Inner-Cycle Phases Can Be Estimated from a Single Inertial Sensor by Long Short-Term Memory Neural Network in Roller-Ski Skating
Abstract
:1. Introduction
2. Materials and Methods
2.1. Participants
2.2. Experimental Setup
3. Calculations
3.1. Reference System
3.2. Machine Learning Model
3.3. Analysis
4. Results
5. Discussion
6. Conclusions
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Acknowledgments
Conflicts of Interest
References
- Halilaj, E.; Rajagopal, A.; Fiterau, M.; Hicks, J.L.; Hastie, T.J.; Delp, S.L. Machine learning in human movement biomechanics: Best practices, common pitfalls, and new opportunities. J. Biomech. 2018, 81, 1–11. [Google Scholar] [CrossRef]
- Palmerini, L.; Mellone, S.; Avanzolini, G.; Valzania, F.; Chiari, L. Quantification of motor impairment in Parkinson’s disease using an instrumented timed up and go test. IEEE Trans. Neural Syst. Rehabil. Eng. 2013, 21, 664–673. [Google Scholar] [CrossRef]
- Buchman, A.S.; Leurgans, S.E.; Weiss, A.; VanderHorst, V.; Mirelman, A.; Dawe, R.; Barnes, L.L.; Wilson, R.S.; Hausdorff, J.M.; Bennett, D.A. Associations between quantitative mobility measures derived from components of conventional mobility testing and parkinsonian gait in older adults. PLoS ONE 2014, 9, e86262. [Google Scholar] [CrossRef]
- Biswas, D.; Cranny, A.; Gupta, N.; Maharatna, K.; Achner, J.; Klemke, J.; Jöbges, M.; Ortmann, S. Recognizing upper limb movements with wrist worn inertial sensors using k-means clustering classification. Hum. Mov. Sci. 2015, 40, 59–76. [Google Scholar] [CrossRef] [Green Version]
- Camomilla, V.; Bergamini, E.; Fantozzi, S.; Vannozzi, G. Trends supporting the in-field use of wearable inertial sensors for sport performance evaluation: A systematic review. Sensors 2018, 18, 873. [Google Scholar] [CrossRef] [Green Version]
- Rindal, O.M.H.; Seeberg, T.M.; Tjønnås, J.; Haugnes, P.; Sandbakk, Ø. Automatic classification of sub-techniques in classical cross-country skiing using a machine learning algorithm on micro-sensor data. Sensors 2018, 18, 75. [Google Scholar] [CrossRef] [Green Version]
- Uddin, M.Z.; Seeberg, T.M.; Kocbach, J.; Liverud, A.E.; Gonzalez, V.; Sandbakk, Ø.; Meyer, F. Estimation of Mechanical Power Output Employing Deep Learning on Inertial Measurement Data in Roller Ski Skating. Sensors 2021, 21, 6500. [Google Scholar] [CrossRef]
- Gløersen, Ø.; Gilgien, M. Classification of Cross-Country Ski Skating Sub-Technique Can Be Automated Using Carrier-Phase Differential GNSS Measurements of the Head’s Position. Sensors 2021, 21, 2705. [Google Scholar] [CrossRef]
- Stetter, B.J.; Ringhof, S.; Krafft, F.C.; Sell, S.; Stein, T. Estimation of knee joint forces in sport movements using wearable sensors and machine learning. Sensors 2019, 19, 3690. [Google Scholar] [CrossRef] [Green Version]
