Detection Of Parkinson’s Disease Based On The Integration Of Audio Features And RFE Algorithm

In recent times, the scientific community has exhibited a notable upsurge in curiosity concerning the potential application of acoustic signals as biomarkers in the detection of Parkinson’s disease. Approximately 90% of individuals diagnosed with Parkinson’s disease demonstrate symptoms of vocal impairment in the early phases of the condition, according to research. In this article, a novel approach is presented for analyzing the effect of a dataset that includes both short-term and long term features, MFCC (melt frequency cepstral coefficients), on the efficacy of a random forest model designed to detect Parkinson’s disease. The research employs Recursive Feature Elimination (RFE) to produce an optimal subset of critical features, thereby enhancing the model’s performance and mitigating overfitting that may occur during training. The results indicate that the integration of MFCC with long-term features led to a Random Forest Classifier accuracy of 89.94%.