| dc.description.abstract | Abstract: Freezing of Gait (FoG) is a common
incapacitating complication in Parkinson’s
patients, which will temporarily hinder the
forward progression and will prevent them from
re-initiating their normal gait. This can lead to
potentially fatal falls and severely affect the
quality of life of the patient. Due to
characteristic changes in their gait, FoG can be
identified by using wearable sensors such as
pressure sensors, Inertial Measurement Units
(IMU), and Electroencephalogram (EEG)
electrodes. Classification models that run on
machine learning algorithms have been
frequently used. Prediction of FoG would be
highly useful for the patients since this identifies
the changes in their gait preceding the event
and the patient can be notified. This will allow
them to overcome FoG. This systematic review
identifies the best sensors, sensor placements,
predictive algorithms, and the limitations of the
existing prediction systems. Out of all the
methods reviewed, combinations of plantar
pressure sensors placed on the insoles and IMUs
placed on the shank produced the highest
accuracies with a specificity of 91.6%. The best
algorithm was identified as Convolutional
Neural Networks. | en_US |
