A new approach for quantitative assessment of upper-extremity impairment in post-stroke patients using wearable sensors: a pilot study.

Stroke is the second leading cause of death in Europe and the leading cause of long therm disability. With the decrease in deaths rate, the number of stroke survivors is growing up and a good post-stroke rehabilitation plan is fundamental for regaining autonomy. Before starting rehabilitation, it is necessary to assess motor function impairment and this is done using clinical scales. In particular Upper-Extremity Fugl-Meyer Assessment is the most used for stroke patients in case of upper limb motor deficits. However, this scale is clinician-dependent, the evaluation is qualitative, gives only partial information and has only poor sensitivity to mild impairment. For those reasons, in this thesis a new approach for quantitative assessment of upper-extremity impairment in post stroke-patients is proposed. XSens Awinda and Myo armband systems were used for testing a total of 11 tasks from the Upper Extremity Fugl-Meyer Assessment scale. Two experiments were performed, the first one involving 9 healthy subjects and the second on one stroke patient. A complete analysis was performed on kinematic data (XSens used on all the 10 subjects) and several parameters were extracted: some of them were deriving from a single segment, while Correlation Coefficient estimation and PCA were done using data from several joints. An exploratory analysis was executed on EMG data, since recording with Myo armband was performed on only 3 healthy subjects. Also in this case, several metrics were chosen, including RMS value, Correlation Coefficient and PCA. Results obtained suggest that the selected parameters can lead to a more complete description of impairment, since they are able to capture differences between stroke and healthy subjects. Moreover the evaluation obtained is quantitative, clinician-independent and more sensitive to mild impairment. It is a method which allows to consider the contribute of the entire upper body section, in order to identify possible compensatory strategies or abnormal synergies. There are some limitations regarding the Myo armband, which only records data from forearm segment, the offline analysis and the fact that, for now, the system was tested on a single patient. However, all those limits can be easily overcome in line with the future perspectives.