This research evaluates the application of Echo State Networks (ESN) for predicting Parkinson’s disease using voice features. The study aims to develop a diagnostic model that ensures high accuracy and minimal false negatives. A feature selection strategy using ANOVA was adopted to identify the most informative features. Various machine learning methods were tested, and ESN showed superior performance, with less than 8% false negatives in 83% of cases. This positions ESN as a suitable choice for Parkinson’s disease diagnosis, particularly with limited data.
Publication date: 31 Jan 2024
Project Page: Not provided
Paper: https://arxiv.org/pdf/2401.15672
