Abstract
Over 25% of people with Multiple Sclerosis (MS) experience tremor, which considerably affects quality of life. There is currently no objective measurement for tremor characteristics. This thesis aimed to develop and validate an objective measurement of tremor in people with MS using a wrist worn accelerometer (AX3).Firstly, a tremor detection and characterisation algorithm based on frequency analysis using Fast Fourier analysis was developed in Matlab. Its performance was tested within the laboratory against the current clinical gold standard of tremor assessment and also motion capture (N=12). Secondly, the algorithm was optimised for use on 24-hour, everyday living records of people with MS (N=12), using results of a non-pathological population (N=25) to refine outcomes. The results of the records of the non-pathological population were compared to people with definite MS tremor. Lastly, the responsiveness of the measure in regards to changes in tremor characteristics due to an intervention was investigated (N=4).
The algorithm performed well against the current clinical gold standard, correctly predicting over 97% of cases as tremor or non-tremor in the MS population (N=12) as recorded in the laboratory. The correlation of calculated movement amplitude [cm] to the clinician’s tremor severity scores was only moderate. However, the correlation to objectively measured tremor amplitude of the motion capture system was very strong. It was not possible to eliminate all false positives from the everyday living records, activities such as tooth brushing for example of the normative population. In the MS tremor population, identified tremor correlated with self-reported tremor but there were no clear differences in signal characteristics between the two groups. Only the overall activity was remarkably lower in those with MS. Finally, the objective measurement
responded to changes in tremor characteristics which were reflected in the participants own perception, but not necessarily in the clinicians severity scores.
It can be concluded that the developed objective measure identified and characterised tremor accurately in a tremor assessment setting. However, further testing is required for everyday living assessments as the number of participants accessed was too low and too many false positive outcomes were identified in the control population. A further study with matched controls would be advisable. Ultimately, the outcomes of the study on the responsiveness of the system highlighted the importance of an objective measure; it identified changes in tremor characteristics which made a difference to the affected person, without being reflected on the clinical rating scale.
| Date of Award | 2016 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Jennifer Preston (Supervisor) & Frederike van Wijck (Supervisor) |