Advertisement

From deep learning to transfer learning for the prediction of skeletal muscle forces

  • Tien Tuan DaoEmail author
Original Article
  • 148 Downloads

Abstract

Skeletal muscle forces may be estimated using rigid musculoskeletal models and neural networks. Neural network (NN) approach has the advantages of real-time estimation ability and promising prediction accuracy. However, most of the developed NN models are based on conventional feedforward NNs, which do not take dynamic temporal relationships of the muscle force profiles into consideration. The objectives of this present paper are twofold: (1) to develop a recurrent deep neural network (RDNN) incorporating dynamic temporal relationships to estimate skeletal muscle forces from kinematics data during a gait cycle; (2) then to establish a transfer learning strategy to improve the accuracy of muscle force estimation. A long short-term memory (LSTM) model as a RDNN was developed and evaluated. A weight transfer strategy was established. Three databases were established for training and evaluation purposes. The predictions of rectus femoris, soleus, and tibialis anterior forces with developed LSTM network show root mean square error range of 2.4–84.6 N. Relative root mean square error (RMSE) deviations for internal and external validations are less than 5% and 10% for all analyzed muscles respectively. Pearson correlation coefficients (R) range of 0.95–0.999 showed perfect waveform similarity between data and predicted muscle forces for all analyzed muscles. The use of weight transfer leads to an improvement of 1.3% for the relative deviation between simulation outcome and LSMT prediction. This present study suggests that the recurrent deep neural network is a powerful and accurate computational tool for the prediction of skeletal muscle forces. Moreover, the coupling between this deep learning approach and a transfer learning strategy leads to improve the prediction accuracy. In future work, this coupling approach will be incorporated into a developed decision support tool for functional rehabilitation with real-time estimation and tracking of skeletal muscle forces.

Graphical abstract

Keywords

Deep learning Transfer learning Long short-term memory (LSTM) network Recurrent neural network Musculoskeletal modeling Skeletal muscle forces 

Notes

Acknowledgements

This work was carried out in the framework of the Labex MS2T. It was supported by the French Government, through the program “Investments for the future” managed by the National Agency for Research (Reference ANR-11-IDEX-0004-02).

Compliance with ethical standards

Conflict of interest

The author declares that there is no conflict of interest.

