Periodic Reporting for period 1 - Rehab-Assessment (Personalised Quantitative Upper Extremity Assessment for Stroke Rehabilitation)
Okres sprawozdawczy: 2022-01-01 do 2023-12-31
With the number of people surviving a stroke soaring, more and more rehabilitation programs are delivered with minimal involvement of a physiotherapist due to their limited availability, and the success of this approach depends on the accurate assessment of stroke patients’ movement impairment. The overall objective of this project is to establish a comprehensive, quantitative, objective and personalised Motor Impairment Index, via benchmarking the impaired arm movement to healthy arm mirrored exercises and evaluating the differences. This approach will help to reduce the long-term demand on overstretched healthcare resources and increase the likelihood of successful rehabilitation leading to improved quality of life for millions of people affected by stroke.
There are three objectives of the project:
O1: Personalised neuro-musculoskeletal modelling: A novel upper limb neuro-musculoskeletal mathematical model will be established to predict individual muscle contribution during movement. A hybrid calibration procedure will be developed to estimate personalised model parameters (i.e. model optimal muscle fibre length, tendon slack length, muscle fibre contraction velocity, muscle damping, and pennation angle, etc.).
O2: Personalised quantitative upper extremity assessment: A quantitative, personalised and objective MII will be developed via dissimilarity analysis to quantify the movement differences between the impaired arm and the healthy arm during mirrored exercises and obtain the objective upper extremity impairment assessment.
O3: Two-stage feasibility study: 10 asymptomatic volunteers (5 males and 5 females) and 20 stroke survivors (individuals with any extent of arm weakness, 10 females and 10 males) will be recruited to evaluate the effectiveness of the developed MII, with the support of National Demonstration Centre for Rehabilitation at Leeds Teaching Hospital NHS Trust.
To achieve O1, a series of physics-informed data-driven musculoskeletal models were proposed for personalised neuro-musculoskeletal modelling. The feasibility and effectiveness of the proposed approaches were verified on 10 healthy volunteers. Different from existing physics-based and data-driven modelling approaches, the proposed approaches can accurately reflect the underlying physical neuro-mechanical processes and have fast execution speed. To achieve O2, a quantitative, personalised, and objective motor impairment index was established by evaluating the difference between the healthy arm and the impaired arm during mirrored exercises to quantify the motor impairment. The effectiveness of the established approach was verified with regard to the conventional clinical approaches, such as FM-UE and ARAT scores. To achieve O3, 10 asymptomatic volunteers were first recruited to evaluate the system at the Leeds’ Rehabilitation Robotics and Sensing Lab. 20 volunteers with stroke were also recruited to further evaluate the system at the lab.
O1: Personalised neuro-musculoskeletal modelling: A novel upper limb neuro-musculoskeletal mathematical model will be established to predict individual muscle contribution during movement. A hybrid calibration procedure will be developed to estimate personalised model parameters (i.e. model optimal muscle fibre length, tendon slack length, muscle fibre contraction velocity, muscle damping, and pennation angle, etc.).
O2: Personalised quantitative upper extremity assessment: A quantitative, personalised and objective MII will be developed via dissimilarity analysis to quantify the movement differences between the impaired arm and the healthy arm during mirrored exercises and obtain the objective upper extremity impairment assessment.
O3: Two-stage feasibility study: 10 asymptomatic volunteers (5 males and 5 females) and 20 stroke survivors (individuals with any extent of arm weakness, 10 females and 10 males) will be recruited to evaluate the effectiveness of the developed MII, with the support of National Demonstration Centre for Rehabilitation at Leeds Teaching Hospital NHS Trust.
To achieve O1, a series of physics-informed data-driven musculoskeletal models were proposed for personalised neuro-musculoskeletal modelling. The feasibility and effectiveness of the proposed approaches were verified on 10 healthy volunteers. Different from existing physics-based and data-driven modelling approaches, the proposed approaches can accurately reflect the underlying physical neuro-mechanical processes and have fast execution speed. To achieve O2, a quantitative, personalised, and objective motor impairment index was established by evaluating the difference between the healthy arm and the impaired arm during mirrored exercises to quantify the motor impairment. The effectiveness of the established approach was verified with regard to the conventional clinical approaches, such as FM-UE and ARAT scores. To achieve O3, 10 asymptomatic volunteers were first recruited to evaluate the system at the Leeds’ Rehabilitation Robotics and Sensing Lab. 20 volunteers with stroke were also recruited to further evaluate the system at the lab.
The research works successfully integrate the physics domain knowledge and data-driven modelling to enhance the personalised musculoskeletal modelling, which are the first attempt to bring the physics information into the data-driven models for musculoskeletal modelling enhancement. These pioneered works unlock the potential of physics-informed neural network in personalised musculoskeletal modelling, and open a new direction for the musculoskeletal modelling using AI, more research can continuously conduct based on these proposed approaches. The organized special issue about the wearable sensing and computing attracted many researchers from the world to submit their papers, which significantly improves the effects of our research. Through the fellowship, Dr. Jie Zhang has built a stronger academic profile and prestige through publications in high-quality journals and talks in prestigious conferences, and the fellowship also builds the foundation for future collaboration with researchers from Europe and North America.