{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,9]],"date-time":"2026-05-09T17:16:55Z","timestamp":1778347015053,"version":"3.51.4"},"reference-count":44,"publisher":"MDPI AG","issue":"4","license":[{"start":{"date-parts":[[2021,2,10]],"date-time":"2021-02-10T00:00:00Z","timestamp":1612915200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Sensors"],"abstract":"<jats:p>Human Locomotion Mode Recognition (LMR) has the potential to be used as a control mechanism for lower-limb active prostheses. Active prostheses can assist and restore a more natural gait for amputees, but as a medical device it must minimize user risks, such as falls and trips. As such, any control system must have high accuracy and robustness, with a detailed understanding of its internal operation. Long Short-Term Memory (LSTM) machine-learning networks can perform LMR with high accuracy levels. However, the internal behavior during classification is unknown, and they struggle to generalize when presented with novel users. The target problem addressed in this paper is understanding the LSTM classification behavior for LMR. A dataset of six locomotive activities (walking, stopped, stairs and ramps) from 22 non-amputee subjects is collected, capturing both steady-state and transitions between activities in natural environments. Non-amputees are used as a substitute for amputees to provide a larger dataset. The dataset is used to analyze the internal behavior of a reduced complexity LSTM network. This analysis identifies that the model primarily classifies activity type based on data around early stance. Evaluation of generalization for unseen subjects reveals low sensitivity to hyper-parameters and over-fitting to individuals\u2019 gait traits. Investigating the differences between individual subjects showed that gait variations between users primarily occur in early stance, potentially explaining the poor generalization. Adjustment of hyper-parameters alone could not solve this, demonstrating the need for individual personalization of models. The main achievements of the paper are (i) the better understanding of LSTM for LMR, (ii) demonstration of its low sensitivity to learning hyper-parameters when evaluating novel user generalization, and (iii) demonstration of the need for personalization of ML models to achieve acceptable accuracy.<\/jats:p>","DOI":"10.3390\/s21041264","type":"journal-article","created":{"date-parts":[[2021,2,12]],"date-time":"2021-02-12T16:12:10Z","timestamp":1613146330000},"page":"1264","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":66,"title":["Understanding LSTM Network Behaviour of IMU-Based Locomotion Mode Recognition for Applications in Prostheses and Wearables"],"prefix":"10.3390","volume":"21","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-3611-4846","authenticated-orcid":false,"given":"Freddie","family":"Sherratt","sequence":"first","affiliation":[{"name":"Department of Mechanical Engineering, University of Bath, Bath BA2 7AY, UK"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0082-3480","authenticated-orcid":false,"given":"Andrew","family":"Plummer","sequence":"additional","affiliation":[{"name":"Department of Mechanical Engineering, University of Bath, Bath BA2 7AY, UK"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4965-0341","authenticated-orcid":false,"given":"Pejman","family":"Iravani","sequence":"additional","affiliation":[{"name":"Department of Mechanical Engineering, University of Bath, Bath BA2 7AY, UK"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1968","published-online":{"date-parts":[[2021,2,10]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"602","DOI":"10.1016\/j.gaitpost.2008.04.005","article-title":"Compensatory mechanisms in below-knee amputee gait in response to increasing steady-state walking speeds","volume":"28","author":"Silverman","year":"2008","journal-title":"Gait Posture"},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"457","DOI":"10.1098\/rspb.2011.1194","article-title":"Bionic ankle-foot prosthesis normalizes walking gait for persons with leg amputation","volume":"279","author":"Herr","year":"2012","journal-title":"Proc. R. Soc. B Biol. Sci."