneeForceCross section set of elements on load cell part20Lower tibiaForceCross section set of elements on lower tibia
2Moment21Moment
3ImpulseIntegral of knee force versus time22StrainMax principal strain from cross section solid elements
4Upper tibiaForceCross section set of elements on upper tibia23StrainMax principal strain distal shell elements
5Moment24ImpulseIntegral of lower tibia force versus time
6StrainMax principal strain from cross section solid elements25Tibia indexFlowertibFc+MlowertibMc
7ImpulseIntegral of upper tibia force versus time26Lower fibulaForceCross section set of elements on lower fibula
8Tibia indexFuppertibFc+MuppertibMc27Moment
9Upper fibulaForceCross section set of elements on upper fibula28ImpulseIntegral of lower fibula force versus time
10Moment29StrainMax principal strain distal shell elements
11ImpulseIntegral of upper fibula force versus time30Whole tibiaEnergyLS-DYNA “matsum” internal energy for tibia parts
12Middle tibiaForceCross section set of elements on middle tibia31Whole fibulaEnergyLS-DYNA “matsum” internal energy for fibula parts
13Moment32TalusForceCross section set of elements on talus
14StrainMax principal strain from cross section solid elements33Moment
15ImpulseIntegral of middle tibia force versus time34ImpulseIntegral of talus force versus time
16Tibia indexFmiddletibFc+MmiddletibMc35StrainMax principal strain of shell elements
17Middle fibulaForceCross section set of elements on middle fibula36EnergyLS-DYNA “matsum” internal energy
18Moment37CalcaneusForceCross section set of elements on calcaneus
19ImpulseIntegral of middle fibula force versus time38Moment
39ImpulseIntegral of calcaneus force versus time
40StrainMax principal strain of shell elements
41EnergyLS-DYNA “matsum”
42PlateForceContact force between the plate and foot/boot
43ImpulseIntegral of plate force versus time
No.LocationMetricMethodNo.MetricMethod
1KneeForceCross section set of elements on load cell part20Lower tibiaForceCross section set of elements on lower tibia
2Moment21Moment
3ImpulseIntegral of knee force versus time22StrainMax principal strain from cross section solid elements
4Upper tibiaForceCross section set of elements on upper tibia23StrainMax principal strain distal shell elements
5Moment24ImpulseIntegral of lower tibia force versus time
6StrainMax principal strain from cross section solid elements25Tibia indexFlowertibFc+MlowertibMc
7ImpulseIntegral of upper tibia force versus time26Lower fibulaForceCross section set of elements on lower fibula
8Tibia indexFuppertibFc+MuppertibMc27Moment
9Upper fibulaForceCross section set of elements on upper fibula28ImpulseIntegral of lower fibula force versus time
10Moment29StrainMax principal strain distal shell elements
11ImpulseIntegral of upper fibula force versus time30Whole tibiaEnergyLS-DYNA “matsum” internal energy for tibia parts
12Middle tibiaForceCross section set of elements on middle tibia31Whole fibulaEnergyLS-DYNA “matsum” internal energy for fibula parts
13Moment32TalusForceCross section set of elements on talus
14StrainMax principal strain from cross section solid elements33Moment
15ImpulseIntegral of middle tibia force versus time34ImpulseIntegral of talus force versus time
16Tibia indexFmiddletibFc+MmiddletibMc35StrainMax principal strain of shell elements
17Middle fibulaForceCross section set of elements on middle fibula36EnergyLS-DYNA “matsum” internal energy
18Moment37CalcaneusForceCross section set of elements on calcaneus
19ImpulseIntegral of middle fibula force versus time38Moment
39ImpulseIntegral of calcaneus force versus time
40StrainMax principal strain of shell elements
41EnergyLS-DYNA “matsum”
42PlateForceContact force between the plate and foot/boot
43ImpulseIntegral of plate force versus time

Figure 4 shows the force variation for the different body regions that were instrumented in the model for injury cases that comprised the “any injury” risk curve. A general trend of increase in force as velocity increases was observed for each region of the model. The average force in the calcaneus was lower in the booted cases for the 4 and 6 m/s impacts compared to the nonbooted cases. The force in the tibia and talus was higher than the calcaneus indicating the presence of the boot alters the loading pattern from the calcaneus to the tibia. There is also a convergence of the load path from the forefoot and calcaneus into the talus and tibia. A general trend of high forces at the distal end of the leg and lower forces at the proximal end of the leg were noted except for the talus which experienced higher forces than the calcaneus. The fibula tended to increase in force the more proximal the measurement. However, the tibia carried a majority of the load compared to the fibula (87%, 75%, and 72% of the load for the lower, middle, and upper locations respectively). The results below break the risk curves out based on injury type: “any injury,” “calcaneus injury,” and “tibia injury” and explore the significance of covariates (boot use, and posture) for the top 3 metrics (Tables 46).

Fig. 4
Peak force variation for different regions instrumented in the model for the “any injury” nonbooted and booted cases
Fig. 4
Peak force variation for different regions instrumented in the model for the “any injury” nonbooted and booted cases
Close modal

“Any Injury” Outcome.

The BMS ranking for all 43 metrics for the “any injury” IRC is shown in Fig. 5. For the “any injury” risk curve, the calcaneus strain using the Weibull distribution had the lowest BMS score. Calcaneus force using the Weibull distribution was the second-best metric and knee force using the Weibull distribution was the third best metric. Following the censoring approach described in the methods section, the data used to generate the risk curve consisted of 145 right censored data and 36 left censored data points. The injury risk curve and input data for the calcaneus strain along with the NCIS curve are show in Fig. 6. Injury probability, NCIS, mean and 95% CI are presented in Table 4. The NCIS or quality index showed “good” values for all injury probabilities greater than 4% and less than 99%. For “any injury” the type of metric selected (force, moment, strain etc.) by the Brier score metric was varied. The top ten metrics were comprised of strain (1), force (5), tibia index (2) and energy (2). For posthoc analysis, the inclusion of covariates (boot use and posture) was considered for the top three metrics ranked by the BMS. Boot use was a significant covariate, but posture was not for all three metrics. Boot use shifted the injury risk curve to the right by 21%, 13%, and 18% for calcaneus strain, calcaneus force and knee force, respectively, highlighting the effectiveness of the boot. Fit parameters, injury risk curves for the second and third best metrics along with the curves including covariates are shown in the Supplemental Materials on the ASME Digital Collection (Figs. S1–S5 and Tables S1–S3).

