Identification of subject-specific responses to footwear during running
Keywords:Support Vector Machines, Layer-wise Relevance Propagation, Ground Reaction Forces, Explainable Artificial Intelligence, XAI, Machine Learning
Placing a stronger focus on subject-specific responses to footwear may lead to a
better functional understanding of footwear effects on running and its influence on comfort
perception, performance, and pathogenesis of injuries. Here, we investigate subject-specific
responses to different footwear conditions within ground reaction force (GRF) data during
running using a machine learning-based approach. We conduct our investigation in three steps,
guided by the following hypotheses: (I) For each subject x footwear combination, unique GRF
patterns can be identified. (II) For each subject, unique GRF characteristics can be identified
across footwear conditions. (III) For each footwear condition, unique GRF characteristics can be
identified across subjects.
Thirty male subjects ran ten times at their preferred (self-selected) speed on a level and
approximately 15 m long runway in five footwear conditions (barefoot, subject’s own running
shoe, and three standardised running shoes). We recorded three-dimensional GRFs for one
right-foot stance phase per running trial and classified the vectorised GRFs using support vector
The highest prediction accuracy was found for the subject x footwear classification
(hypothesis I). The median prediction accuracy was 95.7 %. This is approximately 137 times higher
than the zero-rule baseline (ZRB) of 0.7 %. Across footwear conditions, subjects could be
discriminated with a median prediction accuracy of 89.7 % (approximately 27 times higher than
the ZRB of 3.3 %). Across subjects, footwear conditions could be discriminated with a median
prediction accuracy of 76.3 % (approximately 4 times higher than the ZRB of 20.0 %).
Our results suggest that, during running, responses to footwear are unique to each
individual subject and footwear design. As a result, considering subject-specific responses
contribute to a more differentiated functional understanding of footwear effects. Incorporating
holistic biomechanical data is auspicious for the (subject-specific) evaluation of the footwear
effects, as unique interactions between subjects and footwear manifest in versatile ways.
Machine learning methods have demonstrated their great potential to fathom subject-specific
responses when evaluating and recommending footwear.
Nigg BM, Stefanyshyn D, Cole G, Boyer K. Footwear research - past, present and future. In: Hamill J, Hardin E, Williams K, eds. Proceedings: 7th Symposium on Footwear Biomechanics. Cleveland, Ohio, USA: Case Western Reserve University Printing; 2005:14-17.
Nigg BM. Biomechanics of Sport Shoes. Calgary, Alberta, Canada: Topline Printing; 2010.
Nigg BM, Baltich J, Hoerzer S, Enders H. Running shoes and running injuries: mythbusting and a proposal for two new paradigms: 'preferred movement path' and 'comfort filter'. Br J Sports Med. 2015;49(20):1290-1294. https://doi.org/10.1136/bjsports-2015-095054
Hoitz F, Mohr M, Asmussen M, Lam W‑K, Nigg S, Nigg B. The effects of systematically altered footwear features on biomechanics, injury, performance, and preference in runners of different skill level: a systematic review. Footwear Sci. 2020;12(3):193-215. https://doi.org/10.1080/19424280.2020.1773936
Sterzing T, Schweiger V, Ding R, Cheung JT‑M, Brauner. Influence of rearfoot and forefoot midsole hardness on biomechanical and perception variables during heel-toe running. Footwear Sci. 2013;5(2):71-79. https://doi.org/10.1080/19424280.2012.757810
Sterzing T, Custoza G, Ding R, Cheung JT‑M. Segmented midsole hardness in the midfoot to forefoot region of running shoes alters subjective perception and biomechanics during heel-toe running revealing potential to enhance footwear. Footwear Sci. 2015;7(2):63-79. https://doi.org/10.1080/19424280.2015.1008589
Cole GK, Nigg BM, Fick GH, Morlock MM. Internal loading of the foot and ankle during impact in running. J Appl Biomech. 1995;11(1):25-46. https://doi.org/10.1123/jab.11.1.25
