Parkinson’s disease (PD) is a neurodegenerative condition that affects the normal motor function
of the brain. Automated sensor systems are unique sensor modalities that have been combined
together for a specific function. In this thesis, an automated sensor system that tracks
and records PD motor symptoms in a clinical setting is proposed. The motor symptoms that
were targeted for quantification were rigidity, bradykinesia and tremor. The automated sensor
system produced here is called the Parkinson’s diagnostic device (PDD) and comprised of
a force sensor, motion sensors and muscle activity sensors to measure PD motor symptoms.
All accompanying software interfacing libraries and graphical user interfaces developed were
described in this thesis as well. The Unified Parkinson’s disease Rating Scale (UPDRS) is currently
the gold standard used by clinicians to measure and rate the severity of motor symptoms.
As such, during subject testing, the UPDRS score given by the clinician was used as a target to
track symptom severity using the PDD. Using supervised learning models, the PDD was able to
accurately predict the UPDRS scores on a cohort of 18 subjects with PD with 96.3 % accuracy
for elbow rigidity, 100 % accuracy for wrist rigidity, 94.1 % accuracy for bradykinesia, 93.9
% accuracy for kinetic tremor, 90.9 % accuracy for postural tremor and 90.5 % accuracy for
rest tremor. The supervised learning models developed were utilised further during deep brain
stimulation (DBS) programming sessions. During DBS programming sessions, clinicians can
struggle to track which DBS setting minimised the symptoms best. In addition, the UPDRS
scale is too low resolution to track any incremental changes from DBS settings. An estimated
UPDRS probability model and visualisation was developed to portray the best DBS settings for
each subject during a DBS programming session.