Fig. 5
Human body models ranking of all metrics considered for the “any injury” risk curve
Fig. 5
Human body models ranking of all metrics considered for the “any injury” risk curve
Close modal
Fig. 6
Injury risk curves for “any injury” using peak calcaneus strain as the predictor variable (left) and NCIS as a function of injury probability (right). The equation for the IRC is given by y = 1−e−(x0.00573)3.26.
Fig. 6
Injury risk curves for “any injury” using peak calcaneus strain as the predictor variable (left) and NCIS as a function of injury probability (right). The equation for the IRC is given by y = 1−e−(x0.00573)3.26.
Close modal
Table 3

Peak force values (kN) plus standard deviation for injury cases for the three different IRCs developed

Location“Any injury” (n = 39)“Calcaneus injury” (n = 30)“Tibia injury” (n = 9)
Plate10.89 ± 4.3310.46 ± 4.0414.41 ± 3.69
Calcaneus8.11 ± 2.478.14 ± 2.469.83 ± 1.84
Talus9.63 ± 3.079.47 ± 2.9211.95 ± 2.42
Lower tibia9.93 ± 3.169.66 ± 3.1312.22 ± 1.99
Lower fibula1.38 ± 0.971.43 ± 0.931.92 ± 1.25
Middle tibia9.17 ± 2.988.81 ± 3.0611.17 ± 1.71
Middle fibula2.89 ± 1.593.16 ± 1.623.24 ± 1.81
Upper tibia8.73 ± 2.778.39 ± 2.8810.47 ± 1.54
Upper fibula3.45 ± 1.703.77 ± 1.763.63 ± 1.82
Knee7.48 ± 2.447.32 ± 2.399.28 ± 1.83
Location“Any injury” (n = 39)“Calcaneus injury” (n = 30)“Tibia injury” (n = 9)
Plate10.89 ± 4.3310.46 ± 4.0414.41 ± 3.69
Calcaneus8.11 ± 2.478.14 ± 2.469.83 ± 1.84
Talus9.63 ± 3.079.47 ± 2.9211.95 ± 2.42
Lower tibia9.93 ± 3.169.66 ± 3.1312.22 ± 1.99
Lower fibula1.38 ± 0.971.43 ± 0.931.92 ± 1.25
Middle tibia9.17 ± 2.988.81 ± 3.0611.17 ± 1.71
Middle fibula2.89 ± 1.593.16 ± 1.623.24 ± 1.81
Upper tibia8.73 ± 2.778.39 ± 2.8810.47 ± 1.54
Upper fibula3.45 ± 1.703.77 ± 1.763.63 ± 1.82
Knee7.48 ± 2.447.32 ± 2.399.28 ± 1.83

“Calcaneus Injury.”

The BMS ranking for all 43 metrics for the calcaneus IRC is shown in Fig. 7. For the calcaneus IRC, the calcaneus force using the Weibull distribution had the lowest BMS score followed by the upper tibia, tibia index using the Weibull distribution and the upper fibula impulse using the Weibull distribution. Following the censoring approach described in the methods section, the data used to generate the risk curve consisted of 157 right censored data and 24 left censored data. The injury risk curve with input data for the calcaneus force and NCIS curve are shown in Fig. 8. Injury probability, NCIS, mean and 95% CI are presented in Table 5. The NCIS or quality index was slightly lower than the “any injury” risk. For injury risk values between 8% and 86% the NCIS rating was “good” and “fair” for values < 8% or greater than 86%. The type of metric selected by the Brier Metric Score was mixed again for the “calcaneus injury.” The top ten metrics were comprised of three force measurements, two tibia index measurements, two impulse measurements, two moment measurements, and one strain measurement. For posthoc analysis, the inclusion of covariates (boot use and posture) was considered for the top three metrics ranked by the BMS. Boot use was a significant covariate for calcaneus force and the upper tibia, tibia index but not for the upper fibula impulse. Boot use shifted the injury risk curve to the right by 28% for the calcaneus force and 36% for the upper tibia, tibia index. Posture was a significant covariate for the upper fibula impulse but not for calcaneus force or upper tibia, tibia index. The alternate posture shifted the injury risk curve to the right by 67%. Fit parameters, injury risk curves for the second and third best metrics along with the covariates for all curves are shown in the Supplemental Materials on the ASME Digital Collection (Figs. S6–S10 and Tables S4–S6).

Fig. 7
Human body models ranking of all metrics considered for the “calcaneus injury” risk curve
Fig. 7
Human body models ranking of all metrics considered for the “calcaneus injury” risk curve
Close modal
Fig. 8
Injury risk curves for “calcaneus injury” using peak calcaneus force as the predictor variable (left) and NCIS as a function of injury probability (right). The equation for the IRC is given by y = 1−e−(x8.06)3.65.
Fig. 8
Injury risk curves for “calcaneus injury” using peak calcaneus force as the predictor variable (left) and NCIS as a function of injury probability (right). The equation for the IRC is given by y = 1−e−(x8.06)3.65.
Close modal
Table 4

Summary injury and NCIS values for calcaneus strain for the “any injury” IRC

Injury probabilityNCISLower boundMeanUpper bound
0.050.471.83 × 10−32.31 × 10−32.90 × 10−3
0.100.362.41 × 10−32.88 × 10−33.44 × 10−3
0.250.253.45 × 10−33.91 × 10−34.43 × 10−3
0.500.244.54 × 10−35.12 × 10−35.78 × 10−3
0.750.305.45 × 10−36.34 × 10−37.36 × 10−3
0.900.376.17 × 10−37.40 × 10−38.88 × 10−3
0.950.406.57 × 10−38.03 × 10−39.81 × 10−3
Injury probabilityNCISLower boundMeanUpper bound
0.050.471.83 × 10−32.31 × 10−32.90 × 10−3
0.100.362.41 × 10−32.88 × 10−33.44 × 10−3
0.250.253.45 × 10−33.91 × 10−34.43 × 10−3
0.500.244.54 × 10−35.12 × 10−35.78 × 10−3
0.750.305.45 × 10−36.34 × 10−37.36 × 10−3
0.900.376.17 × 10−37.40 × 10−38.88 × 10−3
0.950.406.57 × 10−38.03 × 10−39.81 × 10−3

“Tibia Injury.”

The BMS ranking for all 43 metrics for the tibia IRC is shown in Fig. 9. For the tibia IRC, the lower tibia force using the Weibull distribution had the lowest BMS score followed by the talus force using the Weibull distribution and calcaneus force using the Weibull distribution. Following the censoring approach described in the methods section, the data used to generate the risk curve consisted of 172 right censored data and 9 left censored data. The injury risk curve with input data and NCIS curves are show in Fig. 10. Injury probability, NCIS, mean and 95% CI are presented in Tables 4 and 6. The NCIS or quality index showed “good” values for all injury probabilities between 3% and 98%. For the “tibia injury,” the metrics that were selected by the Brier Metric Score were less varied. Force measurements comprised the top five metrics, and seven of the top ten measurements. Tibia index made up two measurements and a single strain metric was included in the top 10. For posthoc analysis, the inclusion of covariates (boot use and posture) was considered for the top three metrics ranked by the BMS. However, due to the small sample size (only nine injury cases) the survival models were unable to converge for the boot and posture covariates. Therefore, results for the “tibia injury” risk curve with covariates are not presented. Fit parameters and injury risk curves for the second and third best metrics are shown in the Supplemental Materials on the ASME Digital Collection (Figs. S11 and S12 and Tables S7–S9).