References

  1. 1.
    Eldridge FL (1975) Relationship between respiratory nerve and muscle activity and muscle force output. J Appl Physiol 39(4):567–574PubMedGoogle Scholar
  2. 2.
    Hatze H (1977) The relative contribution of motor unit recruitment and rat coding to the production of static isometric muscle force. Biol Cybern 27(1):21–25PubMedGoogle Scholar
  3. 3.
    Crago PE, Peckham PH, Thrope GB (1980) Modulation of muscle force by recruitment during intramuscular stimulation. IEEE Trans Biomed Eng 27(12):679–684PubMedGoogle Scholar
  4. 4.
    Crowninshield RD, Brand RA (1981) A physiologically based criterion of muscle force prediction in locomotion. J Biomech 14(11):793–801PubMedGoogle Scholar
  5. 5.
    Dao TT (2016) Rigid musculoskeletal models of the human body systems: a review. J Musculoskelet Res 19(3):1630001Google Scholar
  6. 6.
    Eskinazi I, Fregly BJ (2018) A computational framework for simultaneous estimation of muscle and joint contact forces and body motion using optimization and surrogate modeling. Med Eng Phys 54:56–64PubMedGoogle Scholar
  7. 7.
    Erdemir A, McLean S, Herzog W, van den Bogert AJ (2007) Model-based estimation of muscle forces exerted during movements. Clin Biomech 22(2):131–154Google Scholar
  8. 8.
    Trinler U, Hollands K, Jones R, Baker R (2018) A systematic review of approaches to modelling lower limb muscle forces during gait: applicability to clinical gait analyses. Gait Posture 61:353–361PubMedGoogle Scholar
  9. 9.
    Cecchini G, Lozito GM, Schmid M, Conforto S, Fulginei FG, Bibbo D (2014) Neural networks for muscle forces prediction in cycling. Algorithms 7(4):621–634Google Scholar
  10. 10.
    Vilimek M (2014) An artificial neural network approach and sensitivity analysis in predicting skeletal muscle forces. Acta Bioeng Biomech 16(3):119:127Google Scholar
  11. 11.
    Tibold R, Fuglevand AJ (2015) Prediction of muscle activity during loaded movements of the upper limb. J Neuroeng Rehabil 12(6):6PubMedPubMedCentralGoogle Scholar
  12. 12.
    Arjmand N, Ekrami O, Shirazi-Adl A, Plamondon A, Parnianpour M (2013) Relative performances of artificial neural network and regression mapping tools in evaluation of spinal loads and muscle forces during static lifting. J Biomech 46(8):1454–1462PubMedGoogle Scholar
  13. 13.
    Dariani S, Keshavarz M, Parviz M, Raoufy MR, Gharibzadeh S (2007) Modeling force-velocity relation in skeletal muscle isotonic contraction using an artificial neural network. Biosystems 90(2):529–534PubMedGoogle Scholar
  14. 14.
    Liu MM, Herzog W, Savelberg HH (1999) Dynamic muscle force predictions from EMG: an artificial neural network approach. J Electromyogr Kinesiol 9(6):391–400PubMedGoogle Scholar
  15. 15.
    Atzori M, Cognolato M1, Müller H (2016) Deep learning with convolutional neural networks applied to electromyography data: a resource for the classification of movements for prosthetic hands. Front Neurorobot 10(9)Google Scholar
  16. 16.
    Soda P, Mazzoleni S, Cavallo G, Guglielmelli E, Iannello G (2010) Human movement onset detection from isometric force and torque measurements: a supervised pattern recognition approach. Artif Intell Med 50(1):55–61PubMedGoogle Scholar
  17. 17.
    Ameri A, Scheme EJ, Kamavuako EN, Englehart KB, Parker PA (2014) Real-time, simultaneous myoelectric control using force and position-based training paradigms. IEEE Trans Biomed Eng 61(2):279–287PubMedGoogle Scholar
  18. 18.
    Mobasser F, Hashtrudi-Zaad K (2012) A comparative approach to hand force estimation using artificial neural networks. Biomed Eng Comput Biol 4:1–15PubMedPubMedCentralGoogle Scholar
  19. 19.
    Youn W, Kim J (2011) Feasibility of using an artificial neural network model to estimate the elbow flexion force from mechanomyography. J Neurosci Methods 194(2):386–393PubMedGoogle Scholar
  20. 20.
    Uchiyama T, Bessho T, Akazawa K (1998) Static torque-angle relation of human elbow joint estimated with artificial neural network technique. J Biomech 31(6):545–554PubMedGoogle Scholar
  21. 21.
    Choi C, Kwon S, Park W, Lee HD, Kim J Real-time pinch force estimation by surface electromyography using an artificial neural network. Med Eng Phys 32(5):429–436Google Scholar
  22. 22.
    Savelberg HH, Herzog W (1997) Prediction of dynamic tendon forces from electromyographic signals: an artificial neural network approach. J Neurosci Methods 78(1–2):65–74PubMedGoogle Scholar
  23. 23.
    Grandjean B, Hepp-Reymond MC, Maier MA (2007) The functional role of different neural activation profiles during precision grip: an artificial neural network approach. J Physiol 101(1–3):9–21Google Scholar
  24. 24.
    Wang L, Buchanan TS (2002) Prediction of joint moments using a neural network model of muscle activations from EMG signals. IEEE Trans Neural Syst Rehabil Eng 10(1):30–37PubMedGoogle Scholar
  25. 25.
    de Vries WH, Veeger HE, Baten CT, van der Helm FC Determining a long term ambulatory load profile of the shoulder joint: neural networks predicting input for a musculoskeletal model. Hum Mov Sci 31(2):419–428Google Scholar
  26. 26.
    Deb C, Zhang F, Yang J, Lee SE, Shah KW (2017) A review on time series forecasting techniques for building energy consumption. Renew Sust Energ Rev 74:902–924Google Scholar
