An Improved KNN Algorithm to Identify Parkinson's Disease on Voice Ailment

In recent times, researchers have extended their research on Parkinson's Disease (PD) using Artificial Neural Network techniques. PD is an age related progressive neurological ailment that affects millions of older people worldwide. There is a need to identify PD accurately as early as possible. Identifying PD at an early stage is a challenging procedure. This paper aims to use machine learning models such as Artificial Neural Networks (ANN) and an enhanced k-Nearest Neighbour (kNN) techniques to identify PD at an early stage. To detect the PD, an audio data of older people is used in this study. An automatic process classifies an audio data into two categories such as PD patient and healthful individual. The proposed model uses the weights and the nearest neighbourhood concepts to predict the audio samples. The results of the improved kNN classifier produces optimal results by classifying the audio datasets into healthy and unhealthy categories.