Fig. 9
BMS ranking of all metrics considered for the “tibia injury” risk curve
Fig. 9
BMS ranking of all metrics considered for the “tibia injury” risk curve
Close modal
Fig. 10
Injury risk curves for “tibia injury” using peak lower tibia force as the predictor variable (left) and NCIS as a function of injury probability (right). The equation for the IRC is given by y = 1−e−(x7.54)3.49.
Fig. 10
Injury risk curves for “tibia injury” using peak lower tibia force as the predictor variable (left) and NCIS as a function of injury probability (right). The equation for the IRC is given by y = 1−e−(x7.54)3.49.
Close modal
Table 5

Summary injury and NCIS values for calcaneus force for the “calcaneus injury” IRC

Injury probabilityNCISLower boundMeanUpper bound
0.050.602.823.795.10
0.100.443.894.856.04
0.250.295.916.847.91
0.500.327.899.2310.81
0.750.439.4511.7014.48
0.900.5410.6713.9218.16
0.950.6011.3415.2320.45
Injury probabilityNCISLower boundMeanUpper bound
0.050.602.823.795.10
0.100.443.894.856.04
0.250.295.916.847.91
0.500.327.899.2310.81
0.750.439.4511.7014.48
0.900.5410.6713.9218.16
0.950.6011.3415.2320.45
Table 6

Summary injury and NCIS values for the lower tibia force for the “tibia injury” risk curve

Injury probabilityNCISLower boundMeanUpper bound
0.050.407.278.8610.79
0.100.308.589.9511.55
0.250.2210.5111.7113.05
0.500.2511.9113.5015.30
0.750.3312.8015.1117.82
0.900.4113.4116.4020.06
0.950.4513.7117.1121.35
Injury probabilityNCISLower boundMeanUpper bound
0.050.407.278.8610.79
0.100.308.589.9511.55
0.250.2210.5111.7113.05
0.500.2511.9113.5015.30
0.750.3312.8015.1117.82
0.900.4113.4116.4020.06
0.950.4513.7117.1121.35
Table 7

Parameters used for the top three metrics selected by the BMS for the “any injury” risk curve for the Weibull distribution

MetricCovariateβ0β1λK
Calcaneus strainAny−5.16N/A5.73 × 10−33.26
Booted−5.260.246.59 × 10−33.65
Unbooted5.18 × 10−3
Neutral−5.152.52 × 10−35.77 × 10−33.72
Alternate5.79 × 10−3
Calcaneus forceAny2.09N/A8.063.65
Booted2.030.148.834.38
Unbooted7.66
Neutral2.086.13 × 10−28.014.14
Alternate8.51
Knee forceAny2.02N/A7.543.49
Booted1.930.208.404.36
Unbooted6.91
Neutral1.980.107.274.08
Alternate8.04
MetricCovariateβ0β1λK
Calcaneus strainAny−5.16N/A5.73 × 10−33.26
Booted−5.260.246.59 × 10−33.65
Unbooted5.18 × 10−3
Neutral−5.152.52 × 10−35.77 × 10−33.72
Alternate5.79 × 10−3
Calcaneus forceAny2.09N/A8.063.65
Booted2.030.148.834.38
Unbooted7.66
Neutral2.086.13 × 10−28.014.14
Alternate8.51
Knee forceAny2.02N/A7.543.49
Booted1.930.208.404.36
Unbooted6.91
Neutral1.980.107.274.08
Alternate8.04
Table 8

Summary injury values for calcaneus force for the “any injury” risk curve

Injury probabilityNCISLower boundMeanUpper bound
0.050.492.803.574.54
0.100.373.614.355.24
0.250.245.075.736.47
0.500.206.597.298.06
0.750.247.818.829.95
0.900.308.7210.1311.77
0.950.349.2110.8912.88
Injury probabilityNCISLower boundMeanUpper bound
0.050.492.803.574.54
0.100.373.614.355.24
0.250.245.075.736.47
0.500.206.597.298.06
0.750.247.818.829.95
0.900.308.7210.1311.77
0.950.349.2110.8912.88
Table 9

Summary injury values for knee force for the “any injury” risk curve

Injury probabilityNCISLower boundMeanUpper bound
0.050.522.493.224.17
0.100.403.253.964.82
0.250.254.665.285.98
0.500.216.116.797.54
0.750.277.268.289.46
0.900.348.109.5811.32
0.950.388.5610.3312.47
Injury probabilityNCISLower boundMeanUpper bound
0.050.522.493.224.17
0.100.403.253.964.82
0.250.254.665.285.98
0.500.216.116.797.54
0.750.277.268.289.46
0.900.348.109.5811.32
0.950.388.5610.3312.47
Table 10

Parameters used for the top three metrics selected by the BMS for the “calcaneus injury” risk curve for the Weibull distribution

MetricCovariateβ0β1λK
Calcaneus forceAny2.35N/A10.472.93
Booted2.140.3311.873.54
Unbooted8.54
Neutral2.37−0.0510.672.92
Alternate10.17
Upper tibia, tibia indexAny0.88N/A2.412.49
Booted0.590.452.823.07
Unbooted1.80
Neutral0.770.212.172.52
Alternate2.68
Upper fibula impulseAny3.31N/A27.332.22
Booted3.210.2732.412.16
Unbooted24.75
Neutral2.970.5119.552.42
Alternate32.58
MetricCovariateβ0β1λK
Calcaneus forceAny2.35N/A10.472.93
Booted2.140.3311.873.54
Unbooted8.54
Neutral2.37−0.0510.672.92
Alternate10.17
Upper tibia, tibia indexAny0.88N/A2.412.49
Booted0.590.452.823.07
Unbooted1.80
Neutral0.770.212.172.52
Alternate2.68
Upper fibula impulseAny3.31N/A27.332.22
Booted3.210.2732.412.16
Unbooted24.75
Neutral2.970.5119.552.42
Alternate32.58
Table 11

Summary injury values for upper tibia, tibia index for the “calcaneus injury” risk curve

Injury probabilityNCISLower boundMeanUpper bound
0.050.680.520.731.02
0.100.500.760.981.25
0.250.341.231.461.73
0.500.371.732.082.50
0.750.502.142.743.51
0.900.622.483.364.57
0.950.692.673.745.24
Injury probabilityNCISLower boundMeanUpper bound
0.050.680.520.731.02
0.100.500.760.981.25
0.250.341.231.461.73
0.500.371.732.082.50
0.750.502.142.743.51
0.900.622.483.364.57
0.950.692.673.745.24
Table 12

Summary injury values for upper fibula impulse for the ‘calcaneus injury’ risk curve

Injury probabilityNCISLower boundMeanUpper bound
0.050.764.947.1710.41
0.100.577.509.9213.11
0.250.3912.8615.5918.91
0.500.4318.7623.1728.62
0.750.5723.9231.6741.92
0.900.7028.1739.8056.23
0.950.7830.5944.8165.65
Injury probabilityNCISLower boundMeanUpper bound
0.050.764.947.1710.41
0.100.577.509.9213.11
0.250.3912.8615.5918.91
0.500.4318.7623.1728.62
0.750.5723.9231.6741.92
0.900.7028.1739.8056.23
0.950.7830.5944.8165.65
Table 13

Parameters used for the top three metrics selected by the BMS for the “tibia injury” risk curve for the Weibull distribution

MetricCovariateβ0β1λK
Lower tibia forceAny2.66N/A14.336.18
Talus forceAny2.65N/A14.145.71
Calcaneus forceAny2.46N/A11.685.81
MetricCovariateβ0β1λK
Lower tibia forceAny2.66N/A14.336.18
Talus forceAny2.65N/A14.145.71
Calcaneus forceAny2.46N/A11.685.81
Table 14