Oriwol D, Sterzing T, Milani TL. The position of medial dual density midsole elements in running shoes does not influence biomechanical variables. Footwear Sci. 2011;3(2):107-116. https://doi.org/10.1080/19424280.2011.613857
Baltich J, Maurer C, Nigg BM. Increased vertical impact forces and altered running mechanics with softer midsole shoes. PloS One. 2015;10(4):e0125196. https://doi.org/10.1371/journal.pone.0125196
Schöllhorn WI, Nigg BM, Stefanyshyn DJ, Liu W. Identification of individual walking patterns using time discrete and time continuous data sets. Gait Posture. 2002;15(2):180-6. https://doi.org/10.1016/s0966-6362(01)00193-x
Uhl A, Wild P. Personal identification using eigenfeet, ballprint and foot geometry biometrics. In: 2007 First IEEE International Conference on Biometrics: Theory, Applications, and Systems. 2007:1-6. https://doi.org/10.1109/BTAS.2007.4401924
Kanai R, Rees G. The structural basis of inter-individual differences in human behaviour and cognition. Nat Rev Neurosci. 2011;12(4):231-242. https://doi.org/10.1038/nrn3000
van Mechelen W. Running injuries. A review of the epidemiological literature. Sports Med. 1992;14(5):320-335. https://doi.org/10.2165/00007256-199214050-00004
Bates BT, Osternig LR, Sawhill JA, James SL. An assessment of subject variability, subject-shoe interaction, and the evaluation of running shoes using ground reaction force data. J Biomech. 1983;16(3):181-91. https://doi.org/10.1016/0021-9290(83)90125-2
Pataky TC, Mu T, Bosch K, Rosenbaum D, Goulermas JY. Gait recognition: Highly unique dynamic plantar pressure patterns among 104 individuals. J R Soc Interface. 2012;9(69):790-800. https://doi.org/10.1098/rsif.2011.0430
Horst F, Mildner M, Schöllhorn WI. One-year persistence of individual gait patterns identified in a follow-up study - a call for individualised diagnose and therapy. Gait Posture. 2017;58:476-480. https://doi.org/10.1016/j.gaitpost.2017.09.003
Hoitz F, Tscharner V von, Baltich J, Nigg BM. Individuality decoded by running patterns: movement characteristics that determine the uniqueness of human running. PloS One. 2021;16(4): e0249657. https://doi.org/10.1371/journal.pone.0249657
Hoitz F, Fraeulin L, Tscharner V von, Ohlendorf D, Nigg BM, Maurer-Grubinger C. Isolating the unique and generic movement characteristics of highly trained runners. Sensors. 2021;21(21):7145. https://doi.org/10.3390/s21217145
Nigg BM, Vienneau J, Smith AC, Trudeau MB, Mohr M, Nigg SR. The preferred movement path paradigm: influence of running shoes on joint movement. Med Sci Sports Exerc. 2017;49(8):1641-1648. https://doi.org/10.1249/MSS.0000000000001260
Fisher AJ, Medaglia JD, Jeronimus BF. Lack of group-to-individual generalizability is a threat to human subjects research. Proc Natl Acad Sci USA. 2018;115(27):E6106-E6115. https://doi.org/10.1073/pnas.1711978115
Bates BT. Single-subject methodology: an alternative approach. Med Sci Sports Exerc. 1996;28(5):631-638. https://doi.org/10.1097/00005768-199605000-00016
Nigg BM. The role of impact forces and foot pronation: a new paradigm. Clin J Sport Med. 2001;11(1):2-9. https://doi.org/10.1097/00042752-200101000-00002
Hoerzer S, Tscharner V von, Jacob C, Nigg BM. Defining functional groups based on running kinematics using Self-Organizing Maps and Support Vector Machines. J Biomech. 2015;48(10):2072-2079. https://doi.org/10.1016/j.jbiomech.2015.03.017
Horst F, Lapuschkin S, Samek W, Müller K‑R, Schöllhorn WI. Explaining the unique nature of individual gait patterns with deep learning. Sci Rep. 2019;9(1):2391. https://doi.org/10.1038/s41598-019-38748-8
Aeles J, Horst F, Lapuschkin S, Lacourpaille L, Hug F. Revealing the unique features of each individual's muscle activation signatures. J R Soc Interface. 2021;18(174):20200770. https://doi.org/10.1098/rsif.2020.0770
Schöllhorn WI, Bauer HU. Recognition of individual running patterns using neural networks (Erkennung von individuellen Laufmustern mit Hilfe von neuronalen Netzen). In: Mester J, Perl J, eds. Informatik im Sport. Cologne, Germany: Sport Buch Strauss;1998:169-176. [in German].
