Foot shape and plantar pressure relationships in shod and barefoot populations

November 2019 Biomechanics and Modeling in Mechanobiology

DOI: 10.1007/s10237-019-01255-w

この著者の他の研究もチェックした方が良さそう。

母趾角と1−２趾間距離が歩行時の足圧と関係あり

abst

This study presents population-based multivariate regression models for predicting foot plantar pressure from easily measured foot metrics in both shod and barefoot populations for running and walking tasks. Both shod and barefoot models were trained on 50 participants and predicted plantar pressure from anthropometric measurements using a ‘leave-one-out’ validation with R2 values of 0.72–0.78 across walking and running in both populations. When the model was blindly tested on 16 new data sets, the model performed just as well with R2 values of 0.76–0.79 across both populations. Walking and running peak plantar pressure were predicted with similar levels of accuracy in both populations. It was revealed that forefoot plantar pressure was more sensitive to the hallux-toe distance in barefoot people with shod participants showing little response to this foot characteristic. Lateral forefoot plantar pressure was sensitive to the arch index in both shod and barefoot participants but only for walking. During running, the arch index was not a useful determinant of lateral forefoot pressure. Hence, habitually barefoot people who adopt minimalist footwear should consider additional support in the medial forefoot and walking footwear should include forefoot support stratified by arch index (foot type), but running footwear is challenging due to the variability in strike patterns.

この研究では、ランニングタスクとウォーキングタスクの靴と裸足の両方の人口で簡単に測定された足の測定基準から足底圧を予測するための人口ベースの多変量回帰モデルを提示します。靴と裸足モデルの両方が50人の参加者で訓練され、両集団での歩行と走行でのR2値が0.72〜0.78の「leave-one-out」検証を使用して、人体測定から足底圧を予測しました。モデルが16の新しいデータセットで盲目的にテストされたとき、モデルは両方の母集団にわたって0.76〜0.79のR2値で同様に実行されました。ウォーキングとランニングのピーク足底圧は、両方の集団で同様のレベルの精度で予測されました。前足部の足底圧は、裸足の人の母趾-4趾間の距離により敏感であり、靴の参加者はこの足の特徴に対してほとんど反応を示さないことが明らかになりました。前足部外側の足底圧は、歩行時のみ、靴および裸足群の両方のアーチ指数に敏感でした。ランニング中、アーチインデックスは、前足部外側の足圧の有用な決定因子ではありませんでした。したがって、ミニマリストのフットウェアを採用している裸足の習慣的な人々は、内側の前足で追加のサポートを検討する必要があり、歩行用のフットウェアにはアーチインデックス（足型）によって層化された前足のサポートを含める必要がありますが、ランニングシューズはストライクパターンのばらつきのために困難です。

Keywords

Shod · Barefoot · Hallux angle · 1–2 Toe distance · Peak pressure · Walking · Running

intro

人間の足の形態は、二足歩行と環境との相互作用から進化しました（Bennett et al。2009）。具体的には、足の形態は、足根骨から後方に突出した骨、ドーム状の内側および外側縦アーチ、および残りのつま先に対する内転母趾（足の親指）によって特徴付けられます（Bennett et al。2009; Shu et al。2015）。この足の形態は、歩行中の体重伝達やバランス制御など、いくつかの機能を果たします。さらに、前足と後足の間で負荷が共有され、弾性エネルギーが縦アーチ（中足部）に保存され、つま先のグリップ機能によって安定性が維持されます（Bramble and Lieberman 2004; Mei et al。2016）。足の機能的パフォーマンスは、足のタイプ（Hillstrom et al。2013; Mootanah et al。2013）、年齢層（Mülleret al。2012）、民族性（Gurney et al。2009）によって影響を受けることが報告されています。裸足または裸足の習慣（Mei et al。2015、2016; Shu et al。2015; Hollander et al。2017a）、BMIおよび性別（Domjanic et al。2015）。