Summary injury values for talus force for the “tibia injury” risk curve

Injury probabilityNCISLower boundMeanUpper bound
0.050.416.878.4110.28
0.100.318.179.5311.12
0.250.2310.1111.3712.78
0.500.2711.5913.2615.16
0.750.3512.5914.9717.80
0.900.4213.2816.3620.15
0.950.4613.6317.1321.52
Injury probabilityNCISLower boundMeanUpper bound
0.050.416.878.4110.28
0.100.318.179.5311.12
0.250.2310.1111.3712.78
0.500.2711.5913.2615.16
0.750.3512.5914.9717.80
0.900.4213.2816.3620.15
0.950.4613.6317.1321.52
Table 15

Summary injury values for calcaneus force for the “tibia injury” risk curve

Injury probabilityNCISLower boundMeanUpper bound
0.050.425.687.008.63
0.100.316.787.939.27
0.250.238.409.4310.58
0.500.279.5710.9712.57
0.750.3610.3112.3614.82
0.900.4510.8113.4916.82
0.950.4911.0714.1117.99
Injury probabilityNCISLower boundMeanUpper bound
0.050.425.687.008.63
0.100.316.787.939.27
0.250.238.409.4310.58
0.500.279.5710.9712.57
0.750.3610.3112.3614.82
0.900.4510.8113.4916.82
0.950.4911.0714.1117.99

Discussion

The GHBMC M50-O v 6.0 lower extremity was simulated in matched-pair UBB like loading conditions based on physical PMHS experiments to develop injury risk curves for three distinct outcome measures: “any injury,” “calcaneus injury,” and “tibia injury.” In contrast to the PMHS experiments in which the data collection is limited by specific instrumentation (knee load cells and tibia or calcaneus accelerometers) which can be cumbersome to collect and may alter the biomechanical response of the specimen, the GHBMC was instrumented virtually to examine 43 different outcome metrics. Metrics considered were force, moment, energy, impulse, strain, and the tibia index, and these measurements were collected at specific bones (tibia, fibula, calcaneus, and talus). This instrumentation approach with HBM's provides insight into forces that are not easily measured in PMHS experiments without disrupting the biomechanical response.

It was hypothesized that more metrics that are not easily measured in PMHS experiments such as strain might be a better injury predictor. However, this was not always the case based on the Brier Metric Score method, which was used to rank and determine the most appropriate metric. For two of the three injury risk curves developed, the best performing metric was based on force. Calcaneus strain was selected for the “any injury” risk curve but calcaneus force was the second-best metric. The calcaneus force was selected for the “calcaneus injury” risk curve. This is hypothesized because the calcaneus is one of the first bones to be loaded from the floor plate and the force and strain measured through the bone would be most accurate for injury prediction. The BMS metric selection supports this region-specific outcome metric because from the experimental data, calcaneus injuries occurred in 77% of the “any injury” category. For the “tibia injury” risk curve, the lower tibia force metric was selected. This is hypothesized for a similar reason as the distal tibia experiences the highest amount of force in the tibia and is the first part of the tibia to experience the loading imparted from the calcaneus and talus. Danelson et al. noted that in OIF and OEF, pilon fractures occurred in 29% of all distal tibia fractures [6]. Pilon fractures are high energy fractures and often caused by the talus loading into the tibia. The force variation data and BMS ranking supports this injury mechanism as the talus experiences a higher load than the calcaneus and is similar in magnitude to the lower tibia.

Human body models have been used in the UBB environment; however, the use has been limited to exploration and validation studies. This study is unique in that model specific, and injury specific risk curves were developed. Often times, models are used in conjunction with injury risk curves developed from PMHS studies. However, the injury risk curves can be limited to the application and use cases they were developed in. The results also indicate the importance of the boot when studying injury. Boot use was found to be a significant covariate for the top three risk curves for “any injury” and two out of the top three curves for “calcaneus injury.” The use of a model specific injury risk curve can be used to guide design of experiments studies where physical testing is not possible and provide accurate injury outcome predictions based on a wide variety of PPE or environmental design modifications. Furthermore, the covariate inclusion can be used to study the effect of force mitigation by the boot and aid future boot material improvements, or alternate seating posture in vehicles.

It is important to put the output risk curves in context. Compared to other injury risk curves developed from PMHS studies developed for the automotive or UBB communities, the model falls within the range of reported values. The GHBMC risk curves are presented below in Fig. 11 are for the best (lowest) BMS force location. The any, calcaneus, and tibia risk curves are based on force measured at the calcaneus, calcaneus, and lower tibia, respectively. The Chirvi data included age as a significant covariate and the comparison is based on a 40-year-old subject. The “any injury” risk curve is slightly more conservative compared to the Chirvi PMHS data but falls well within the range of reported values in the literature. The GHBMC “calcaneus injury” risk curve is comprised of both minor and major calcaneus injuries and for most of the injury probabilities falls between the minor and major specific curves. The GHBMC “tibia injury” curve also consists of both tibia I and tibia II injuries. The curve is shifted slightly rightward at higher injury probabilities. This is hypothesized because the force is measured at the lower tibia which will experience a higher force compared to the knee load cell force that the Chirvi tibia I and II risk curves were developed from. The GHBMC model also shows sensitivity to specific injuries and highlights the difference in magnitude for the different injuries. The resulting risk curves in the paper may very slightly than those developed using another FE HBM; however, given the vast amount of previous PMHS injury risk curves developed in Fig. 11, the GHBMC model curves are well within the range of reported values.