  27. 27.
    De Gooijer JG, Hyndman RJ (2006) 25 years of time series forecasting. Int J Forecast 22:443–473Google Scholar
  28. 28.
    Hou Y, Zurada JM, Karwowski W, Marras WS, Davis K (2007) Estimation of the dynamic spinal forces using a recurrent fuzzy neural network. IEEE Trans Syst Man Cybern B 37(1):100–109Google Scholar
  29. 29.
    Hochreiter S, Schmidhuber J (1997) Long short-term memory. Neural Comput 9(8):1735–1780PubMedPubMedCentralGoogle Scholar
  30. 30.
    Gers FA, Schmidhuber J (2000) Recurrent nets that time and count. Proc. of the IEEE-INNS-ENNS International Joint Conference on Neural Networks: 189–194Google Scholar
  31. 31.
    Kingma DP, Ba J (2015) Adam: a method for stochastic optimization. International Conference on Learning Representations, 1–13Google Scholar
  32. 32.
    Liu MQ, Anderson FC, Schwartz MH, Delp SL (2008) Muscle contributions to support and progression over a range of walking speeds. J Biomech 41(15):3243–3252PubMedPubMedCentralGoogle Scholar
  33. 33.
    Delp SL, Anderson FC, Arnold AS, Loan P, Habib A, John CT, Guendelman E, Thelen DG (2007) OpenSim: open-source software to create and analyze dynamic simulations of movement. IEEE Trans Biomed Eng 54(11):1940–1950PubMedGoogle Scholar
  34. 34.
    Bischoff W, Cremers H, Fieger W (1991) Normal distribution assumption and least squares estimation function in the model of polynomial regression. J Multivar Anal 36(1):1–17Google Scholar
  35. 35.
    Dao TT, Ho B, Tho MC (2015) Assessment of parameter uncertainty in rigid musculoskeletal simulation using a probabilistic approach. J Musculoskelet Res 18(3):1550013Google Scholar
  36. 36.
    Dao TT, Pouletaut P, Charleux F, Lazáry Á, Eltes P, Varga PP, Ho B, Tho MC (2015) Multimodal medical imaging (CT and dynamic MRI) data and computer-graphics multi-physical model for the estimation of patient specific lumbar spine muscle forces. Data Knowl Eng 96-97:3–18Google Scholar
  37. 37.
    Geronilla KB, Miller GR, Mowrey KF, Wu JZ, Kashon ML, Brumbaugh K, Reynolds J, Hubbs A, Cutlip RG (2003) Dynamic force responses of skeletal muscle during stretch-shortening cycles. Eur J Appl Physiol 90(1–2):144–153PubMedGoogle Scholar
  38. 38.
    Rajagopal A, Dembia CL, DeMers MS, Delp DD, Hicks JL, Delp SL (2016) Full body musculoskeletal model for muscle-driven simulation of human gait. IEEE Trans Biomed Eng 63(10):2068–2079PubMedPubMedCentralGoogle Scholar
  39. 39.
    Delp SL, Zajac FE (1992) Force- and moment-generating capacity of lower-limb muscles before and after tendon lengthening. Clin Orthop Relat Res 284:247–259Google Scholar
  40. 40.
    Hinton GE, Salakhutdinov RR (2006) Reducing the dimensionality of data with neural networks. Science 313(5786):504–507PubMedPubMedCentralGoogle Scholar
  41. 41.
    Fan AX, Dakpé S, Dao TT, Pouletaut P, Rachik M, Ho B, Tho MC (2017) MRI-based finite element modeling of facial mimics: a case study on the paired zygomaticus major muscles. Computer Methods in Biomechanics and Biomedical Engineering 20(9):919–928PubMedGoogle Scholar
  42. 42.
    Dao TT, Rassineux A, Charleux F, Ho Ba Tho MC (2015) A robust protocol for the creation of patient specific finite element models of the musculoskeletal system from medical imaging data. Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization 3(3):136–146Google Scholar
  43. 43.
    Liang L, Liu M, Martin C, Sun W A deep learning approach to estimate stress distribution: a fast and accurate surrogate of finite-element analysis. J R Soc Interface 15(138):20170844, 2018Google Scholar
  44. 44.
    Nazemi SM, Amini M, Kontulainen SA, Milner JS, Holdsworth DW, Masri BA, Wilson DR, Johnston JD (2017) Optimizing finite element predictions of local subchondral bone structural stiffness using neural network-derived density-modulus relationships for proximal tibial subchondral cortical and trabecular bone. Clin Biomech 41:1–8Google Scholar
  45. 45.
    Xiao C, Choi E, Sun J (2018. In Press) Opportunities and challenges in developing deep learning models using electronic health records data: a systematic review. J Am Med Inform Assoc 25:1419–1428.  https://doi.org/10.1093/jamia/ocy068 PubMedPubMedCentralGoogle Scholar
  46. 46.
    Camacho DM, Collins KM, Powers RK, Costello JC, Collins JJ (2018) Next-generation machine learning for biological networks. Cell 173(7):1581–1592PubMedGoogle Scholar
  47. 47.
    Faust O, Hagiwara Y, Hong TJ, Lih OS, Acharya UR (2018) Deep learning for healthcare applications based on physiological signals: a review. Comput Methods Prog Biomed 161:1–13Google Scholar
  48. 48.
    TT Dao, H Tannous, P Pouletaut, D Gamet, D Istrate, MC Ho Ba Tho. Interactive and connected rehabilitation systems for E-health Innovation and Research in BioMedical Engineering. IRBM 37(5–6): 289–296, 2006Google Scholar

Copyright information

© International Federation for Medical and Biological Engineering 2018

Authors and Affiliations

  1. 1.Sorbonne University, Université de Technologie de Compiègne, CNRS, UMR 7338 Biomechanics and Bioengineering, Centre de recherche RoyallieuCompiègne CedexFrance

Personalised recommendations