Periodic Reporting for period 2 - REPAIRS (Relearning Perception Action In Rehabilitation from a Systems perspective)
Période du rapport: 2023-01-01 au 2025-07-31
Rehabilitation of movement disorders aims to restore functional ability through re-learning perception-action couplings. However, current day rehabilitation leads to marginal improvements of daily functioning at best. To improve effectiveness of rehabilitation training we started from the idea that functional behaviour emerges from non-linear interactions within and between the neural, movement, agent-environment and the agent-agent systems, a so-called systems perspective. We used this perspective to develop fundamental knowledge on learning perception-action couplings. To this end, we performed experiments involving individual and dyadic perception-action behaviour. We also used mathematical models to reveal mechanisms producing the behaviours. An important focus of the project was on changes in perception-action behaviour when learning a novel skill. This knowledge was then translated to training applications for rehabilitation to improve rehabilitation effectiveness. We trained a new generation of scientists to exploit a systems perspective in understanding perception-action learning and its application in rehabilitation training. The knowledge developed aimed to provide the basis for developing applications for rehabilitation that could improve the daily lives of people suffering from stroke, people with a visual impairment, with anterior cruciate ligament injury, and with autism spectrum disorder
The first objective of REPAIRS entailed developing fundamental knowledge on perception-action learning. It is generally assumed that variability in practice is helpful in learning a new perception-action skill, hence, this was the focus of several REPAIRS ESRs. To this end, we addressed how learning was affected by presenting training conditions in a random order, as compared to a blocked order. We found positive effects of variable training when learning a novel coordination pattern among joints in the arm in a task where joint angles controlled an avatar in a computer game, and when blind-folded people used a vibrotactile sensory substitution device that helped them grasp an object. The variable practice prompted exploration of coordination patterns and the generation of vibrotactile patterns that made the behaviour more flexible and adaptable.
To study learning in joint actions we developed a ‘doubles pong’ task. In this task two players each move one virtual paddle along a shared interception axis, so that an approaching ball is intercepted by one of the paddles without collision with the other paddle. We found that 1)) catch-ableness of the ball can be perceived for oneself and for another person, 2) this ability does not change over learning, 3) the catching kinematics can be described with a model, 4) verbal communications between people in the dyads does not affect performance, and 5) an overall framework was developed of social facilitation effects at different brain scales.
Other joint action studies focussed on how coordination patterns at the task level and at the level of the joint angles emerged when two participants jointly controlled a plank or a board to roll a ball into a target region. In one series of studies, it was shown that solo performance only moderately transferred to joint action, indicating that the interactions present in joint action added to the performance. Another series of studies demonstrated that coordination patterns in the legs re-organised when participants learned the task.
The second objective of REPAIRS entailed developing understanding on requirements for translating knowledge on perception-action learning to rehabilitation practice. The findings on variability of practice can be used to develop trainings for people suffering from stroke and people with visual impairments. The findings in the doubles pong task provide the basis for developing an artificial agent that can be used to train people from autism spectrum disorder. The board balancing tasks indicate that training people suffering from stroke or from an anterior cruciate ligament injury can benefit from training in joint action settings.
We applied data science techniques to data from rehabilitation clinics and we found that the patients treated for different injuries had different pain levels 6 months after discharge and that the interaction between the injury and the pain levels depended on the type of injury. Also, we focused on improving accuracy and applicability of rehabilitation games through implementing novel algorithms to read out motion capture units often used in these games. Finally, we provided a conceptual framework on differences in behavioural settings that are used to understand difficulties in translating knowledge from one domain to another.
To widely share our findings and insights, we developed toolboxes on theories, methodologies and analyses that are open for public use and can be found on our website, eg. we give background on the Uncontrolled Manifold Method and we give a perspective on knowledge translation and coordination of practices when people of different disciplines work together. Two toolboxes are particularly relevant for translating the knowledge of REPAIRS to the clinical practice. These toolboxes provide interviews with clinicians and frame the outcomes of the interviews in a systems perspective on rehabilitation. These interviews present experiences and obstacles clinicians encounter when translating the knowledge and potential strategies to overcome obstacles of translations to rehabilitation. Both toolboxes emphasise that clinical practice should be early involved in the research, hence co-creation is advised to improve translation to rehabilitation.