人間の足は、ウォーキングやランニングなどの特定のタスクを実行するように進化しています。たとえば、足は、小さなアキレス腱モーメントアームを備えた短い骨の塊茎の長さ（Raichlen et al。2011）と、より少ない機械的作業を必要とする足趾屈筋力の低減を備えた短いつま先の長さでの持久走行用に設計されていることが報告されています（Rolian et al al。2009）。現代の狭い履物の導入により、母趾の内転が増加し、外反母趾が形成された（Shu et al。2015）。具体的には、母趾外転が増加した習慣的な裸足集団は前足部の負荷分散を強化しましたが、これは中足骨に集中した圧力を示す小足集団では観察されませんでした（Mei et al。2015）。これは最近の研究で確認されており、つま先は狭い履物をシミュレートし、つま先の負荷を制限し、中足骨の負荷を増加させました（Mei et al。2016）。

足の種類と年齢層が異なると、足底の圧力パターンに影響します。多くの場合、足は、動的な足機能のための足分類方法（ラゼギおよびラテギとその他）を使用して、扁平（低アーチ型および回内）、ノーマル（中立）および凹足（高アーチ型および回外）（Hillstrom et al.2013; Mootanah et al.2013） Batt 2002）。興味深いことに、歩行運動学と時空間パラメータは3つの足タイプ間で違いはありませんが、扁平足は内側に集中した圧力を示し、凹足はノーマルと比較して増加した横圧を示します（Hillstrom et al。2013; Mootanah et al。2013）。年齢も足の圧力パターンに関与しており、6歳の子供は成人の足底圧力パターンに向かって増加傾向を示しています（Mülleret al。2012）。

異なる民族、靴を履く習慣、性別、BMIの人口は、異なる足の形態と関連する足底圧パターンを示します。たとえば、西部の白人とインドの裸足の個人の比較は、異なる足の形と足底の圧力を明らかにしました。細い白人の足はかかとと前足に集中した圧力を示しましたが、より広い裸足のインドの足は均等に分散した圧力を示しました（D’Aoûtet al。2009）。これらの発見は、中国の小足とインドの裸足ランナーの生体力学的違いを調査した最近の研究と一致していました（Mei et al。2015）。民族性の違いを正規化して、インドの靴下とインドの裸足の参加者を比較すると、靴を履くと足底圧の集中が増加することが明らかになりました（D’Aoûtet al。2009）。足の長さ、幅、アーチ指数、母趾角も靴を履く習慣の影響を受け、子供や青年の足の発達に影響を与える要因になる可能性があります（Hollander et al。2017a）。性別およびBMI因子も足の形で報告されており、小柄な女性は細い足を示し（Shu et al。2015）、高BMIの人はより広い足を示し（Domjanic et al。2015）、高齢男性のアーチ指数の増加（Aurichio et al。2011）。

足底圧を含む歩行パラメータの予測は、時間のかかる実験を必要とせずに足の力学を迅速に推定する機会を提供します。これは、足底圧変動の最大47％を予測するために、足構造と人体計測の組み合わせを使用して以前に試みられました（Mootanah et al。2013）。別の研究では、BMI、身長、体重を含む単純な測定から、足の長さ、幅、アーチ指数、足姿勢指数を含む足の測定基準を予測しました（Aurichio et al。2011; Domjanic et al。2015）。牽引力を獲得する一般的な手法は、変動を説明するパラメーターの最小セットを決定するために使用される、変数削減法である主成分分析（PCA）です。たとえば、最近の50フィートのPCA分析では、最初の4つの主成分が足の形状の変化の78.3％を占めていることが示されました（Fernandez et al。2019）。さらに重要な順に最初の4つのPCAモードは足のボリューム、回内/回外、足の幅、足のアーチであることが明らかになりました。これらのPCAモードは、母集団全体の歩行パラメータの予測子として使用される可能性があります。

