Essential Tremor Severity Assessment Using Handwriting Analysis and Machine Learning.

Area of Science:

• Neurology
• Biomedical Engineering
• Data Science

Background:

• Essential tremor (ET) is a common neurological disorder.
• Accurate diagnosis and severity assessment are crucial for effective ET management.

Purpose of the Study:

• To develop and validate a machine learning pipeline for assessing ET severity.
• To utilize the Archimedes spiral test for objective tremor evaluation.
• To establish a robust diagnostic tool for clinical and telemedicine use.

Main Methods:

• A machine learning pipeline combining Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and Support Vector Machines (SVMs).
• Analysis of Archimedean spiral radius data from a family-based dataset spanning all ET severity levels.
• Integration of the Fahn-Tolosa-Marin Tremor Rating Scale (FMT-TRS) for enhanced classification.

Main Results:

• The pipeline effectively distinguishes between tremor presence and severity.
• Robustness confirmed through cross-validation and Gaussian noise perturbation tests.
• Demonstrated potential for non-invasive ET diagnosis and progression tracking.

Conclusions:

• The machine learning pipeline shows significant potential as a non-invasive diagnostic tool for ET.
• The framework combines geometric features, clinical scores, and machine learning for interpretable and clinically meaningful results.
• This approach supports clinical practice and telemedicine applications for essential tremor management.