Fig. 11
Comparison of the GHBMC risk curves developed for any, calcaneus, and tibia injuries to previously reported curves
Fig. 11
Comparison of the GHBMC risk curves developed for any, calcaneus, and tibia injuries to previously reported curves
Close modal
Fig. 12
Injury risk curves for “any injury” using peak calcaneus force as the predictor variable (left) and NCIS as a function of injury probability (right)
Fig. 12
Injury risk curves for “any injury” using peak calcaneus force as the predictor variable (left) and NCIS as a function of injury probability (right)
Close modal
Fig. 13
Injury risk curves for “any injury” using peak knee force as the predictor variable (left) and NCIS as a function of injury probability (right)
Fig. 13
Injury risk curves for “any injury” using peak knee force as the predictor variable (left) and NCIS as a function of injury probability (right)
Close modal
Fig. 14
Calcaneus strain covariates for “any injury.” Boot use (significant) is shown on the left and posture (not significant) on the right
Fig. 14
Calcaneus strain covariates for “any injury.” Boot use (significant) is shown on the left and posture (not significant) on the right
Close modal
Fig. 15
Calcaneus force covariates for “any injury.” Boot use (significant) is shown on the left and posture (not significant) on the right
Fig. 15
Calcaneus force covariates for “any injury.” Boot use (significant) is shown on the left and posture (not significant) on the right
Close modal
Fig. 16
Knee force covariates for “any injury.” Boot use (significant) is shown on the left and posture (not significant) on the right
Fig. 16
Knee force covariates for “any injury.” Boot use (significant) is shown on the left and posture (not significant) on the right
Close modal
Fig. 17
Injury risk curves for “calcaneus injury” using upper tibia, tibia index as the best predictor variable (left) and NCIS as a function of injury probability (right)
Fig. 17
Injury risk curves for “calcaneus injury” using upper tibia, tibia index as the best predictor variable (left) and NCIS as a function of injury probability (right)
Close modal
Fig. 18
Injury risk curves for “calcaneus injury” using upper fibula impulse as the best predictor variable (left) and NCIS as a function of injury probability (right)
Fig. 18
Injury risk curves for “calcaneus injury” using upper fibula impulse as the best predictor variable (left) and NCIS as a function of injury probability (right)
Close modal
Fig. 19
Calcaneus force covariates for “calcaneus injury.” Boot use (significant) is shown on the left and posture (not significant) on the right
Fig. 19
Calcaneus force covariates for “calcaneus injury.” Boot use (significant) is shown on the left and posture (not significant) on the right
Close modal
Fig. 20
Upper tibia, tibia index covariates for “calcaneus injury.” Boot use (significant) is shown on the left and posture (not significant) on the right
Fig. 20
Upper tibia, tibia index covariates for “calcaneus injury.” Boot use (significant) is shown on the left and posture (not significant) on the right
Close modal
Fig. 21
Upper fibula impulse covariates for “calcaneus injury.” Boot use (not significant) is shown on the left and posture (significant) on the right
Fig. 21
Upper fibula impulse covariates for “calcaneus injury.” Boot use (not significant) is shown on the left and posture (significant) on the right
Close modal
Fig. 22
Injury risk curves for “tibia injury” using talus force as the predictor variable (left) and NCIS as a function of injury probability (right)
Fig. 22
Injury risk curves for “tibia injury” using talus force as the predictor variable (left) and NCIS as a function of injury probability (right)
Close modal
Fig. 23
Injury risk curves for “tibia injury” using calcaneus force as the predictor variable (left) and NCIS as a function of injury probability (right)
Fig. 23
Injury risk curves for “tibia injury” using calcaneus force as the predictor variable (left) and NCIS as a function of injury probability (right)
Close modal

Limitations

The authors acknowledge that there are some limitations to the study. First, the underlying experimental data was collected from older subjects. It is commonly known that fracture tolerance decreases with age and BMD and both have been shown to be a significant effect on lower extremity injury risk curves [9,54,55]. The model age was not adjusted for specific subjects and the injury outcomes from the PMHS test data were based on older subjects. This effect is seen in the “any injury” risk curve as it is shifted to the left compared to the 40-year-old Chirvi injury curve. Furthermore, factors such as sex and anthropometry also play a role in fracture risk and injury outcomes [56]. However, the goal of this work was to use the validated GHBMC 50th percentile male model to develop model specific injury risk curves that could be used for injury prediction. The GHBMC is “ageless” in that the material properties are comprised of underlying data from all different ages. Aged adjusted GHBMC models have been developed [57,58] and can be used in future studies to address the question of aging and anthropometry. The material properties of the model could be adjusted to capture age effects but was outside of the scope of the present study. Additionally, Chirvi et al. showed that the injury risk curve is sensitive to the type of injury that is being predicted. The current study did not distinguish between severity of calcaneus or tibia injuries. The goal was to provide baseline injury risk curves that model users and PPE manufactures could use for injury prediction. Furthermore, modeling and predicting fracture in a deterministic sense can depend on several factors including mesh density, bone material properties, and failure stresses and strains. Therefore, a probabilistic approach was used to determine injury risk values. Furthermore, given stability concerns and lack of validated element elimination criteria, it is assumed the general usage of the GHBMC model in this environment would be to disable element elimination (explicit simulation of fracture).

Conclusion

While a great deal of work has been done in the development and validation of injury risk data for PMHS, relatively little such work has been done for computational human body models. The GHBMC M50-O v. 6.0 lower extremity was used to simulate 181 cases under two loading regimes for which it was previously validated. Injury outcome in a matched pair sense was used to develop risk curves for three distinct injury outcomes, “any injury,” “calcaneus injury,” and “tibia injury.” 43 unique biomechanical metrics were instrumented on the model and ranked using a previously published method based on the Brier Metric Score. The best performing metric for the “any injury,” ”calcaneus injury,” and “tibia injury” cases were calcaneus strain, calcaneus force, and lower tibia force, respectively. For the six different injury risk curves were covariates were considered, the presence of the boot was found to be a significant covariate reducing injury risk in five out of six cases. Posture was a significant for only one curve. The injury risk curves developed from this study can serve as a baseline for model injury prediction, PPE evaluation, and can aid in larger scale testing and experimental protocols

Acknowledgment

Funding for this work was provided by U.S. Army Med R & D Command, under BAA W911NF-17-S-0003, Exploring Physics-based Finite Element Analysis of Service Members Subjected to Extreme Environments. The authors acknowledge the contributions and support of Dr. Michael Kleinberger, Dr. Caitlin Weaver of the U.S. Army Futures Command. The authors also acknowledge Dr. Frank Pintar and Mr. Michael Schlick at the Medical College of Wisconsin for their help and guidance with all experimental data. Drs. F. Scott Gayzik and Zachary Hostetler are members of Elemance, LLC which provides academic and commercial licenses of GHBMC-owned models.

Funding Data

  • U.S. Army Med R & D Command (Grant No. BAA W911NF-17-S-0003; Funder ID: 10.13039/100000182).

Data Availability Statement

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.