この研究の目的は、足の足底圧を迅速に予測し、特定の足の形態が足底圧に関係することを説明するために、大量の靴と裸足の足データを利用することです。これは、履物の設計、足の生体力学、および足の病理を評価する際の考慮事項に影響を与えます。この研究では、歩行とランニングのタスクでの足底圧のピークを迅速に予測するために、足と足の両方の足の形状と足底圧の関係を訓練する多変量回帰モデルを提示します。足底圧の迅速な予測のためのパラメータの最適なサブセットを選択するために、足底圧が最も敏感である形状メトリックを評価します。（1）足の形状からのピーク足底圧の予測は、歩行ではより正確になりますが、ランニングストライクパターンの変動によりランニングを予測するときの精度が低下すると仮定します。（2）足の親指からつま先までの距離は、裸足の人の前足の圧力を説明するかもしれませんが、靴の個人ではあまり役に立ちません。（3）アーチインデックスは、すべての集団で前足の外側圧を説明する可能性があります。

Methods

2.1 Participants

A total of 136 age-matched physically active males participated in this study, with 68 habitually barefoot (BF) males (chosen from a south Indian population) and 68 habitually shod (SH) males (chosen from a Chinese population). Participants of Indian ethnicity have exhibited barefoot gait since birth and only wore non-toe-restricted shoes, while participants of Chinese ethnicity wore different kinds of shoes throughout life. Both groups had experience with running and sports for at least 3 h per week (defined as physically active). Participants were informed of the requirements and procedures of this study, which was approved by the Ethics Committee in Ningbo University (RAGH20170306).

2.2 Test protocol and data collection

https://www.researchgate.net/profile/Qichang_Mei/publication/337101874/figure/fig1/AS:823520376393728@1573353966163/Protocol-of-foot-scanning-and-plantar-pressure-collection_W640.jpg

Participants visited the motion analysis laboratory for foot shape scanning and plantar pressure measurement using a randomised order following familiarisation with the laboratory environment (Fig. 1). A 3D foot scanner (Easy- Foot-Scan, EFS, OrthoBaltic, Kaunas, Lithuania) was used to collect foot morphology data following our previously established protocol (Shu et al. 2015), and an EMED plate (Novel, Germany) was employed to record the plantar pressure at 100 Hz, which was fixed in the middle of the gait runway. To capture participants’ natural walking and running gait patterns, a self-selected speed was performed during both walking and running over four successful trials for each task.

Participant measurements collected included body mass (kg), height (m), body mass index (BMI, kg/m2), foot length (mm), foot width (mm), width/length ratio, heel width (mm), first–second toe distance (mm), hallux angle (degree) and arch index. Hallux angle was formulated as positive (adduction or varus) and negative (abduction or valgus) based on the local foot axis. Arch index was calculated from the ratio of midfoot area divided by the whole foot area excluding toes (Cavanagh and Rodgers 1987). Peak plantar pressure was selected from four trials of each participant averaged to minimise inter-trial error. Differences in anthropometrics, foot metrics and peak pressures were tested using independent samples t test, with significance level of 0.05.

2.3 Partial least squares regression (PLSR) prediction model 多変量回帰分析モデルの式

Partial least squares regression (PLSR) (Wold et al. 1984) was used to model the relationships between 10 predictors, including foot metrics and anthropometrics, and 11 responses, including peak pressure in 11 plantar regions. Foot metrics included length, width, width/length ratio, heel width, hallux- toe distance, hallux angle and arch index. Participant measurements included height and mass to compute body mass index (BMI). The 11 plantar regions were hallux (HA), other toes (OT), first to fifth metatarsals (M1, M2, M3, M4 and M5), medial midfoot (MM), lateral midfoot (MF), medial heel (MH) and lateral heel (LH), which are depicted in the study modelling pipeline (Fig. 1).