- Stetter, B.J.; Krafft, F.C.; Ringhof, S.; Stein, T.; Sell, S. A Machine Learning and Wearable Sensor Based Approach to Estimate External Knee Flexion and Adduction Moments During Various Locomotion Tasks. Front. Bioeng. Biotechnol. 2020, 8, 9. [Google Scholar] [CrossRef]
- Dorschky, E.; Nitschke, M.; Martindale, C.F.; van den Bogert, A.J.; Koelewijn, A.D.; Eskofier, B.M. CNN-Based Estimation of Sagittal Plane Walking and Running Biomechanics From Measured and Simulated Inertial Sensor Data. Front. Bioeng. Biotechnol. 2020, 8, 604. [Google Scholar] [CrossRef] [PubMed]
- Tan, T.; Strout, Z.A.; Shull, P.B. Accurate Impact Loading Rate Estimation during Running via a Subject-Independent Convolutional Neural Network Model and Optimal IMU Placement. IEEE J. Biomed. Health Inform. 2021, 25, 1215–1222. [Google Scholar] [CrossRef] [PubMed]
- Wang, J.; Wang, Z.; Qiu, S.; Xu, J.; Zhao, H.; Fortino, G.; Habib, M. A selection framework of sensor combination feature subset for human motion phase segmentation. Inf. Fusion 2021, 70, 1–11. [Google Scholar] [CrossRef]
- Lempereur, M.; Rousseau, F.; Rémy-néris, O.; Pons, C.; Houx, L. Short communication A new deep learning-based method for the detection of gait events in children with gait disorders: Proof-of-concept and concurrent validity. J. Biomech. 2019, 98, 109490. [Google Scholar] [CrossRef]
- Meyer, F.; Borrani, F. Estimating alpine skiers’ energetics and turn radius using different morphological points. Front. Physiol. 2018, 9, 1541. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Alcantara, R.S.; Day, E.M.; Hahn, M.E.; Grabowski, A.M. Sacral acceleration can predict whole-body kinetics and stride kinematics across running speeds. PeerJ 2021, 9, e11199. [Google Scholar] [CrossRef] [PubMed]
- Patoz, A.; Lussiana, T.; Breine, B.; Gindre, C.; Malatesta, D. A Single Sacral-Mounted Inertial Measurement Unit to Estimate Peak Vertical Ground Reaction Force, Contact Time, and Flight Time in Running. Sensors 2022, 22, 784. [Google Scholar] [CrossRef]
- Fasel, B.; Favre, J.; Chardonnens, J.; Gremion, G.; Aminian, K. An inertial sensor-based system for spatio-temporal analysis in classic cross-country skiing diagonal technique. J. Biomech. 2015, 48, 3199–3205. [Google Scholar] [CrossRef] [PubMed]
- Meyer, F.; Seeberg, T.M.; Kocbach, J.; Danielsen, J.; Sandbakk, Ø.; Austeng, A. Validation of temporal parameters within the skating sub-techniques when roller skiing on a treadmill, using inertial measurement units. PLoS ONE 2022, 17, e0270331. [Google Scholar] [CrossRef] [PubMed]
- Meyer, F.; Lund-Hansen, M.; Kocbach, J.; Seeberg, T.; Austeng, A.; Sandback, Ø. Inertial sensors-based estimation of temporal events in skating sub-techniques while in-field roller skiing. Preprints 2022, 2022110477. [Google Scholar] [CrossRef]
- Andersson, E.; Supej, M.; Sandbakk, Ø.; Sperlich, B.; Stöggl, T.; Holmberg, H.C. Analysis of sprint cross-country skiing using a differential global navigation satellite system. Eur. J. Appl. Physiol. 2010, 110, 585–595. [Google Scholar] [CrossRef] [PubMed]