Schöllhorn WI. Applications of artificial neural nets in clinical biomechanics. Clin Biomech. 2004;19(9):876-898. https://doi.org/10.1016/j.clinbiomech.2004.04.005
Prakash C, Kumar R, Mittal N. Recent developments in human gait research: parameters, approaches, applications, machine learning techniques, datasets and challenges. Artif Intell Rev. 2018;49(1):1-40. https://doi.org/10.1007/s10462-016-9514-6
Phinyomark A, Petri G, Ibáñez-Marcelo E, Osis ST, Ferber R. Analysis of big data in gait biomechanics: current trends and future directions. J Med Biol Eng. 2018;38(2):244-260. https://doi.org/10.1007/s40846-017-0297-2
Halilaj E, Rajagopal A, Fiterau M, Hicks JL, Hastie TJ, Delp SL. Machine learning in human movement biomechanics: best practices, common pitfalls, and new opportunities. J Biomech. 2018;81:1-11. https://doi.org/10.1016/j.jbiomech.2018.09.009
Cortes C, Vapnik V. Support-vector networks. Mach Learn. 1995;20(3):273-297. https://doi.org/10.1007/BF00994018
Fan RE, Chang K-W, Hsieh C-J, Wang X-R, Lin C-J. Liblinear: A library for large linear classification. J Mach Learn Res. 2008;9:1871-1874.
Bach S, Binder A, Montavon G, Klauschen F, Müller K‑R, Samek W. On pixel-wise explanations for non-linear classifier decisions by Layer-Wise Relevance Propagation. PloS One. 2015;10(7):e0130140. https://doi.org/10.1371/journal.pone.0130140
Bechtold B, Fletcher P, Holden S, Gorur-Shandilya S. Violin plots for Matlab, Available at: https://zenodo.org/record/4559847. [accessed 31.10.2022]. https://doi.org/10.5281/zenodo.4559847
Hug F, Vogel C, Tucker K, et al. Individuals have unique muscle activation signatures as revealed during gait and pedaling. J Appl Physiol. 2019;127(4):1165-1174. https://doi.org/10.1152/japplphysiol.01101.2018
van der Worp H, Vrielink JW, Bredeweg SW. Do runners who suffer injuries have higher vertical ground reaction forces than those who remain injury-free? A systematic review and meta-analysis. Br J Sports Med. 2016;50(8):450-457. https://doi.org/10.1136/bjsports-2015-094924
Johnson CD, Tenforde AS, Outerleys J, Reilly J, Davis IS. Impact-related ground reaction forces are more strongly associated with some running injuries than others. Am J Sports Med. 2020;48(12):3072-3080. https://doi.org/10.1177/0363546520950731
Schmida EA, Wille CM, Stiffler-Joachim MR, Kliethermes SA, Heiderscheit BC. Vertical loading rate is not associated with running injury, regardless of calculation method. Med Sci Sports Exerc. 2022;54(8):1382-1388. https://doi.org/10.1249/MSS.0000000000002917
Nigg BM, Mohr M, Nigg SR. Muscle tuning and preferred movement path – a paradigm shift. Curr Issues Sport Sci. 2017;2:007. https://doi.org/10.36950/CISS_2017.007
Trudeau MB, Willwacher S, Weir G, et al. A novel method for estimating an individual’s deviation from their habitual motion path when running. Footwear Sci. 2019;11(3):135-145. https://doi.org/10.1080/19424280.2019.1615004
Willwacher S, Mählich D, Trudeau MB, et al. The habitual motion path theory: Evidence from cartilage volume reductions in the knee joint after 75 minutes of running. Sci Rep. 2020;10(1):1363. https://doi.org/10.1038/s41598-020-58352-5
Ahn AN, Brayton C, Bhatia T, Martin P. Muscle activity and kinematics of forefoot and rearfoot strike runners. J Sport Health Sci. 2014;3(2):102-112. https://doi.org/10.1016/j.jshs.2014.03.007
Gruber AH, Boyer KA, Derrick TR, Hamill J. Impact shock frequency components and attenuation in rearfoot and forefoot running. J Sport Health Sci. 2014;3(2):113-121. https://doi.org/10.1016/j.jshs.2014.03.004
Slijepcevic D, Horst F, Lapuschkin S, et al. Explaining machine learning models for clinical gait analysis. ACM Trans Comput Healthc. 2022;3(2):1-27. https://doi.org/10.1145/3474121
Copyright (c) 2022 Fabian Horst, Fabian Hoitz, Nicolas Schons, Hendrik Beckmann, Benno M. Nigg, Wolfgang I. Schöllhorn
This work is licensed under a Creative Commons Attribution 4.0 International License.