References

1.
Ramasamy
,
A.
,
Masouros
,
S. D.
,
Newell
,
N.
,
Hill
,
A. M.
,
Proud
,
W. G.
,
Brown
,
K. A.
,
Bull
,
A. M.
, and
Clasper
,
J. C.
,
2011
, “
In-Vehicle Extremity Injuries From Improvised Explosive Devices: Current and Future Foci
,”
Philos. Trans. R. Soc., B
,
366
(
1562
), pp.
160
170
.10.1098/rstb.2010.0219
2.
Singleton
,
J. A.
,
Gibb
,
I. E.
,
Hunt
,
N. C.
,
Bull
,
A. M.
, and
Clasper
,
J. C.
,
2013
, “
Identifying Future ‘Unexpected’ Survivors: A Retrospective Cohort Study of Fatal Injury Patterns in Victims of Improvised Explosive Devices
,”
BMJ Open
,
3
(
8
), p.
e003130
.10.1136/bmjopen-2013-003130
3.
Owens
,
B. D.
,
Kragh
,
J. F.
, Jr
,
Wenke
,
J. C.
,
Macaitis
,
J.
,
Wade
,
C. E.
, and
Holcomb
,
J. B.
,
2008
, “
Combat Wounds in Operation Iraqi Freedom and Operation Enduring Freedom
,”
J. Trauma Acute Care Surg.
,
64
(
2
), pp.
295
299
.10.1097/TA.0b013e318163b875
4.
Ramasamy
,
A.
,
Hill
,
A.-M.
, and
Clasper
,
J. C.
,
2009
, “
Improvised Explosive Devices: Pathophysiology, Injury Profiles and Current Medical Management
,”
BMJ Mil. Health
,
155
(
4
), pp.
265
272
.10.1136/jramc-155-04-05
5.
Chirvi
,
S.
,
Pintar
,
F.
,
Yoganandan
,
N.
,
Banerjee
,
A.
,
Schlick
,
M.
,
Curry
,
W.
, and
Voo
,
L.
,
2017
, “
Human Foot-Ankle Injuries and Associated Risk Curves From Under Body Blast Loading Conditions
,”
Stapp Car Crash J.
,
61
, pp.
157
173
.10.4271/2017-22-0006
6.
Danelson
,
K.
,
Watkins
,
L.
,
Hendricks
,
J.
,
Frounfelker
,
P.
,
Pizzolato-Heine
,
K.
,
Valentine
,
R.
, and
Loftis
,
K.
,
2018
, “
Analysis of the Frequency and Mechanism of Injury to Warfighters in the Under-Body Blast Environment
,”
SAE
Paper No. 2018-22-0014.10.4271/2018-22-0014
7.
Pintar
,
F. A.
,
Schlick
,
M. B.
,
Yoganandan
,
N.
,
Voo
,
L.
,
Merkle
,
A. C.
, and
Kleinberger
,
M.
,
2016
, “
Biomechanical Response of Military Booted and Unbooted Foot-Ankle-Tibia From Vertical Loading
,”
SAE
Paper No. 2016-22-0010.10.4271/2016-22-0010
8.
Yoganandan
,
N.
,
Pintar
,
F. A.
,
Schlick
,
M.
,
Humm
,
J. R.
,
Voo
,
L.
,
Merkle
,
A.
, and
Kleinberger
,
M.
,
2015
, “
Vertical Accelerator Device to Apply Loads Simulating Blast Environments in the Military to Human Surrogates
,”
J. Biomech.
,
48
(
12
), pp.
3534
3538
.10.1016/j.jbiomech.2015.06.008
9.
Bailey
,
A. M.
,
McMurry
,
T. L.
,
Poplin
,
G. S.
,
Salzar
,
R. S.
, and
Crandall
,
J. R.
,
2015
, “
Survival Model for Foot and Leg High Rate Axial Impact Injury Data
,”
Traffic Injury Prev.
,
16
(
sup2
), pp.
S96
S102
.10.1080/15389588.2015.1061185
10.
Ramasamy
,
A.
,
Hill
,
A. M.
,
Masouros
,
S.
,
Gibb
,
I.
,
Bull
,
A. M.
, and
Clasper
,
J. C.
,
2011
, “
Blast-Related Fracture Patterns: A Forensic Biomechanical Approach
,”
J. R. Soc. Interface
,
8
(
58
), pp.
689
698
.10.1098/rsif.2010.0476
11.
Bird
,
R.
,
Swinton
,
R.
, and
Krstic
,
A.
,
2001
, “
Protection of Lower Limbs Against Floor Impact in Army Vehicles Experiencing Landmine Explosion
,”
J. Battlefield Technol.
,
4
, pp.
8
12
.10.21203/rs.3.rs-1765590/v1
12.
Pietsch
,
H. A.
,
Bosch
,
K. E.
,
Weyland
,
D. R.
,
Spratley
,
E. M.
,
Henderson
,
K. A.
,
Salzar
,
R. S.
,
Smith
,
T. A.
, et al.,
2016
, “
Evaluation of WIAMan Technology Demonstrator Biofidelity Relative to Sub-Injurious PMHS Response in Simulated Under-Body Blast Events
,”
Stapp Car Crash J.
,
60
, p.
199
.
13.
Scherer
,
R.
,
Felczak
,
C.
, and
Halstad
,
S.
,
2011
, “
Vehicle and Crash Dummy Response to an Underbelly Blast Event
,” RDECOM-TARDEC Unclassified Public Release.
14.
Gayzik
,
F.
,
Marcus
,
I.
,
Danelson
,
K.
,
Rupp
,
J.
,
Bass
,
C.
,
Yoganandan
,
N.
, and
Zhang
,
J.
,
2015
, “
A Point-Wise Normalization Method for Development of Biofidelity Response Corridors
,”
J. Biomech.
,
48
(
15
), pp.
4173
4177
.10.1016/j.jbiomech.2015.09.017
15.
Pietsch
,
H. A.
,
Bosch
,
K. E.
,
Weyland
,
D. R.
,
Spratley
,
E. M.
,
Henderson
,
K. A.
,
Salzar
,
R. S.
,
Smith
,
T. A.
, et al.,
2016
, “
Evaluation of WIAMan Technology Demonstrator Biofidelity Relative to Sub-Injurious PMHS Response in Simulated Under-Body Blast Events
,”
SAE
Paper No. 2016-22-0009.10.4271/2016-22-0009
16.
Marie Bailey
,
A.
,
Christopher
,
J. J.
,
Salzar
,
R. S.
, and
Brozoski
,
F.
,
2015
, “
Comparison of Hybrid-III and Postmortem Human Surrogate Response to Simulated Underbody Blast Loading
,”
ASME J. Biomech. Eng.
,
137
(
5
), p.
051009
.10.1115/1.4029981
17.
Danelson
,
K. A.
,
Kemper
,
A. R.
,
Mason
,
M. J.
,
Tegtmeyer
,
M.
,
Bolte
, IV
,
J. H.
, and
Hardy
,
W. N.
,
2015
, “
Comparison of ATD to PMHS Response in the Under-Body Blast Environment
,”
Stapp Car Crash J.
,
59
, p.
445
.10.4271/2015-22-0017
18.
Barbat
,
S.
,
Fu
,
Y.
,
Zhan
,
Z.
,
Yang
,
R.-J.
, and
Gehre
,
C.
,
2013
, “
Objective Rating Metric for Dynamic Systems
,” 23rd International Technical Conference on the Enhanced Safety of Vehicles (
ESV
), Seoul, Republic Korea, May 27–30, Paper No. 13–0448.https://wwwesv.nhtsa.dot.gov/Proceedings/23/files/23ESV-000448.PDF
19.
Ott
,
K.
,
Drewry
,
D.
,
Luongo
,
M.
,
Andrist
,
J.
,
Armiger
,
R.
,
Titus
,
J.
, and
Demetropoulos
,
C.
,
2020
, “
Comparison of Human Surrogate Responses in Underbody Blast Loading Conditions
,”
ASME J. Biomech. Eng.
,
142
(
9
), p.
091010
.10.1115/1.4046638
20.
Funk
,
J. R.
,
Crandall
,
J. R.
,
Tourret
,
L. J.
,
MacMahon
,
C. B.
,