- Meyer, F.; Kocbach, J.; Tjønnås, J.; Seeberg, T.M.; Austeng, A.; Sandbakk, Ø.; Danielsen, J. Temporal and kinematic patterns distinguishing the G2 from the G4 skating sub-technique. Sport. Biomech. 2021, 1–18. [Google Scholar] [CrossRef] [PubMed]
- Falbriard, M.; Meyer, F.; Mariani, B.; Millet, G.P.; Aminian, K. Accurate estimation of running temporal parameters using foot-worn inertial sensors. Front. Physiol. 2018, 9, 610. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Seeberg, T.M.; Kocbach, J.; Danielsen, J.; Noordhof, D.A.; Skovereng, K.; Meyer, F.; Sandbakk, Ø. Physiological and Biomechanical Responses to Cross-Country Skiing in Varying Terrain: Low- vs. High-Intensity. Front. Physiol. 2021, 12, 741573. [Google Scholar] [CrossRef] [PubMed]
A | REF Cycles | ML Cycles | PON Error (ms) Mean ± SD | Missed IC (%) | Extra IC (%) | POFF Error (ms) Mean ± SD | Missed TC (%) | Extra TC (%) | CT Error (ms) Mean ± SD | FT Error (ms) Mean ± SD | ||||||||
SJ1 | 1120 | 1157 | 29 | ± | 40 | 3 | 6 | 78 | ± | 73 | 9 | 12 | 57 | ± | 84 | −59 | ± | 87 |
SJ2 | 1080 | 1079 | 30 | ± | 57 | 6 | 6 | −41 | ± | 93 | 10 | 10 | −69 | ± | 114 | 65 | ± | 110 |
SJ3 | 1063 | 762 | 30 | ± | 60 | 31 | 4 | 39 | ± | 67 | 31 | 4 | 12 | ± | 75 | −10 | ± | 71 |
SJ4 | 1043 | 734 | −85 | ± | 63 | 35 | 7 | −47 | ± | 83 | 34 | 6 | 23 | ± | 77 | −26 | ± | 78 |
SJ5 | 1108 | 851 | −50 | ± | 71 | 39 | 20 | 7 | ± | 113 | 40 | 21 | 35 | ± | 103 | −44 | ± | 100 |
SJ6 | 1760 | 1120 | 95 | ± | 153 | 79 | 67 | 155 | ± | 129 | 71 | 54 | 7 | ± | 81 | −11 | ± | 69 |
SJ7 | 1095 | 597 | −27 | ± | 70 | 46 | 2 | 42 | ± | 92 | 49 | 6 | 53 | ± | 91 | −46 | ± | 92 |
SJ8 | 1265 | 1203 | 6 | ± | 70 | 15 | 11 | −43 | ± | 96 | 15 | 10 | −39 | ± | 115 | 45 | ± | 116 |
SJ9 | 987 | 317 | 121 | ± | 162 | 89 | 67 | 130 | ± | 163 | 89 | 66 | 4 | ± | 59 | 0 | ± | 16 |
All | 10521 | 7820 | 3 | ± | 90 | 38.0 | 21.0 | 17 | ± | 101 | 38.5 | 20.9 | 9 | ± | 117 | −9 | ± | 119 |
B | REF cycles | ML cycles | PON error (ms) mean ± SD | missed IC (%) | extra IC (%) | POFF error (ms) mean ± SD | missed TC (%) | extra TC (%) | CT error (ms) mean ± SD | FT error (ms) mean ± SD | ||||||||
SJ1 | 1120 | 1108 | 6 | ± | 43 | 8 | 7 | 59 | ± | 69 | 8 | 7 | 58 | ± | 84 | −59 | ± | 85 |
SJ2 | 1080 | 1044 | 25 | ± | 32 | 7 | 4 | 46 | ± | 90 | 8 | 5 | 21 | ± | 99 | −24 | ± | 98 |
SJ3 | 1063 | 1101 | 13 | ± | 45 | 4 | 8 | −35 | ± | 66 | 7 | 10 | −43 | ± | 77 | 42 | ± | 78 |
SJ4 | 1043 | 1003 | −24 | ± | 31 | 21 | 18 | 17 | ± | 64 | 21 | 18 | 34 | ± | 68 | −36 | ± | 67 |
SJ5 | 1108 | 1212 | 1 | ± | 60 | 8 | 16 | 35 | ± | 113 | 14 | 22 | 22 | ± | 119 | −32 | ± | 122 |
SJ6 | 1760 | 1498 | −51 | ± | 39 | 18 | 4 | 18 | ± | 65 | 20 | 6 | 58 | ± | 76 | −56 | ± | 76 |
SJ7 | 1095 | 1106 | 12 | ± | 59 | 11 | 12 | −14 | ± | 91 | 11 | 12 | −25 | ± | 82 | 29 | ± | 76 |
SJ8 | 1265 | 1205 | 7 | ± | 55 | 11 | 7 | −32 | ± | 76 | 12 | 8 | −35 | ± | 83 | 39 | ± | 81 |