The results generated in REPAIRS have been presented by our ESRs at local, national and international scientific meetings. By the end of the project, we published 21 scientific manuscripts. We organised a collective participation at the European Researcher’s Night 2024 where 5 ESRs demonstrated their set ups and presented our findings to the public.
To study learning in joint actions we developed a ‘doubles pong’ task. In this task two players each move one virtual paddle along a shared interception axis, so that an approaching ball is intercepted by one of the paddles without collision with the other paddle. We found that 1)) catch-ableness of the ball can be perceived for oneself and for another person, 2) this ability does not change over learning, 3) the catching kinematics can be described with a model, 4) verbal communications between people in the dyads does not affect performance, and 5) an overall framework was developed of social facilitation effects at different brain scales.
Other joint action studies focussed on how coordination patterns at the task level and at the level of the joint angles emerged when two participants jointly controlled a plank or a board to roll a ball into a target region. In one series of studies, it was shown that solo performance only moderately transferred to joint action, indicating that the interactions present in joint action added to the performance. Another series of studies demonstrated that coordination patterns in the legs re-organised when participants learned the task.
The second objective of REPAIRS entailed developing understanding on requirements for translating knowledge on perception-action learning to rehabilitation practice. The findings on variability of practice can be used to develop trainings for people suffering from stroke and people with visual impairments. The findings in the doubles pong task provide the basis for developing an artificial agent that can be used to train people from autism spectrum disorder. The board balancing tasks indicate that training people suffering from stroke or from an anterior cruciate ligament injury can benefit from training in joint action settings.
We applied data science techniques to data from rehabilitation clinics and we found that the patients treated for different injuries had different pain levels 6 months after discharge and that the interaction between the injury and the pain levels depended on the type of injury. Also, we focused on improving accuracy and applicability of rehabilitation games through implementing novel algorithms to read out motion capture units often used in these games. Finally, we provided a conceptual framework on differences in behavioural settings that are used to understand difficulties in translating knowledge from one domain to another.
To widely share our findings and insights, we developed toolboxes on theories, methodologies and analyses that are open for public use and can be found on our website, eg. we give background on the Uncontrolled Manifold Method and we give a perspective on knowledge translation and coordination of practices when people of different disciplines work together. Two toolboxes are particularly relevant for translating the knowledge of REPAIRS to the clinical practice. These toolboxes provide interviews with clinicians and frame the outcomes of the interviews in a systems perspective on rehabilitation. These interviews present experiences and obstacles clinicians encounter when translating the knowledge and potential strategies to overcome obstacles of translations to rehabilitation. Both toolboxes emphasise that clinical practice should be early involved in the research, hence co-creation is advised to improve translation to rehabilitation.
The results generated in REPAIRS have been presented by our ESRs at local, national and international scientific meetings. By the end of the project, we published 21 scientific manuscripts. We organised a collective participation at the European Researcher’s Night 2024 where 5 ESRs demonstrated their set ups and presented our findings to the public.
The REPAIRS project has advanced both scientific understanding and translational applications in rehabilitation. We identified key challenges faced by visually impaired individuals in hotel environments and proposed actionable guidelines for improving sensory substitution devices, with potential uptake by assistive technology developers. Our research revealed injury-specific pain trajectories post-rehabilitation, informing treatment pathway adjustments now under review by our clinical partner MEDIAN. Novel training games and coordination paradigms were developed, showing promise for stroke recovery, autism therapy, and anterior cruciate ligament injury rehabilitation. Additionally, we created and openly shared advanced calibration algorithms of magnetic based motion capture systems, now integrated into Euleria’s rehabilitation tools, supporting broader innovation in virtual therapy solutions. These outcomes highlight the project's socio-economic and societal impact through improved rehabilitation strategies and inclusive technology design.
REPAIRS provided interdisciplinary and intersectoral training to the ESRs, equipping them with a wide range of scientific and transferable skills preparing them to advance the scientific field of perception-action learning and translating these insights to rehabilitation practice.
REPAIRS provided interdisciplinary and intersectoral training to the ESRs, equipping them with a wide range of scientific and transferable skills preparing them to advance the scientific field of perception-action learning and translating these insights to rehabilitation practice.