Predicting Recovery of Patients after Stroke

9daccab7 5d5f 47cd a902 866ccd182c59.png 500 500

Organization: Advanced Interactive Technologies

Involved Host(s): Kaufmann Manuel

Abstract: The aim of this project is to develop a model that predicts how well after-stroke patients recover their upper limb motor functions 3 months after stroke onset.

Description: After stroke onset, patients suffer from impaired upper limb motor functions, such as loss of dexterity of the hand and reduced shoulder abduction (i.e. lifting the arm). Some patients can recover from these deficits. To tailor rehabilitation plans to individual patients and to reduce costs, it is thus desirable to predict how well a patient will change their motor functions. Previous studies have shown that it is indeed possible to predict upper limb recovery, which led to the development of predictive models like SAFE and PREP2. The Department of Neurology of the University Hospital Zurich is currently running a clinical observational study, aRISE, aimed towards predicting the outcome of upper limb capacity 3 months after the stroke based on measurements taken very early after stroke onset (i.e. within less than 48 hours). To do so, measurements of 40 patients were taken at 3 different times during their recovery process using a Myo armband, which measures EMG signals as well as inertial data.

Goal: We now looking for a master student to analyze the collected data and develop a model in the style of SAFE or PREP2 with increased predictive performance. Depending on the student’s interests the goal of the thesis can be adjusted. For example, a tool that allows the clinician to visualize and manage the recorded data is also of potential interest.

Contact Details: Dr. Jeremia Held (USZ) Manuel Kaufmann (ETH)

EMG IMU Machine Learning Healthcare

Labels: Master Thesis CLS Student Project (MPG ETH CLS) ETH Organization's Labels (ETHZ)
Topics: Medical and Health Sciences Information, Computing and Communication Sciences