File
Meirovich Apelrot K1,2, Dr. Mendelsohn A3, Dr. Kafri M1
1University of Haifa, Department of Physical Therapy, Faculty of Social Welfare and Health Sciences, Haifa, Israel, 2'Clalit' Health Services, Haifa, Israel, 3University of Haifa, Sagol Department of Neurobiology, Haifa, Israel
Background: Vertical landing is related in many sports to lower limb injuries. An established body of knowledge links particular kinematic features of the landing and lower limb injuries, and emphasizes the need to modify landing patterns as a suitable prevention or rehabilitation intervention. Recently, there is a growing interest in the sports and rehabilitation fields in the potential of reinforcement learning (RL) as a tool for promoting motor behavior changes. Studies that examine how RL features affect ´real life´ motor skills are scarce.
Purpose: The goal of this study was to compare between the effects of motor training with positive reinforcement (social and monetary) vs. verbal feedback regarding performance (i.e. knowledge of performance) on changes in drop-jump landing patterns and in the kinematic variability within a training session.
Methods: Forty-two self-declared healthy young adults of both sexes were recruited and randomly assigned to either a monetary and social positive reinforcement group (PR group) or a summary verbal knowledge of performance (KP) feedback group (KP group). Participants from both groups performed three baseline trials of drop vertical jump test (DVJT), followed by 25 trials and a retention test 30 minutes thereafter. Each landing was assessed using the Landing Error Scoring System (LESS), a clinical tool designed to provide estimates of movement patterns in a jump-and-drop task and identify items related to the landing kinematics. For the purpose of the study, we used the PhysiMaxTM system, consisting of an immediate computerized motion analysis system based on the LESS items. During the training session, subjects got the PR or the KP feedback according to the kinematic assessment of the PhysiMaxTM system.
Results: There was a significant training effect (p 0.0001) in the variables of medio-lateral knee position, which consisted of knee valgus angle and knee flexion displacement, as the valgus angles were lower and the flexion angles of the knees were higher from baseline to the retention test. There was also a significant group effect (Pleft knee= 0.045, Pright knee=0.0002), such that the valgus angles in the KP group were lower than in the PR group. In these variables, there was also a significant interaction such that the participants who received KP improved more in their landing kinematics (p 0.0001 vs. p=0.0003). Both groups showed a significant improvement in the LESS total score (P 0.0001). There were no significant differences in kinematic variability within training sessions between groups.
Conclusion(s): RL is a possible approach in order to improve the kinematic pattern of landings in the frontal and sagittal planes. However, in the short term (30 min.), it was not proved to be better than KP. There is a need to keep investigating the characteristics of this form of learning in 'real-life' tasks and search for the optimal reward frequency, number of trials and sessions of training and their long-term effect.
Implications: Enhancing rehabilitation protocols and expanding the understanding of RL in diverse populations such as stroke patients whose rehabilitation process massively involves motor training that focuses at changing motor patterns is warranted.
Keywords: reinforcement learning, reward, vertical jump landing
Funding acknowledgements: This study was partially funded by an internal grant from the Faculty of Social Welfare and Health Sciences, Haifa University.
Purpose: The goal of this study was to compare between the effects of motor training with positive reinforcement (social and monetary) vs. verbal feedback regarding performance (i.e. knowledge of performance) on changes in drop-jump landing patterns and in the kinematic variability within a training session.
Methods: Forty-two self-declared healthy young adults of both sexes were recruited and randomly assigned to either a monetary and social positive reinforcement group (PR group) or a summary verbal knowledge of performance (KP) feedback group (KP group). Participants from both groups performed three baseline trials of drop vertical jump test (DVJT), followed by 25 trials and a retention test 30 minutes thereafter. Each landing was assessed using the Landing Error Scoring System (LESS), a clinical tool designed to provide estimates of movement patterns in a jump-and-drop task and identify items related to the landing kinematics. For the purpose of the study, we used the PhysiMaxTM system, consisting of an immediate computerized motion analysis system based on the LESS items. During the training session, subjects got the PR or the KP feedback according to the kinematic assessment of the PhysiMaxTM system.
Results: There was a significant training effect (p 0.0001) in the variables of medio-lateral knee position, which consisted of knee valgus angle and knee flexion displacement, as the valgus angles were lower and the flexion angles of the knees were higher from baseline to the retention test. There was also a significant group effect (Pleft knee= 0.045, Pright knee=0.0002), such that the valgus angles in the KP group were lower than in the PR group. In these variables, there was also a significant interaction such that the participants who received KP improved more in their landing kinematics (p 0.0001 vs. p=0.0003). Both groups showed a significant improvement in the LESS total score (P 0.0001). There were no significant differences in kinematic variability within training sessions between groups.
Conclusion(s): RL is a possible approach in order to improve the kinematic pattern of landings in the frontal and sagittal planes. However, in the short term (30 min.), it was not proved to be better than KP. There is a need to keep investigating the characteristics of this form of learning in 'real-life' tasks and search for the optimal reward frequency, number of trials and sessions of training and their long-term effect.
Implications: Enhancing rehabilitation protocols and expanding the understanding of RL in diverse populations such as stroke patients whose rehabilitation process massively involves motor training that focuses at changing motor patterns is warranted.
Keywords: reinforcement learning, reward, vertical jump landing
Funding acknowledgements: This study was partially funded by an internal grant from the Faculty of Social Welfare and Health Sciences, Haifa University.
Topic: Sport & sports injuries
Ethics approval required: Yes
Institution: The University of Haifa
Ethics committee: The Faculty of Social Welfare and Health Sciences
Ethics number: 087/17
All authors, affiliations and abstracts have been published as submitted.