The two fundamental equations in PLSR are the predictor matrix (X) and the response matrix (Y) given by

XNM =TNLPTML+ENM, (1) and

YNP = UNL QTPL + FNP, (2)

where subscript N is the number of data sets (50 training samples from each group in this study), subscript M is the number of predictor variables (10 metrics), subscript P is the number of response variables (11 plantar regions), and subscript L is the number of components. T and U are the projection matrices (also known as scores), and P and Q are the transposed orthogonal loading matrices (where the rows are created from eigenvectors or principal components), and E and F are the error or residual terms. The score vectors are related using a linear function,

U = f (T) + H

where H is the vector of residuals.

The PLSR model (Matlab statistics and machine learning toolbox) was trained with a data set of 50 shod and 50 barefoot participants, respectively. The model was firstly evaluated using a ‘leave-one-out’ cross-validation analysis and then tested on a data set of 18 new participants (from each group) that were not used in the training of the PLSR model. This was repeated for both walking and running tasks. To assist visualisation of predicted peak pressures in the eleven anatomical regions during walking and run- ning tasks, customised colour plots were created from the scanned foot print for SH and BF participants, respec- tively. Comparison of EMED measured and PLSR model predicted errors in peak plantar pressure is presented to show the accuracy of the trained model. Secondly, a sen- sitivity analysis was conducted to evaluate the influence of foot shape metrics on peak plantar pressures by perturba- tion of foot metrics in the PLSR model to observe vari- ations (changes in percentage) in predicted peak plantar pressure of each foot region. Specifically, we evaluate the influence of hallux-toe distance and foot arch index on peak plantar pressure in SH and BF populations.

Results

https://gyazo.com/dec68e3f24595e4870b9bc362391d764

Comparison of metrics between SH and BF participants is presented in Table 1. There were differences observed in body mass, height, foot width, width/length ratio, heel width, 1–2 toe (hallux-toe) distance and hallux angle between SH and BF groups. These 10 metrics were used in the PLSR model to estimate 11 peak plantar pressures.

Peak plantar pressures during walking for SH and BF groups are shown in Table 2 for 11 regions. There were observed differences in the OT, M1, M2, M3, M4, MM, LM, MH and LH regions. Figure 2 exhibits the mean peak pressure for walking in each region for both SH and BF populations. High peak pressures were observed at the heel (MH, 506.4 kPa and LH, 452.5 kPa), medial forefoot (M1, 525.5 kPa and M2, 531.9 kPa) and hallux (521.7 kPa) regions in the SH group (left), but evenly distributed across the rear and forefoot regions in the BF group (right), includ- ing MH (385.1 kPa), LH (372.1 kPa), M1 (370.9 kPa), M2 (373.5 kPa) and M3 (352.1 kPa).

Peak plantar pressures during running for SH and BF groups are shown in Table 3 for 11 regions. There were dif- ferences observed in M1–M5, MM, LM, MH and LH. Fig- ure 3 shows the mean peak pressure distribution during run- ning. Similar to walking, high peak pressures were observed in the MH (822.1 kPa), LH (752.7 kPa), M1 (790.8 kPa), M2 (688.9 kPa), M3 (582.1 kPa) and hallux (651.6 kPa) regions in the SH population. However, low pressure was exhibited in the rear foot, MH (384.5 kPa) and LH (371.4 kPa), with higher forefoot pressure, M1 (536.9 kPa), M2 (461.1 kPa) and M3 (436.5 kPa) and hallux (716.3 kPa) for the BF population.

https://www.researchgate.net/profile/Qichang_Mei/publication/337101874/figure/fig2/AS:823520376418304@1573353966237/Training-black-and-testing-blue-accuracy-of-walking-top-row-and-running-bottom_W640.jpg

https://www.researchgate.net/profile/Qichang_Mei/publication/337101874/figure/fig3/AS:823520376410112@1573353966295/Training-black-and-testing-blue-accuracy-of-walking-top-row-and-running-bottom_W640.jpg