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"191","DOI":"10.1016\/j.scispo.2017.07.011","article-title":"Assesment of physical activity in amputees: A systematic review of the literature","volume":"32","author":"Piazza","year":"2017","journal-title":"Sci. Sport."},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"349","DOI":"10.1016\/j.gaitpost.2018.03.040","article-title":"Energy expenditure in people with transtibial amputation walking with crossover and energy storing prosthetic feet: A randomized within-subject study","volume":"62","author":"McDonald","year":"2018","journal-title":"Gait Posture"},{"key":"ref_5","unstructured":"Ottobock (2020, December 15). Empower | Ottobock UK. Available online: https:\/\/www.ottobock.co.uk\/prosthetics\/lower-limb-prosthetics\/prosthetic-product-systems\/empower-ankle\/."},{"key":"ref_6","unstructured":"Blatchford (2020, December 15). ElanIC\u2014Waterproof Microprocessor Prosthetic Foot\u2014Blatchford. Available online: https:\/\/www.blatchford.co.uk\/products\/elanic\/."},{"key":"ref_7","unstructured":"Ossur (2020, December 15). PROPRIO FOOT\u00ae. Available online: https:\/\/www.ossur.com\/en-gb\/prosthetics\/feet\/proprio-foot."},{"key":"ref_8","doi-asserted-by":"crossref","unstructured":"Montgomery, J.R., and Grabowski, A.M. (2018). Use of a powered ankle\u2013foot prosthesis reduces the metabolic cost of uphill walking and improves leg work symmetry in people with transtibial amputations. J. R. Soc. Interface, 15.","DOI":"10.1098\/rsif.2018.0442"},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"1011","DOI":"10.1109\/TMECH.2019.2911685","article-title":"The Design, Control, and Testing of an Integrated Electrohydrostatic Powered Ankle Prosthesis","volume":"24","author":"Yu","year":"2019","journal-title":"IEEE\/ASME Trans. Mechatron."},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Labarri\u00e8re, F., Thomas, E., Calistri, L., Optasanu, V., Gueugnon, M., Ornetti, P., and Laroche, D. (2020). Machine Learning Approaches for Activity Recognition and\/or Activity Prediction in Locomotion Assistive Devices\u2014A Systematic Review. Sensors, 20.","DOI":"10.3390\/s20216345"},{"key":"ref_11","doi-asserted-by":"crossref","unstructured":"Murad, A., and Pyun, J.Y. (2017). Deep recurrent neural networks for human activity recognition. Sensors, 17.","DOI":"10.3390\/s17112556"},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"Samprita, S., Koshy, A.S., Megharjun, V.N., and Talasila, V. (2020, January 20\u201323). LSTM-Based Analysis of a Hip-Hop Movement. Proceedings of the 2020 6th International Conference on Control, Automation and Robotics (ICCAR), Singapore.","DOI":"10.1109\/ICCAR49639.2020.9108052"},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"486","DOI":"10.1109\/TBME.2004.840727","article-title":"Assessment of walking features from foot inertial sensing","volume":"52","author":"Sabatini","year":"2005","journal-title":"IEEE Trans. Biomed. Eng."},{"key":"ref_14","doi-asserted-by":"crossref","unstructured":"Tucker, M.R., Olivier, J., Pagel, A., Bleuler, H., Bouri, M., Lambercy, O., Del Mill\u00e1n, J.R., Riener, R., Vallery, H., and Gassert, R. (2015). Control strategies for active lower extremity prosthetics and orthotics: A review. J. Neuroeng. Rehabil., 12.","DOI":"10.1186\/1743-0003-12-1"},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"5","DOI":"10.1186\/s12938-016-0284-9","article-title":"Active lower limb prosthetics: A systematic review of design issues and solutions","volume":"15","author":"Windrich","year":"2016","journal-title":"Biomed. Eng. Online"},{"key":"ref_16","first-page":"1","article-title":"Investigation of timing to switch control mode in powered knee prostheses during task transitions","volume":"10","author":"Zhang","year":"2015","journal-title":"PLoS ONE"},{"key":"ref_17","doi-asserted-by":"crossref","unstructured":"Pedroli, E., Serino, S., Cipresso, P., Leo, G.D., Goulene, K., Morelli, S., Avenio, G.D., Stramba-badiale, M., Grigioni, M., and Gaggioli, A. (2019). An Immersive Cognitive Rehabilitation Program: A Case Study. International Conference on NeuroRehabilitation, Proceedings of the ICNR 2018: Converging Clinical and Engineering Research on Neurorehabilitation III, Pisa, Italy, 16\u201320 October 2018, Springer International Publishing.","DOI":"10.1007\/978-3-030-01845-0_142"},{"key":"ref_18","doi-asserted-by":"crossref","first-page":"90","DOI":"10.1177\/0309364610397087","article-title":"Factors affecting quality of life in lower limb amputees","volume":"35","author":"Sinha","year":"2011","journal-title":"Prosthetics Orthot. Int."