Prediction of freezing of gait using analysis of brain effective connectivity

Freezing of gait (FOG) is a debilitating symptom of Parkinson's disease (PD), in which patients experience sudden difficulties in starting or continuing locomotion. It is described by patients as the sensation that their feet are suddenly glued to the ground. This, disturbs their balance, and hence often leads to falls. In this study, directed transfer function (DTF) and partial directed coherence (PDC) were used to calculate the effective connectivity of neural networks, as the input features for systems that can detect FOG based on a Multilayer Perceptron Neural Network, as well as means for assessing the causal relationships in neurophysiological neural networks during FOG episodes. The sensitivity, specificity and accuracy obtained in subject dependent analysis were 82%, 77%, and 78%, respectively. This is a significant improvement compared to previously used methods for detecting FOG, bringing this detection system one step closer to a final version that can be used by the patients to improve their symptoms.