A Novel Algorithm For Real-time Detection Of Freezing Of Gait Using Wrist-based Accelerometer Sensor Data.

Area of Science:

• Neurology
• Biomedical Engineering
• Signal Processing

Background:

• Aging populations are increasing due to medical advancements, leading to a rise in age-related neurological disorders.
• Parkinson's disease (PD) is a significant neurological disorder, with Freezing of Gait (FoG) being a particularly debilitating symptom.
• Current methods for FoG detection may lack real-time applicability or user convenience.

Purpose of the Study:

• To develop a novel, subject-independent algorithm for real-time Freezing of Gait (FoG) detection in Parkinson's disease (PD).
• To utilize data from wrist-worn accelerometer sensors for enhanced feasibility and usability.
• To improve the noise resilience of motion data using Variable Mode Decomposition (VMD).

Main Methods:

• A novel algorithm was developed for real-time FoG detection.
• Variable Mode Decomposition (VMD) was integrated to enhance noise resilience in accelerometer data.
• The algorithm was designed as a subject-independent model using wrist-worn sensors.

Main Results:

• The proposed algorithm demonstrated significant potential in detecting FoG.
• Evaluation on a public dataset yielded an average Sensitivity of 0.80.
• The model achieved an average Specificity of 0.81, indicating effective performance.

Conclusions:

• The developed algorithm shows promise for effective and user-friendly real-time FoG detection in PD.
• Wrist-worn sensors offer a practical solution for monitoring FoG symptoms.
• The subject-independent nature and noise resilience enhance the clinical applicability of this approach.