Long-term activity monitoring and movement classification in stroke rehabilitation

Stroke patients receive intensive therapy in rehabilitation centers. Once they are at home, no objective information of their activity is available. Especially the tasks they perform with the affected upper limbs are mainly unknown. There is no good mean to compare the patient’s behavior in all-day life before and after intensive therapy sessions.

In this project we develop small sensor modules and algorithms to record and classify the patient’s activities over long periods. To guarantee an operating time of several weeks required for this project, we develop our proper hardware. The processor and sensors can be selected suiting exactly our application and total hardware access enables for controlling the power consumption of the components during operation.
These modules will later be used to study upper limb movements over longer periods. Algorithms to classify these movements have to be researched.

A small low-power sensor module has been developed and is currently evaluated. First tests have shown promising results regarding low power consumption and precision of the measurements.

The hardware and algorithms developed within this project can be powerful tools for assessment and show the doctors how therapies transfer to each subject.

Leuenberger, K., and Gassert, R. (2011). Low-power sensor module for long-term activity monitoring. Proc. IEEE Engineering in Medicine and Biology Conference (EMBC), pp. 2237–2241.

Contacts

Kaspar Leuenberger
ETH Zurich
Rehabilitation Engineering Lab
ML G 53.1
8092 Zurich
lkaspar@ethz.ch