SJ9 | 987 | 877 | −5 | ± | 78 | 39 | 32 | 43 | ± | 116 | 42 | 35 | 53 | ± | 89 | −51 | ± | 91 |
All | 10521 | 10154 | −5 | ± | 55 | 13.9 | 11.7 | 14 | ± | 89 | 15.8 | 13.5 | 19 | ± | 100 | −19 | ± | 101 |
C | REF cycles | ML cycles | PON error (ms) mean ± SD | missed IC (%) | extra IC (%) | POFF error (ms) mean ± SD | missed TC (%) | extra TC (%) | CT error (ms) mean ± SD | FT error (ms) mean ± SD | ||||||||
SJ1 | 1120 | 1156 | 56 | ± | 29 | 1 | 4 | 63 | ± | 52 | 1 | 4 | 13 | ± | 66 | −14 | ± | 66 |
SJ2 | 1080 | 1104 | 90 | ± | 30 | 2 | 4 | 69 | ± | 68 | 3 | 5 | −20 | ± | 76 | 19 | ± | 78 |
SJ3 | 1063 | 1056 | −18 | ± | 42 | 7 | 6 | −32 | ± | 40 | 7 | 6 | −14 | ± | 63 | 11 | ± | 60 |
SJ4 | 1043 | 984 | −61 | ± | 32 | 8 | 2 | −28 | ± | 52 | 8 | 2 | 29 | ± | 58 | −31 | ± | 60 |
SJ5 | 1108 | 1127 | −40 | ± | 60 | 2 | 3 | 3 | ± | 45 | 2 | 4 | 38 | ± | 85 | −38 | ± | 89 |
SJ6 | 1760 | 1572 | −5 | ± | 31 | 13 | 3 | 36 | ± | 44 | 16 | 6 | 36 | ± | 54 | −35 | ± | 54 |
SJ7 | 1095 | 1042 | −36 | ± | 36 | 6 | 1 | −17 | ± | 45 | 6 | 1 | 20 | ± | 59 | −19 | ± | 58 |
SJ8 | 1265 | 1292 | 18 | ± | 51 | 2 | 5 | −5 | ± | 68 | 3 | 5 | −24 | ± | 63 | 24 | ± | 63 |
SJ9 | 987 | 977 | −26 | ± | 45 | 6 | 5 | −15 | ± | 76 | 6 | 5 | 9 | ± | 83 | −9 | ± | 84 |
All | 10521 | 10310 | −1 | ± | 64 | 5.0 | 3.7 | 11 | ± | 69 | 5.6 | 4.2 | 11 | ± | 73 | −12 | ± | 74 |
A | REF cycles | ML cycles | PON error (ms) mean ± SD | missed IC (%) | extra IC (%) | POFF error (ms) mean ± SD | missed TC (%) | extra TC (%) | CT error (ms) mean ± SD | FT error (ms) mean ± SD | ||||||||
SJ1 | 1120 | 1157 | 52 | ± | 65 | 10 | 5 | −9 | ± | 56 | 10 | 4 | −56 | ± | 74 | 54 | ± | 75 |
SJ2 | 1080 | 1079 | −66 | ± | 60 | 13 | 7 | −62 | ± | 60 | 15 | 9 | 1 | ± | 82 | 0 | ± | 81 |
SJ3 | 1063 | 762 | 29 | ± | 72 | 16 | 7 | 13 | ± | 45 | 15 | 6 | −14 | ± | 77 | 16 | ± | 78 |
SJ4 | 1043 | 734 | −24 | ± | 62 | 6 | 3 | −14 | ± | 35 | 5 | 3 | 9 | ± | 64 | −10 | ± | 65 |
SJ5 | 1108 | 851 | −13 | ± | 92 | 29 | 18 | −44 | ± | 78 | 29 | 17 | −14 | ± | 82 | 18 | ± | 91 |
SJ6 | 1760 | 1120 | 62 | ± | 152 | 95 | 94 | −60 | ± | 165 | 96 | 95 | 1 | ± | 10 | 1 | ± | 23 |
SJ7 | 1095 | 597 | 53 | ± | 94 | 20 | 13 | 7 | ± | 53 | 15 | 8 | −33 | ± | 93 | 38 | ± | 96 |
SJ8 | 1265 | 1203 | 10 | ± | 72 | 7 | 7 | 55 | ± | 47 | 5 | 5 | 42 | ± | 88 | −40 | ± | 90 |
SJ9 | 987 | 317 | −32 | ± | 167 | 88 | 81 | −27 | ± | 145 | 91 | 86 | 3 | ± | 30 | 5 | ± | 56 |
All | 10521 | 7820 | 5 | ± | 91 | 31.5 | 26.0 | −5 | ± | 72 | 31.2 | 25.8 | −7 | ± | 92 | 9 | ± | 95 |
B | REF cycles | ML cycles | PON error (ms) mean ± SD | missed IC (%) | extra IC (%) | POFF error (ms) mean ± SD | missed TC (%) | extra TC (%) | CT error (ms) mean ± SD | FT error (ms) mean ± SD | ||||||||
SJ1 | 1120 | 1108 | 41 | ± | 68 | 18 | 7 | 6 | ± | 46 | 17 | 6 | −27 | ± | 67 | 28 | ± | 72 |
SJ2 | 1080 | 1044 | −1 | ± | 88 | 12 | 6 | −43 | ± | 52 | 11 | 4 | −33 | ± | 89 | 32 | ± | 88 |