Bass
,
C. R.
,
Patrie
,
J. T.
,
Khaewpong
,
N.
, and
Eppinger
,
R. H.
,
2002
, “
The Axial Injury Tolerance of the Human Foot/Ankle Complex and the Effect of Achilles Tension
,”
ASME J. Biomech. Eng.
,
124
(
6
), pp.
750
757
.10.1115/1.1514675
21.
Yoganandan
,
N.
,
Pintar
,
F. A.
,
Gennarelli
,
T. A.
,
Seipel
,
R.
, and
Marks
,
R.
,
1999
, “
Biomechanical Tolerance of Calcaneal Fractures
,”
Proceedings of Annual Proceedings/Association for the Advancement of Automotive Medicine, Association for the Advancement of Automotive Medicine
, Barcelona, Spain, Oct., p.
345
.https://trid.trb.org/view/639786
22.
Crandall
,
J. R.
,
Kuppa
,
S. M.
,
Klopp
,
G.
,
Hall
,
G.
,
Pilkey
,
W. D.
, and
Hurwitz
,
S. R.
,
1998
, “
Injury Mechanisms and Criteria for the Human Foot and Ankle Under Axial Impacts to the Foot
,”
Int. J. Crashworthiness
,
3
(
2
), pp.
147
162
.10.1533/cras.1998.0068
23.
Henderson
,
K.
,
Bailey
,
A. M.
,
Christopher
,
J.
,
Brozoski
,
F.
, and
Salzar
,
R. S.
,
2013
, “
Biomechanical Response of the Lower Leg Under High Rate Loading
,”
Proceedings of IRCOBI Conference
, Sweden, Sept. 9, pp.
13
24
.http://www.ircobi.org/wordpress/downloads/irc13/pdf_files/24.pdf
24.
McKay
,
B. J.
, and
Bir
,
C. A.
,
2009
, “
Lower Extremity Injury Criteria for Evaluating Military Vehicle Occupant Injury in Underbelly Blast Events
,”
Stapp Car Crash J.
,
53
, p.
229
.10.4271/2009-22-0009
25.
Quenneville
,
C. E.
,
Fraser
,
G. S.
, and
Dunning
,
C. E.
,
2010
, “
Development of an Apparatus to Produce Fractures From Short-Duration High-Impulse Loading With an Application in the Lower Leg
,”
ASME J. Biomech. Eng.
,
132
(
1
), p.
014502
.10.1115/1.4000084
26.
Bass
,
C.
,
Folk
,
B.
,
Salzar
,
R.
,
Davis
,
M.
,
Harris
,
R.
,
Rountree
,
M. S.
,
Harcke
,
T.
, et al.,
2004
,
Development of a Test Methodology to Evaluate Mine Protective Footwear
,
Virginia University
,
Charlottesville, VA
.
27.
Gallenberger
,
K.
,
Yoganandan
,
N.
, and
Pintar
,
F.
,
2013
, “
Biomechanics of Foot/Ankle Trauma With Variable Energy Impacts
,”
Ann. Adv. Automot. Med.
,
57
, p.
123
.https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3861818/
28.
Yoganandan
,
N.
,
Pintar
,
F. A.
,
Boynton
,
M.
,
Begeman
,
P.
,
Prasad
,
P.
,
Kuppa
,
S. M.
,
Morgan
,
R. M.
, and
Eppinger
,
R. H.
,
1996
, “
Dynamic Axial Tolerance of the Human Foot-Ankle Complex
,”
SAE Trans.
, 105, pp.
1887
1898
.10.4271/962426
29.
Yoganandan
,
N.
,
Arun
,
M. W.
,
Pintar
,
F. A.
, and
Szabo
,
A.
,
2014
, “
Optimized Lower Leg Injury Probability Curves From Postmortem Human Subject Tests Under Axial Impacts
,”
Traffic Injury Prev.
,
15
(
sup1
), pp.
S151
S156
.10.1080/15389588.2014.935357
30.
Kuppa
,
S.
,
Wang
,
J.
,
Haffner
,
M.
, and
Eppinger
,
R.
,
2001
, “
Lower Extremity Injuries and Associated Injury Criteria
,”
SAE
Paper No. 2001-06-0160.10.4271/2001-06-0160
31.
Decker
,
W.
,
Koya
,
B.
,
Pak
,
W.
,
Untaroiu
,
C. D.
, and
Gayzik
,
F. S.
,
2019
, “
Evaluation of Finite Element Human Body Models for Use in a Standardized Protocol for Pedestrian Safety Assessment
,”
Traffic Injury Prev.
,
20
(
sup2
), pp.
S32
S36
.10.1080/15389588.2019.1637518
32.
Meyer
,
F.
,
Humm
,
J.
,
Yoganandan
,
N.
,
Leszczynski
,
A.
,
Bourdet
,
N.
,
Deck
,
C.
, and
Willinger
,
R.
,
2021
, “
Development of a Detailed Human Neck Finite Element Model and Injury Risk Curves Under Lateral Impact
,”
J. Mech. Behav. Biomed. Mater.
,
116
, p.
104318
.10.1016/j.jmbbm.2021.104318
33.
Gayzik
,
F. S.
,
Moreno
,
D. P.
,
Vavalle
,
N. A.
,
Rhyne
,
A. C.
, and
Stitzel
,
J. D.
,
2011
, “
Development of the Global Human Body Models Consortium Mid-Sized Male Full Body Model
,”
Proceedings of International Workshop on Human Subjects for Biomechanical Research
,
National Highway Traffic Safety Administration,
Washington, DC
, Nov., 39(12).https://wwwnrd.nhtsa.dot.gov/pdf/bio/proceedings/2011_39/39-12.pdf
34.
Comley
,
K.
, and
Fleck
,
N.
,
2012
, “
The Compressive Response of Porcine Adipose Tissue From Low to High Strain Rate
,”
Int. J. Impact Eng.
,
46
, pp.
1
10
.10.1016/j.ijimpeng.2011.12.009
35.
Katzenberger
,
M. J.
,
Albert
,
D. L.
,
Agnew
,
A. M.
, and
Kemper
,
A. R.
,
2020
, “
Effects of Sex, Age, and Two Loading Rates on the Tensile Material Properties of Human Rib Cortical Bone
,”
J. Mech. Behav. Biomed. Mater.
,
102
, p.
103410
.10.1016/j.jmbbm.2019.103410
36.
Butz
,
K.
,
Spurlock
,
C.
,
Roy
,
R.
,
Bell
,
C.
,
Barrett
,
P.
,
Ward
,
A.
,
Xiao
,
X.
, et al.,
2017
, “
Development of the CAVEMAN Human Body Model: Validation of Lower Extremity Sub-Injurious Response to Vertical Accelerative Loading
,”
Stapp Car Crash J.
,
61
, pp.
175
209
.10.4271/2017-22-0007
37.
Hampton
,
C. E.
,
Kleinberger
,
M.
,
Schlick
,
M.
,
Yoganandan
,
N.
, and
Pintar
,
F. A.
,
2020
, “
Analysis of Force Mitigation by Boots in Axial Impacts Using a Lower Leg Finite Element Model
,”
SAE
Paper No. 2019-22-0011.10.4271/2019-22-0011
38.
Rebelo
,
E. A.
,
Grigoriadis
,
G.
,
Carpanen
,
D.
,
Bull
,
A. M.
, and
Masouros
,
S. D.
,
2021
, “
An Experimentally Validated Finite Element Model of the Lower Limb to Investigate the Efficacy of Blast Mitigation Systems
,”
Front. Bioeng. Biotechnol.
,
9
, p.
410
.10.3389/fbioe.2021.665656
39.
Funk
,
J. R.
,
Tourret
,
L. J.
,
George
,
S. E.
, and
Crandall
,
J. R.
,
2000
, “
