Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
MOVEMENTS CLASSIFICATION BASED ON PRESSURE-SENSING MATTRESS DATA
This Master Thesis aims to develop an algorithm to classify different movements that could occur during sleep. Data is readily available from previous recordings but can be extended using additional measurements.
One third of human life is spent sleeping. Sleep movement disorders like restless legs syndrome (RLS), periodic limb movement disorder (PLMD), REM sleep behaviour disorder (RBD) as well as toss and turning events are shown to be prodromal markers for neurodegeneration like Parkinson's disease. This emphasizes the importance of monitoring movements during sleep with reliable tools. The state-of-the-art to monitor sleep movement disorders is video polysomnography. However, this cumbersome device requires cables attached to the users' bodies and often interferes with natural sleep. A pressure-sensor textile placed on the mattress has the potential to detect not only the lying position but also nocturnal movements and classify them based on known patterns.
One third of human life is spent sleeping. Sleep movement disorders like restless legs syndrome (RLS), periodic limb movement disorder (PLMD), REM sleep behaviour disorder (RBD) as well as toss and turning events are shown to be prodromal markers for neurodegeneration like Parkinson's disease. This emphasizes the importance of monitoring movements during sleep with reliable tools. The state-of-the-art to monitor sleep movement disorders is video polysomnography. However, this cumbersome device requires cables attached to the users' bodies and often interferes with natural sleep. A pressure-sensor textile placed on the mattress has the potential to detect not only the lying position but also nocturnal movements and classify them based on known patterns.
This Master Thesis aims to develop an algorithm to classify different movements that could occur during sleep. The following changes of lying positions should be distinguished: left to supine, supine to the right, right to prone, prone to left, left to prone, prone to right, right to supine, and supine to the left. Moreover, periodic upper and lower limb movements and aggressive moments like RBD should be detected. Data is readily available from previous recordings but can be extended using additional measurements.
Work packages:
• Literature Research: Compare existing algorithms for movements classifications during sleep
• WP1 Pilot Data Analysis: Visualization and evaluation of data recorded in previous work with pressure-sensing mattrasses
• WP2 Algorithm implementation: Design of an algorithm for movements-in-bed classification
• WP3 Additional data collection and classification: Collect additional data of movements in bed from healthy volunteers and report the algorithm's performances
• WP4 Validation: Validate the movement classification algorithm with an in-hospital patients dataset.
This Master Thesis aims to develop an algorithm to classify different movements that could occur during sleep. The following changes of lying positions should be distinguished: left to supine, supine to the right, right to prone, prone to left, left to prone, prone to right, right to supine, and supine to the left. Moreover, periodic upper and lower limb movements and aggressive moments like RBD should be detected. Data is readily available from previous recordings but can be extended using additional measurements.
Work packages:
• Literature Research: Compare existing algorithms for movements classifications during sleep
• WP1 Pilot Data Analysis: Visualization and evaluation of data recorded in previous work with pressure-sensing mattrasses
• WP2 Algorithm implementation: Design of an algorithm for movements-in-bed classification
• WP3 Additional data collection and classification: Collect additional data of movements in bed from healthy volunteers and report the algorithm's performances
• WP4 Validation: Validate the movement classification algorithm with an in-hospital patients dataset.
Not specified
You should be familiar with sensors data analysis and machine learning tasks or highly motivated to learn it
You should be familiar with sensors data analysis and machine learning tasks or highly motivated to learn it
Please, send your application (cv and recent transcript) to both supervisors: alexander.breuss@hest.ethz.ch and oriella.gnarra@hest.ethz.ch
Please, send your application (cv and recent transcript) to both supervisors: alexander.breuss@hest.ethz.ch and oriella.gnarra@hest.ethz.ch