SJ3 | 1063 | 1101 | −5 | ± | 66 | 13 | 5 | 12 | ± | 36 | 13 | 5 | 15 | ± | 73 | −15 | ± | 74 |
SJ4 | 1043 | 1003 | 20 | ± | 58 | 6 | 25 | 13 | ± | 40 | 5 | 25 | −5 | ± | 69 | 4 | ± | 72 |
SJ5 | 1108 | 1212 | −19 | ± | 77 | 22 | 16 | −38 | ± | 69 | 22 | 16 | −11 | ± | 75 | 15 | ± | 80 |
SJ6 | 1760 | 1498 | −22 | ± | 47 | 8 | 4 | −18 | ± | 47 | 7 | 4 | 4 | ± | 56 | −3 | ± | 57 |
SJ7 | 1095 | 1106 | 26 | ± | 82 | 18 | 17 | 13 | ± | 70 | 16 | 15 | −11 | ± | 65 | 12 | ± | 70 |
SJ8 | 1265 | 1205 | −1 | ± | 57 | 4 | 4 | 28 | ± | 39 | 4 | 4 | 28 | ± | 61 | −27 | ± | 64 |
SJ9 | 987 | 877 | 8 | ± | 103 | 28 | 39 | 4 | ± | 82 | 26 | 38 | −13 | ± | 78 | 8 | ± | 81 |
All | 10521 | 10154 | 3 | ± | 74 | 14.2 | 13.8 | −3 | ± | 70 | 13.5 | 13.1 | −5 | ± | 76 | 5 | ± | 78 |
C | REF cycles | ML cycles | PON error (ms) mean ± SD | missed IC (%) | extra IC (%) | POFF error (ms) mean ± SD | missed TC (%) | extra TC (%) | CT error (ms) mean ± SD | FT error (ms) mean ± SD | ||||||||
SJ1 | 1120 | 1156 | 67 | ± | 53 | 15 | 10 | 24 | ± | 39 | 12 | 7 | −37 | ± | 60 | 40 | ± | 63 |
SJ2 | 1080 | 1104 | 38 | ± | 64 | 11 | 6 | 39 | ± | 44 | 9 | 4 | 3 | ± | 72 | −2 | ± | 77 |
SJ3 | 1063 | 1056 | −24 | ± | 57 | 14 | 6 | −23 | ± | 46 | 14 | 5 | 4 | ± | 66 | −3 | ± | 68 |
SJ4 | 1043 | 984 | −35 | ± | 51 | 6 | 27 | −32 | ± | 46 | 5 | 27 | 7 | ± | 69 | −5 | ± | 69 |
SJ5 | 1108 | 1127 | −47 | ± | 63 | 12 | 17 | −44 | ± | 61 | 12 | 17 | 2 | ± | 61 | −2 | ± | 66 |
SJ6 | 1760 | 1572 | −5 | ± | 50 | 6 | 4 | 13 | ± | 49 | 6 | 4 | 18 | ± | 57 | −17 | ± | 59 |
SJ7 | 1095 | 1042 | 0 | ± | 78 | 21 | 18 | −14 | ± | 74 | 19 | 16 | −13 | ± | 65 | 16 | ± | 67 |
SJ8 | 1265 | 1292 | 17 | ± | 47 | 4 | 5 | 35 | ± | 40 | 4 | 4 | 18 | ± | 48 | −17 | ± | 52 |
SJ9 | 987 | 977 | 3 | ± | 84 | 25 | 36 | 0 | ± | 83 | 25 | 36 | −1 | ± | 67 | −5 | ± | 68 |
All | 10521 | 10310 | 2 | ± | 70 | 12.5 | 14.2 | 2 | ± | 62 | 11.8 | 13.4 | 0 | ± | 66 | 0 | ± | 69 |
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |
© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Meyer, F.; Lund-Hansen, M.; Seeberg, T.M.; Kocbach, J.; Sandbakk, Ø.; Austeng, A. Inner-Cycle Phases Can Be Estimated from a Single Inertial Sensor by Long Short-Term Memory Neural Network in Roller-Ski Skating. Sensors 2022, 22, 9267. https://doi.org/10.3390/s22239267
Meyer F, Lund-Hansen M, Seeberg TM, Kocbach J, Sandbakk Ø, Austeng A. Inner-Cycle Phases Can Be Estimated from a Single Inertial Sensor by Long Short-Term Memory Neural Network in Roller-Ski Skating. Sensors. 2022; 22(23):9267. https://doi.org/10.3390/s22239267
Chicago/Turabian StyleMeyer, Frédéric, Magne Lund-Hansen, Trine M. Seeberg, Jan Kocbach, Øyvind Sandbakk, and Andreas Austeng. 2022. "Inner-Cycle Phases Can Be Estimated from a Single Inertial Sensor by Long Short-Term Memory Neural Network in Roller-Ski Skating" Sensors 22, no. 23: 9267. https://doi.org/10.3390/s22239267