The Role of Axial Loading in Malleolar Fractures
,”
SAE Trans.
, 109, pp.
212
223
.https://www.jstor.org/stable/44686868
40.
Rudd
,
R.
,
Crandall
,
J.
,
Millington
,
S.
,
Hurwitz
,
S.
, and
Höglund
,
N.
,
2004
, “
Injury Tolerance and Response of the Ankle Joint in Dynamic Dorsiflexion
,”
SAE
Paper No. 2004-22-0001.10.4271/2004-22-0001
41.
Siegler
,
S.
,
Chen
,
J.
, and
Schneck
,
C.
,
1988
, “
The Three-Dimensional Kinematics and Flexibility Characteristics of the Human Ankle and Subtalar Joints—Part I: Kinematics
,”
ASME J. Biomech. Eng.
, 110(4), pp.
364
373
.10.1115/1.3108455
42.
Kerrigan
,
J. R.
,
Bhalla
,
K. S.
,
Madeley
,
N. J.
,
Funk
,
J. R.
,
Bose
,
D.
, and
Crandall
,
J. R.
,
2003
, “
Experiments for Establishing Pedestrian-Impact Lower Limb Injury Criteria
,”
SAE
Paper No. 0148-7191.10.4271/0148-7191
43.
Untaroiu
,
C.
,
Darvish
,
K.
,
Crandall
,
J.
,
Deng
,
B.
, and
Wang
,
J.-T.
,
2005
, “
A Finite Element Model of the Lower Limb for Simulating Pedestrian Impacts
,”
SAE
Paper No. 2005-22-0008.10.4271/2005-22-0008
44.
Gabler
,
L. F.
,
Panzer
,
M. B.
, and
Salzar
,
R. S.
, “
High-Rate Mechanical Properties of Human Heel Pad for Simulation of a Blast Loading Condition
,”
Proceedings of the IRCOBI Conference
, Berlin, Germany, Sept. 12, pp.
796
808
.https://www.researchgate.net/publication/288095691_Highrate_mechanical_properties_of_human_heel_pad_for_simulation_of_a_blast_loading_condition
45.
Weaver
,
C.
,
Guleyupoglu
,
B.
,
Miller
,
A.
,
Kleinberger
,
M.
, and
Stitzel
,
J. D.
,
2021
, “
Pelvic Response of a Total Human Body Finite Element Model During Simulated Injurious Under Body Blast Impacts
,”
ASCE-ASME J. Risk Uncert. Eng. Sys., B
,
7
(
2
), p.
021004
.10.1115/1.4049105
46.
Hostetler
,
Z. S.
,
Aira
,
J.
,
Stitzel
,
J. D.
, and
Gayzik
,
F. S.
,
2019
, “
A Computational Study of the Biomechanical Response of the Human Lower Extremity Subjected to High Rate Vertical Accelerative Loading
,”
Proceedings of IRCOBI Conference
,
Florence
, Italy, Sept., pp.
662
673
.
47.
Hostetler
,
Z. S.
,
2022
, “
Lower Extremity Validation of a Human Body Model for High Rate Axial Loading in the Underbody Blast Environment
,” Stapp Car Crash J.,
66
, No.
2022-22-0004
.https://www.sae.org/publications/technicalpapers/content/2022-22-0004/
48.
DeVogel
,
N.
,
Yoganandan
,
N.
,
Banerjee
,
A.
, and
Pintar
,
F.
,
2020
, “
Hierarchical Process Using Brier Score Metrics for Lower Leg Injury Risk Curves in Vertical Impact
,”
BMJ Mil. Health
,
166
(
5
), pp.
318
323
.10.1136/jramc-2018-001124
49.
Hostetler
,
Z. S.
,
Hsu
,
F.-C.
,
Yoganandan
,
N.
,
Pintar
,
F. A.
,
Banerjee
,
A.
,
Voo
,
L.
, and
Gayzik
,
F. S.
,
2021
, “
An Improved Method for Developing Injury Risk Curves Using the Brier Metric Score
,”
Ann. Biomed. Eng.
,
49
(
11
), pp.
3091
3098
.10.1007/s10439-020-02686-8
50.
Petitjean
,
A.
,
Trosseille
,
X.
,
Yoganandan
,
N.
, and
Pintar
,
F. A.
,
2015
, “
Normalization and Scaling for Human Response Corridors and Development of Injury Risk Curves
,”
Accidental Injury
,
Springer
, New York, pp.
769
792
.
51.
Hostetler
,
Z. S.
,
Caffrey
,
J.
,
Aira
,
J.
, and
Gayzik
,
F. S.
,
2023
, “
Lower Extremity Validation of a Human Body Model for High Rate Axial Loading in the Underbody Blast Environment
,”
SAE
Paper No. 2022-22-0004.10.4271/2022-22-0004
52.
Baker
,
W. A.
,
Chowdhury
,
M. R.
, and
Untaroiu
,
C. D.
,
2018
, “
Validation of a Booted Finite Element Model of the WIAMan ATD Lower Limb in Component and Whole-Body Vertical Loading Impacts With an Assessment of the Boot Influence Model on Response
,”
Traffic Injury Prev.
,
19
(
5
), pp.
549
554
.10.1080/15389588.2018.1433829
53.
Hostetler
,
Z. S.
,
Aira
,
J.
,
Stitzel
,
J. D.
, and
Gayzik
,
F. S.
, “
A Computational Study of the Biomechanical Response of the Human Lower Extremity Subjected to High Rate Vertical Accelerative Loading
,”
Proceedings Conference Proceedings of International Research Council on Biomechanics of Injury
, Florence, Italy, Sept. 11, pp.
662
673
.
54.
Yoganandan
,
N.
,
Chirvi
,
S.
,
Voo
,
L.
,
DeVogel
,
N.
,
Pintar
,
F. A.
, and
Banerjee
,
A.
,
2017
, “
Foot-Ankle Complex Injury Risk Curves Using Calcaneus Bone Mineral Density Data
,”
J. Mech. Behav. Biomed. Mater.
,
72
, pp.
246
251
.10.1016/j.jmbbm.2017.05.010
55.
Yoganandan
,
N.
,
Chirvi
,
S.
,
Pintar
,
F. A.
,
Uppal
,
H.
,
Schlick
,
M.
,
Banerjee
,
A.
,
Voo
,
L.
, et al.,
2016
, “
Foot–Ankle Fractures and Injury Probability Curves From Post-Mortem Human Surrogate Tests
,”
Ann. Biomed. Eng.
,
44
(
10
), pp.
2937
2947
.10.1007/s10439-016-1598-2
56.
Frazer
,
L.
, and
Nicolella
,
D. P.
,
2020
,
Human Response to High Rate Loading
,
Southwest Research Institute
,
San Antonio, TX
.
57.
Schoell
,
S. L.
,
Weaver
,
A. A.
,
Urban
,
J. E.
,
Jones
,
D. A.
,
Stitzel
,
J. D.
,
Hwang
,
E.
,
Reed
,
M. P.
, and
Rupp
,
J. D.
,
2015
, “
Development and Validation of an Older Occupant Finite Element Model of a Mid-Sized Male for Investigation of Age-Related Injury Risk
,”
Stapp Car Crash J.
,
59
, p.
359
.10.4271/2015-22-0014
58.
Umale
,
S.
,
Khandelwal
,
P.
,
Humm
,
J.
, and
Yoganandan
,
N.
,
2020
, “
Development and Validation of an Elderly Human Body Model for Frontal Impacts
,”
Traffic Injury Prev.
,
21
(
sup1
), pp.
S147
S149
.10.1080/15389588.2020.1829922

Supplementary data