Introduction

Background

Parkinson’s disease (PD) is a chronicle neurodegenerative disorder that results from a progressive loss of dopaminergic and other sub-cortical neurons [1]. Clinical manifestation of PD involves a loss of mobility, increased rigidity, tremor, and postural instability, which lead to a common symptom called Freezing of Gaits (FOG) [2]. FOG is characterized by a reduction in step length and velocity, decreased angular displacement and velocity of lower and upper limbs, high variability of step timing, poor bilateral coordination and asymmetric leg function [3]. FOG is a common cause of falls in PD, and severely impairs the quality of life.

However, the management of FOG is ineffective, and current assessment of FOG mainly bases on subjective reports by physicians, which is problematic as FOG symptoms are not always evident in clinical settings. A continuous monitor of FOG events will be beneficial to PD patients because real-time detection and prediction of FOG will help patients be aware of freezing episodes and avoid possible falls.

Project Scope

  1. Detection of FOG

The first step in detecting FOG is to determine the motion information of the inertial measurement unit (IMU) mounted on the patient’s shoe. Such information can be used to calculate the center of mass displacement, the center of pressure, the ground reaction force of the patient, and other parameters that are important in assessing the stability of gaits [4].

  1. Classification of FOG

FOG can be mainly classified into three categories [5]: a) Initiation freeze: When this occurs, the patient is incapable to start walking; b) Turning freeze: When this occurs, the patient is not able to make a sharp turn; c) Freeze: This type of freezing consists of shuffling forward with very short steps. Automatic classification of these three types of FOG from IMU data is beneficial to patients as they will be given specific alarms and instructions during the onset of a specific type of FOG.

  1. Prediction of FOG

Based on the progress of FOG detection and classification, possibilities of FOG prediction will be studied. Prediction of FOG onset will be a preemptive cue that can help patients avoid freeze and possible falls. Current prediction methods focuses on electrocardiography and skin-conductance [6] measurements that are expensive and not accommodative to everyday settings. Based on results from FOG detection and classification, we plan to use motion information from IMU data before the onset of each FOG event to predict occurrences of FOG.