ACURe
Automatic Classification and Monitoring of Upper-limb Rehabilitation after Stroke
The aim of this project is to precisely and objectively measure patient rehabilitation movements, as well as to use modelling and machine learning techniques to assess the patient's condition and support clinicians in the decision-making process. The project aims to provide access to personalised data on progression and increase patient motivation during rehabilitation of stroke patients.
Motivation
Yearly, more than 15 000 people in Norway experience a stroke, and the number is expected to rise by 50% over the next years because of an aging population. Early treatment, close follow-up and monitoring, and motivating patients’ rehabilitation effort, is closely linked to success in regaining motor functions after stroke. A range of clinical measures (e.g. Fugl-Meyer Assessment (FMA), Action Research Arm Test (ARAT), Wolf Motor Function Test (WMFT)) are available for assessing patients’ level of impairment shortly after stroke, and to classify and monitor patients’ rehabilitation progress.
Currently, therapist’s assessments of patient’s functional movements are subject to each therapist’s subjective assessment and scoring. There is a large potential for more automatic and objective functional assessment of stroke patients both for classification; to decide on the best rehabilitation training strategy, and for monitoring; to closely follow up on progress and adjust training plans and difficulty level for exercises. The state-of-the-art on automatic objective functional assessments of stroke rehabilitation progress shows that efforts using either optical measurement systems with image analysis, or studies using wearable sensors (such as Inertial Measurement Units (IMUs)), are able to automate parts of the functional assessment process, but still cannot achieve sufficient success rate in classification when compared to therapists’ expert assessments.
Aim and scope of project
The primary goal of the project is to develop an automated system for human motion estimation and assessment, specifically tailored for the rehabilitation of stroke patients. This system will integrate data from a range of sensors, employing a data fusion approach and a musculoskeletal model to generate comprehensive movement estimates. The resulting estimates will be utilised to analyse the patient’s motor function and rehabilitation progress through the application of ML algorithms. A simple overview of the proposed system can be seen in fig. 1. The expected outcome of this research is a novel approach for extracting and interpreting motion data which will provide insights to support clinicians in their decision-making processes.
Figure 1: Proposed System Overview - Automatic Human Motion Estimation and Assessment System for Rehabilitation of Stroke Patients