},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Ponce, H., Miralles-Pechu\u00e1n, L., and De Lourdes Mart\u00ednez-Villase\u00f1or, M. (2016). A flexible approach for human activity recognition using artificial hydrocarbon networks. Sensors, 16.","DOI":"10.3390\/s16111715"},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"287","DOI":"10.1016\/j.gaitpost.2004.08.008","article-title":"Stair climbing detection during daily physical activity using a miniature gyroscope","volume":"22","author":"Coley","year":"2005","journal-title":"Gait Posture"},{"key":"ref_21","doi-asserted-by":"crossref","first-page":"1500","DOI":"10.1109\/TNSRE.2016.2636367","article-title":"A Real-Time Gait Event Detection for Lower Limb Prosthesis Control and Evaluation","volume":"25","author":"Maqbool","year":"2017","journal-title":"IEEE Trans. Neural Syst. Rehabil. Eng."},{"key":"ref_22","doi-asserted-by":"crossref","first-page":"2015","DOI":"10.1109\/TNSRE.2018.2870152","article-title":"Real-Time On-Board Recognition of Continuous Locomotion Modes for Amputees with Robotic Transtibial Prostheses","volume":"26","author":"Xu","year":"2018","journal-title":"IEEE Trans. Neural Syst. Rehabil. Eng."},{"key":"ref_23","doi-asserted-by":"crossref","first-page":"277","DOI":"10.1109\/TBME.2019.2912466","article-title":"A Comparison of Control Strategies in Commercial and Research Knee Prostheses","volume":"67","author":"Fluit","year":"2020","journal-title":"IEEE Trans. Biomed. Eng."},{"key":"ref_24","unstructured":"Goodfellow, I., Bengio, Y., and Courville, A. (2015). Deep Learning, MIT Press."},{"key":"ref_25","first-page":"5","article-title":"Supervised Sequence Labelling with Recurrent Neural Networks","volume":"Volume 385","author":"Graves","year":"2012","journal-title":"Studies in Computational Intelligence"},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"1735","DOI":"10.1162\/neco.1997.9.8.1735","article-title":"Long Short-Term Memory","volume":"9","author":"Hochreiter","year":"1997","journal-title":"Neural Comput."},{"key":"ref_27","unstructured":"Olah, C. (2020, December 07). Understanding LSTM Networks [Blog]. Available online: http:\/\/colah.github.io\/posts\/2015-08-Understanding-LSTMs\/."},{"key":"ref_28","unstructured":"Anguita, D., Ghio, A., Oneto, L., Parra, X., and Reyes-Ortiz, J.L. (2013, January 24\u201326). A public domain dataset for human activity recognition using smartphones. Proceedings of the ESANN 2013 Proceedings, 21st European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, Bruges, Belgium."},{"key":"ref_29","doi-asserted-by":"crossref","first-page":"3101","DOI":"10.1109\/JSEN.2019.2956901","article-title":"Human Action Recognition Using Deep Learning Methods on Limited Sensory Data","volume":"20","author":"Tufek","year":"2020","journal-title":"IEEE Sens. J."},{"key":"ref_30","doi-asserted-by":"crossref","first-page":"42","DOI":"10.1186\/s12984-018-0379-6","article-title":"Biomechanics and energetics of walking in powered ankle exoskeletons using myoelectric control versus mechanically intrinsic control","volume":"15","author":"Koller","year":"2018","journal-title":"J. Neuroeng. Rehabil."