PLSR models for BF (Fig. 4) and SH (Fig. 5) populations were trained separately for walking and running tasks. A ‘leave-one-out’ analysis showed plantar pressure prediction accuracy of 72.5% for walking and 74.9% for running in the BF population. Testing the model on 18 new BF cases showed a similar accuracy with 76.3% for walking and 78.7% for running. In the SH population a ‘leave-one-out’ analysis exhibited prediction accuracy of 78.2% for walking and 76.4% for running. Testing the SH model on 18 new cases showed a similar accuracy of 78.9% for walking and 79% for running.

https://www.researchgate.net/profile/Qichang_Mei/publication/337101874/figure/fig4/AS:823520376389637@1573353966360/Peak-plantar-pressure-ground-truth-left-versus-prediction-error-right-for-walking_W640.jpg

To evaluate the PLSR model spatial prediction, we plotted the ground-truth EMED measured pressures versus the PLSR predicted error for walking (Fig. 6) and running (Fig. 7) of the median participant. The average prediction errors for walking were 17.4% for SH and 13.9% for BF populations, respectively. The model was a good predictor with low errors in M3 (1.7%) and M4 (4.9%) for SH walking and M3 (1%), M2 (5.4%) and M1 (7.1%) for BF walking. However, the model was a poor predictor of M5 with errors of 46.4% for SH and 55.9% for BF. For running, average prediction errors were similar to walking with 14.5% for SH and 13.3% for BF populations, respectively. The regions that were best predicted included HA (2.3%), M2 (5.1%) and M3 (6.4%) for SH running, and HA (2%) and M1 (0.4%) for BF running. In contrast to walking, M4 (25.6%) and the heel

region (MH: 16.2% and LH: 11.3%) were predicted poorly for SH running. Poor predictions were also observed in BF running for the M4 (20.7%), MH (19.8%) and LH (22.3%).

Body mass and height influenced plantar pressure magnitude significantly due to ground reaction force being associated with mass and height. However, the changes were uniform across the foot and not sensitive to foot morphology. Sensitivity analysis of foot morphology revealed that plantar pressure was less sensitive to total foot length, forefoot width, width/length ratio and heel width, and more sensitive to 1–2 toe distance and arch index. Figures 8 and 9 exhibit the response of foot pressure during walking and running to a 5% perturbation of 1–2 toe distance and arch index. BF walking (Fig. 8) was most sensitive to a perturbation of 1–2 toe distance in the medial forefoot pressure (M1 and M2), whereas SH walking revealed little response across the foot. However, both SH and BF walking were sensitive in forefoot pressure (M3 and M5) to perturbation of arch index. BF walking was more dominant on the lateral forefoot (increased M5), whereas SH walking presented more uniform forefoot pressure response. Considering running (Fig. 9), both BF and SH populations exhibited little response across the foot to changes in 1–2 toe distance; however, BF running revealed a slight increase in the medial forefoot. In contrast, BF running revealed large fore- and rearfoot pressure response to arch index perturbation, whereas SH running only exhibited medial midfoot response.

Discussion

This study presents population-based multivariate regression models for predicting foot plantar pressure from easily measured foot metrics in both SH and BF populations for running and walking tasks. Both SH and BF models were trained on 50 participants and predicted plantar pressure from anthropometric measurements using a ‘leave-one-out’ validation with R2 values of 0.72–0.78 across walking and running in both populations. When the model was blindly tested on 16 new data sets, the model performed just as well with R2 values of 0.76–0.79 across both populations. Walking and running peak plantar pressure were predicted with similar levels of accuracy in both populations. It was revealed that forefoot plantar pressure was more sensitive to the hallux 1–2 toe distance in BF people with SH participants showing little response to this foot characteristic. Lateral forefoot plantar pressure was sensitive to the arch index in both SH and BF participants but only for walking. During running, the arch index was not a useful determinant of lateral fore- foot pressure.

