Identifying Parkinson's disease and its stages using static standing balance

Area of Science:

• Neurology
• Biomedical Engineering
• Data Science

Background:

• Current Parkinson's disease (PD) diagnosis relies on dynamic motor tasks, posing accessibility challenges.
• Objective and quantitative assessment methods are needed for early PD detection and staging.

Purpose of the Study:

• To develop an accessible method for identifying Parkinson's disease and its stages using static standing balance.
• To leverage machine learning and time-series data analysis for PD assessment.

Main Methods:

• Recruited 210 participants (control and five PD stages).
• Collected 10-second static standing balance data (center of pressure trajectories).
• Extracted features using representation learning and handcrafted methods; trained a Transformer encoder classifier.

Main Results:

• Achieved an F1-score of 0.963 in classifying six groups (control and five PD stages).
• Demonstrated high accuracy in differentiating PD stages based on static balance data.

Conclusions:

• Static standing balance combined with machine learning offers an accessible and accurate approach for PD assessment.
• The novel data mining framework enables early detection and timely intervention for Parkinson's disease.
• This time-series data-driven approach signifies a advancement in digital healthcare for neurological disorders.