},{"key":"ref_31","doi-asserted-by":"crossref","first-page":"22089","DOI":"10.3390\/s150922089","article-title":"Inertial sensor-based gait recognition: A review","volume":"15","author":"Sprager","year":"2015","journal-title":"Sensors"},{"key":"ref_32","doi-asserted-by":"crossref","unstructured":"Chung, S., Lim, J., Noh, K.J., Kim, G., and Jeong, H. (2019). Sensor data acquisition and multimodal sensor fusion for human activity recognition using deep learning. Sensors, 19.","DOI":"10.3390\/s19071716"},{"key":"ref_33","doi-asserted-by":"crossref","unstructured":"Abbaspour, S., Fotouhi, F., Sedaghatbaf, A., Fotouhi, H., Vahabi, M., and Linden, M. (2020). A comparative analysis of hybrid deep learning models for human activity recognition. Sensors, 20.","DOI":"10.3390\/s20195707"},{"key":"ref_34","doi-asserted-by":"crossref","first-page":"179028","DOI":"10.1109\/ACCESS.2020.3027979","article-title":"A Deep Learning Approach for Human Activities Recognition From Multimodal Sensing Devices","volume":"8","author":"Ihianle","year":"2020","journal-title":"IEEE Access"},{"key":"ref_35","doi-asserted-by":"crossref","unstructured":"Mutegeki, R., and Han, D.S. (2020, January 19\u201321). A CNN-LSTM Approach to Human Activity Recognition. Proceedings of the 2020 International Conference on Artificial Intelligence in Information and Communication, ICAIIC 2020, Fukuoka, Japan.","DOI":"10.1109\/ICAIIC48513.2020.9065078"},{"key":"ref_36","first-page":"1","article-title":"Wearable Sensor-Based Human Activity Recognition Using Hybrid Deep Learning Techniques","volume":"2020","author":"Wang","year":"2020","journal-title":"Secur. Commun. Netw."},{"key":"ref_37","first-page":"8","article-title":"Smartwatch-based Human Activity Recognition Using Hybrid LSTM Network","volume":"2020","author":"Mekruksavanich","year":"2020","journal-title":"Proc. IEEE Sens."},{"key":"ref_38","doi-asserted-by":"crossref","unstructured":"Wang, C., Wu, X., Ma, Y., Wu, G., and Luo, Y. (2018). A flexible lower extremity exoskeleton robot with deep locomotion mode identification. Complexity, 2018.","DOI":"10.1155\/2018\/5712108"},{"key":"ref_39","doi-asserted-by":"crossref","first-page":"1032","DOI":"10.1109\/TNSRE.2019.2909585","article-title":"A cnn-based method for intent recognition using inertial measurement units and intelligent lower limb prosthesis","volume":"27","author":"Su","year":"2019","journal-title":"IEEE Trans. Neural Syst. Rehabil. Eng."},{"key":"ref_40","unstructured":"Dehghani, A., Glatard, T., and Shihab, E. (2019). Subject Cross Validation in Human Activity Recognition. arXiv."},{"key":"ref_41","doi-asserted-by":"crossref","unstructured":"Roggen, D., Calatroni, A., Rossi, M., Holleczek, T., Forster, K., Troster, G., Lukowicz, P., Bannach, D., Pirkl, G., and Ferscha, A. (2010, January 15\u201318). Collecting complex activity datasets in highly rich networked sensor environments. Proceedings of the 2010 Seventh International Conference on Networked Sensing Systems (INSS), Kassel, Germany.","DOI":"10.1109\/INSS.2010.5573462"},{"key":"ref_42","doi-asserted-by":"crossref","unstructured":"Sztyler, T., and Stuckenschmidt, H. (2017, January 13\u201317). Online personalization of cross-subjects based activity recognition models on wearable devices. Proceedings of the 2017 IEEE International Conference on Pervasive Computing and Communications, PerCom 2017, Kona, HI, USA.","DOI":"10.1109\/PERCOM.2017.7917864"},{"key":"ref_43","first-page":"249","article-title":"Understanding the difficulty of training deep feedforward neural networks","volume":"9","author":"Glorot","year":"2010","journal-title":"J. Mach. Learn. Res."},{"key":"ref_44","unstructured":"Kingma, D.P., and Ba, J.L. (2015, January 7\u20139). Adam: A method for stochastic optimization. Proceedings of the 3rd International Conference on Learning Representations, ICLR 2015\u2014Conference Track Proceedings, San Diego, CA, USA."}],"container-title":["Sensors"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1424-8220\/21\/4\/1264\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,11]],"date-time":"2025-10-11T05:22:30Z","timestamp":1760160150000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1424-8220\/21\/4\/1264"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,2,10]]},"references-count":44,"journal-issue":{"issue":"4","published-online":{"date-parts":[[2021,2]]}},"alternative-id":["s21041264"],"URL":"https:\/\/doi.org\/10.3390\/s21041264","relation":{},"ISSN":["1424-8220"],"issn-type":[{"value":"1424-8220","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,2,10]]}}}