Several limitations should be considered when interpret- ing the findings from this study. Firstly, we did not account for subject-specific strike patterns for the barefoot population in our model, but this is primarily important for running. We did note that less than 20% of the barefoot population was a forefoot striker during the running test, which is consistent with previously reported percentage of forefoot strikes with self-selected speed (Hatala et al. 2013). Secondly, participants walked and ran with self-selected speed, and this will lead to variation in plantar pressures. This may also explain why our prediction was ~ 78% at best as the remaining variation in plantar pressure is likely explained by participant speed. However, the design of this model was to be used in practice where walking and running speed is not controlled for. Further, our participants exhibited walk- ing speeds (~ 1.3 m/s) (Sun et al. 2018) and running speeds (~ 3.0 m/s) (Hatala et al. 2013) consistent with previous stud- ies. Lastly, only male participants were recruited for this study, to control foot metric variations due to gender differ- ence, as we previously found that deviated hallux angle was highly pronounced in females (Shu et al. 2015). A separate female population-based model is recommended using the same pipeline as presented in this study.

Previous studies investigating habitually SH and BF foot morphology found similar foot characteristics to our study, specifically, wider feet, particular in the toe region, could

be observed in BF populations, but slender feet are typical in SH populations (Hoffmann et al. 1905; Thompson and Zipfel 2005; Kadambande et al. 2006; Zipfel and Berger 2007; D’Août et al. 2009; Shu et al. 2015; Hollander et al. 2017a). This study selected SH participants from a Chinese population and BF participants from an Indian population. Ethnic differences in body mass and height are well known (Deurenberg et al. 1998), but mass was normalised using height to give BMI, which showed no difference between the SH and BF groups in our study, consistent with previ- ous studies (Thompson and Zipfel 2005; Butterworth et al. 2015).

Foot length and width of the BF and SH populations in this study are consistent with those previously reported (Shu et al. 2015; Hollander et al. 2017b) (D’Août et al. 2009; Ashizawa et al. 1997) for Indian, African, Caucasian and Asian populations; hence, this provides confidence in the width/length ratios we report. Furthermore, the BF heel width in this study (which is wider than SH groups) is consistent with previous work (Weinans et al. 2018; Moore et al. 2019) and may be associated with functional adaptation (Gu et al. 2015; Reznikov et al. 2017; Mei et al. 2018). However, this study focussed on the articulate features of the foot that is associated with BF and SH foot morphology, namely, the arch and hallux region (1–2 toe distance).

The arch index, initially developed to classify foot types (Cavanagh and Rodgers 1987), was not different between the SH and BF groups in this study. It should be noted that we removed atypical and pathologic feet during patient recruit- ment. In contrast, the static arch index was reported to be different between habitually SH and BF children and adolescents (Hollander et al. 2017a). Our study showed no difference, which might be attributed to all participants being adults, physically active with less plastic foot features com- pared to children and adolescents.

Differences in hallux angle and hallux-toe distance between SH and BF populations in the present study are consistent with those reported previously (Hoffmann et al. 1905; Shu et al. 2015; Hollander et al. 2017a). Observing the measured plantar pressures in this study, we found SH feet exhibited concentrated plantar pressure in the heel and medial forefoot, whereas BF feet presented a more even dis- tribution, except for the hallux region. These findings are consistent with previous studies for walking (D’Août et al. 2009) and running (Mei et al. 2015).

部分最小二乗回帰PLSRは、予測子と応答を線形相関させるためのsupervised 多変量回帰統計手法であり、これまでは、約90％の精度（Fernandez et al.2018）でmuscle mechanicsを予測するために利用されており、大腿骨の形状変動は約2.3 mm RMSエラー（ Zhang et al。2016）およびモーションキャプチャ（Johnson et al。2018）から最大98％の精度で外部荷重（地面反力と関節モーメント）。私たちの知識から、これはPLおよびSRの母集団における足の測定基準を歩行およびランニングの足底圧と相関させるPLSRモデルを提示する最初の研究です。精度が約78％と低いのは、2つの理由が考えられます。最初に、足底圧の主要な特徴である参加者の速度とストライクパターンが含まれていませんでした。 2番目に、足の力学の機能は非線形である可能性が高く、PLSRは線形予測子です。私たちのモデルは空間的な足底の圧力パターンをうまく予測しましたが、圧力の大きさについてはあまり正確ではありませんでした。

