Prediction of Sagittal Curve Profile of Rollover Footwear Based on Ankle Kinematics while Walking by Applying Neural Network Techniques

Document Type : Original Articles


1 Musculoskeletal Research Center, School of Rehabilitation Sciences, Isfahan University of Medical Sciences, Isfahan, Iran

2 Associate Professor, Medical Image and Signal Processing Research Center, Isfahan University of Medical Sciences AND Department of Control, School of Electrical and Computer Engineering, Isfahan University of Technology, Isfahan, Iran

3 Lecturer, Musculoskeletal Research Center, School of Rehabilitation Sciences, Isfahan University of Medical Sciences, Isfahan, Iran

4 Associate Professor, Medical Image and Signal Processing Research Center AND Musculoskeletal Research Center, School of Rehabilitation Sciences, Isfahan University of Medical Sciences, Isfahan, Iran



Introduction: Sagittal rocker profiles are one of the most commonly prescribed therapeutic footwear interventions to alter or adapt lower limb joints’ kinematics and kinetics. However, the prescription criteria for rocker profiles are commonly based on theoretical considerations. Thus, conducting experimental studies and experiment and error may result in their better prescription and use. A complementary approach is to use intelligent technology to predict curve profile to suit a specific joint position. The aim of this study was to predict sagittal curve profile of the rollover footwear from ankle kinematics while walking by applying an artificial neural network (ANN).Materials and Methods: In the present study, 20 healthy participants (with mean age of 33.1 years) walked on a straight path for 10 meters wearing two different shoes with two different sole curved profiles and ankle kinematic data were collected using reflective markers. The ANN was trained to associate set of ankle sagittal plane motions during stance phase with outsole curve profiles, and then, predict the latter based on the former. The ANN was trained using the data from 13 participants (control group) to obtain the model and the data from the remaining participants (intervention group) was used for the validation of the study purposes.Results: The achieved accuracy was very satisfactory, since the correlation coefficients between the predicted output and the actual curve profile in the validation data were higher than 0.95 for both types of rollover footwear.Conclusion: In this study, a novel algorithm was proposed for sole curve profile characterization of rollover footwear using an ANN model. The results of this study may be useful to designers of footwear, lower limb prostheses, orthoses, and walking casts/boots.


  1. Forghany S, Nester CJ, Richards B, Hatton AL, Liu A. Rollover footwear affects lower limb biomechanics during walking. Gait Posture 2014; 39(1): 205-12.
  2. Pol F, Forghany S, Nester C, Rahimi A. The effect of rollover footwear on head and trunk posture during standing. J Foot Ankle Res 2014; 7(Suppl 1): A21.
  3. Hutchins S, Bowker P, Geary N, Richards J. The biomechanics and clinical efficacy of footwear adapted with rocker profiles--evidence in the literature. Foot (Edinb) 2009; 19(3): 165-70.
  4. Forghany S, Nester CJ, Richards B. The effect of rollover footwear on the rollover function of walking. J Foot Ankle Res 2013; 6(1): 24.
  5. Findlow A, Goulermas JY, Nester C, Howard D, Kenney LP. Predicting lower limb joint kinematics using wearable motion sensors. Gait Posture 2008; 28(1): 120-6.
  6. Chapman JD. Improving the design of the curved rocker shoe for people with diabetes [PhD Thesis]. Salford, UK: University of Salford; 2014.
  7. Germani M, Mandolini M, Mengoni M, Nester C, Raffaeli R. Tools for design and validation of shoe lasts for diabetic patients. Footwear Science 2012; 4(3): 221-41.
  8. Winter DA. Biomechanics and motor control of human movement. 4th ed. Hoboken, NJ: Wiley; 2009.
  9. Levenberg K. A method for the solution of certain non-linear problems in least squares. Quart Appl Math 1944; 2(2): 164-8.
  10. Marquardt DW. An algorithm for least-squares estimation of nonlinear parameters. SIAM J Appl Math 1963; 11(2): 431-41.
  11. Hahn ME, Chou LS. A model for detecting balance impairment and estimating falls risk in the elderly. Ann Biomed Eng 2005; 33(6): 811-20.
  12. Oh SE, Choi A, Mun JH. Prediction of ground reaction forces during gait based on kinematics and a neural network model. J Biomech 2013; 46(14): 2372-80.
  13. Chan CW, Wang N, Cha KC. A parametric study of artificial neural network as a surrogate model for heel-toe running computation. J Biomech 2007; 40: S565.
  14. Lugade V, Lin V, Farley A, Chou LS. An artificial neural network estimation of gait balance control in the elderly using clinical evaluations. PLoS One 2014; 9(5): e97595.
  15. Mostafavizadeh M, Sadri AR, Zekri M. Walking pattern classification in children with cerebral palsy: A wavelet network approach. Proceedings of the 16th CSI International Symposium on Artificial Intelligence and Signal Processing (AISP 2012); 2012 May 2-3; Shiraz, Iran. p. 243-9.
  16. Barton JG, Lees A. Comparison of shoe insole materials by neural network analysis. Med Biol Eng Comput 1996; 34(6): 453-9.
  17. Rupereza MJ, Martin-Guerrerob JD, Monserrat C, Alcaniza M. Artificial neural networks for predicting dorsal pressures on the foot surface while walking. Expert Syst Appl 2012; 39(5): 5349-57.
  18. McKee PR, Rivard A. Biopsychosocial approach to orthotic intervention. Hand Ther 2011; 24(2): 155-63.
Volume 12, Issue 4 - Serial Number 4
September 2016
Pages 221-226
  • Receive Date: 16 January 2017
  • Revise Date: 19 April 2024
  • Accept Date: 22 May 2022