Assessing rehabilitative reach and grasp movements with Singular Spectrum Analysis

An objective and accurate clinical assessment of a patient's movement disabilities are needed to provide the type and intensity of therapy for them. Current assessments in use require therapists to subjectively grade various actions performed by patients, but this is an onerous and error prone process. A common solution is to use the sensors built into consumer devices to obtain patient metrics. However we motivate for our approach of embedding sensors into objects used in established assessments which provide patient metrics in a ratified way. A significant contribution to this approach is using force sensors to capture subtle but important movements which are not easily obtained otherwise and to provide movement cues. Standard frequency domain analyses can lead to misdiagnosis which motivates for data driven approaches. like Singular Spectrum Analysis. This is used in a novel way to identify the most informative eigentriple to assess intraclass movements which are measures of the quality of a patient's upper limb movements. This is done on a set of healthy subjects who also simulated movement disorders and on some actual patients undergoing rehabilitation.