私たちのモデルでは、内側と中央の前足領域（M1からM4）が高精度で予測されることが観察されました。これは、健康な足のこれらの足底圧領域を測定する高い信頼性と一致しています（De Cock et al。 2006; McKay et al。2017）。さらに、M5のモデル予測精度が低いのは、この領域で測定されたピーク圧力のクラス内相関が低いという報告と一致しています（De Cock et al。2006）。さらに、ランニング中の離陸前の中央体重移動（Bennett et al。2009）は、姿勢の異なる人々の間で非常に変動します（Buldt et al。2018）。

The 1–2 toe distance (associated with the hallux angle) is a commonly used classifier to differentiate BF and SH feet (Mei et al. 2015; Shu et al. 2015; Hollander et al. 2017a). BF morphology presents large variation in the hallux region including the 1–2 toe distance, which affects the medial forefoot articulation with the ground, especially during walking. Hence, our finding that forefoot plantar pressure was sensi- tive to the 1–2 toe distance for BF walking is consistent with known BF morphology (Hoffmann et al. 1905; Branthwaite et al. 2013; Mei et al. 2016). However, SH toe spacing is typ- ically closed with minimal variation; hence, plantar pressure is independent of toe morphology in SH walking (Hoffmann et al. 1905; Branthwaite et al. 2013; Mei et al. 2016). During BF running, we observed that only the medial forefoot plantar pressure was sensitive to the 1–2 toe distance (SH walking showed no sensitivity). This is consistent with the hallux gripping function reported to be more pronounced during running. Hence, habitually BF people who adopt minimalist footwear should consider additional support in the medial forefoot.（BFの実行中、前足の内側の足底圧のみが1〜2の足指の距離に敏感であることがわかりました（SH歩行は感度を示しませんでした）。これは、走行中により顕著になると報告されている母趾把持機能と一致しています。したがって、ミニマリストの靴を採用する習慣的にBFの人々は、内側の前足での追加のサポートを検討する必要があります。

Arch indexは当初、異なる足型を分類するための静的な指標として開発されましたが（Cavanagh and Rodgers 1987）、動的運動中のアーチ指数の変動を明らかにしようとした研究はほとんどありません（Hollander et al.2017a）。前足圧は、BFとSHの両歩行においてアーチ指数に敏感であり、BF参加者では横方向のプリサイスが増加することが観察された。これは、靴によってより拘束されるSH群と比較して、体重移動中に前足のより多様な関節を示すBF群と一致します（Hoffmannら1905；Lambri-nudi 1932）。一方、SH走では、中足部足底圧のみがアーチ指数に感応するのに対し、BF走では、前足部と後足部の足底圧がアーチ指数に大きく感応することが観察されました。これは、BFの集団は前足部と後足部のストライカーが混在しており（Mei et al.2015）、そのため、同様のアーチ指数に対する圧力に大きなばらつきがあることが説明できます。一方、私たちのSH集団は一貫した打撃パターンを持ち、SHの足はアーチサポート付きフットウェアに適応したアーチプロファイルを持つため、アーチ領域の圧力のみが影響を受けたのです。したがって、ウォーキングフットウェアは、アーチ指数（足型）によって層別された前足部サポートを含むべきであるが、ランニングシューズは、接地パターンのばらつきのために困難である。

Funding This study is supported by the National Natural Science Foundation of China (No. 81772423), NSFC (Natural Science Foun- dation of China)—RSE (The Royal Society of Edinburgh) Joint Project (No. 81911530253), National Key R&D Program of China (2018YFF0300903) and K. C. Wong Magna Fund in Ningbo Univer- sity. Qichang Mei is supported by the New Zealand–China Doctoral Research Scholarship issued from the Ministry of Foreign Affairs and Trade (New Zealand).