Parkinson’s disease is a neurogenerative disorder and it is difficult to diagnose as no therapies may slow down its progression. This paper contributes a novel analytic system for Parkinson’s Disease Prediction mechanism using Improved Radial Basis Function Neural Network (IRBFNN). Particle swarm optimization (PSO) with Kmeans is used to find the hidden neuron’s centers to improve the accuracy of IRBFNN. The performance of RBFNN is seriously affected by the centers of hidden neurons. Conventionally Kmeans was used to find the centers of hidden neurons. The problem of sensitiveness to the random initial centroid in Kmeans degrades the performance of RBFNN. Thus, a metaheuristic algorithm called PSO integrated with Kmeans alleviates initial random centroid and computes optimal centers for hidden neurons in IRBFNN. The IRBFNN uses Particle swarm optimization Kmeans to find the centers of hidden neurons and the PSO Kmeans was designed to evaluate the fitness measures such as Intracluster distance and Intercluster distance. Experimentation have been performed on three Parkinson’s datasets obtained from the UCI repository. The proposed IRBFNN is compared with other variations of RBFNN, conventional machine learning algorithms and other Parkinson’s Disease prediction algorithms. The proposed IRBFNN achieves an accuracy of 98.73%, 98.47% and 99.03% for three Parkinson’s datasets taken for experimentation. The experimental results show that IRBFNN maximizes the accuracy in predicting Parkinson’s disease with minimum root mean square error.
Parkinson’s Disease (PD) is the second neurogenerative disorder after Alzheimer’s disease. It affects nearly 1% of individuals who are in or above the age of 60 [
Two main quantitative measures that measure the progression of PD is i) the Hoehn and Yahr Scale [
The motivation behind using IRBFNN is to predict PD with maximum accuracy, positive predictive value, negative predictive value, and minimum error. The problem with conventional RBFNN is that performance of the classifier lies in the center of the hidden neurons [
The main contributions of the paper are:
— PSO Kmeans is used to find the centers with the fitness value of maximizing the Intercluster distance and minimizing the intracluster distance
— The centers given by PSO Kmeans are used in the hidden neurons of IRBFNN
— Experimentation of IRBFNN is done on 3 Parkinson’s datasets
IRBFNN is compared with other variations of RBFNN such as RBFNN3 where centers are found using Whale Optimization Algorithm (WOA) Kmeans, RBFNN2 where centers are determined using Sine Cosine Algorithm (SCA) Kmeans, RBFNN1 in which centers are calculated using Genetic Algorithm (GA) based Kmeans and RBFNN where centers are found using Kmeans in terms of accuracy, positive predictive value, negative predictive value, root mean square error, Fscore
— IRBFNN is also compared with conventional machine learning algorithms such as {Kmeans}, Random Forest, Decision Tree, and Support Vector Machine
— Mean, Best, and Worst fitness values are also compared for proposed PSO Kmeans, WOA Kmeans, SCA Kmeans, and GA Kmeans, which is used in IRBFNN, RBFNN3, RBFNN2 and RBFNN1, respectively.
— Also, IRBFNN is compared with other machine learning algorithms used for PD prediction
— Experimentation is carried out 30 times, and the mean value is taken for performance analysis
The rest of the paper is organized as follows: Section 2 describes the related study on applying different machine learning algorithms for predicting Parkinson’s disease. Section 3 details the proposed system for the prediction of Parkinson’s disease. Section 4 details IRBFNN along with centers determined using PSO Kmeans. Section 5 details the experimental results obtained by comparing the proposed Improved radial basis function neural network (IRBFNNRBFNN + PSO + Kmeans) with other machine learning algorithms using the Parkinson’s dataset taken from UCI repository [
Freezing of Gait (FoG) in Parkinson’s disease was predicted using FoG prediction algorithm, which considers various metrics such as sensor positions, sensor axis, sampling window length [
FoG prediction model using AdaBoost was designed using impaired gait features. In order to correctly identify gait, a preFog phase was used based on the slope of the impaired gait pattern [
FoG prediction was made using Electroencephalography (EEG) features, which was determined using Fourier and wavelet analysis using data gathered from 16 patients [
From the literature, it is observed that there are several approaches present for prediction of Parkinson’s disease. Also, there is wide use of particle swarm optimization algorithms to find the number of neurons, their centers and weights of RBFNN and the methods were applied to various realworld problems. With the goal to still maximizing the accuracy, in this paper, PSO with Kmeans is designed to find the optimal centers for RBFNN structure and the proposed approach is used for optimal prediction of Parkinson’s disease.
The system design for the proposed prediction of Parkinson’s disease is shown in
where
The preprocessor does the process of normalizing the data to the range [0, 1]. Normalization of attributes represented in
IRBFNN classifies the input sample by sending each input vector
The numbers of neurons in the input layer are initialized to the number of dimensions of the dataset. Let
where
Each RBF neuron is designed using PSO Kmeans. The RBF neuron prototype plays a prominent role in the optimal allocation of class label to the instance that results in maximizing accuracy. Thus, it is necessary to choose a good prototype for maximizing accuracy. The metaheuristic clustering is used as a RBF neuron prototype where each instance is trained for the optimal class assignment. PSO is evaluated against the metrics such as Intracluster distance and Intercluster distance.
Having computed the hidden neurons’ centers using PSO Kmeans, the next step is computing the variance of each hidden neuron using
The initial weights between the hidden neuron and the summation layer neuron are assigned by the pseudo inverse method represented in
Error for the
where
When the error value is converged, the IRBFNN maximizes the accuracy in the prediction of Parkinson’s disease for the test dataset.
The radial basis function is designed using particle swarm optimizationbased Kmeans. Algorithm 1 illustrates the working procedure of IRBFNN. Section 4 details the computation of the radial basis function for IRBFNN. The combined fitness function for the particle swarm optimization is represented in
where
Intercluster distance
where
The PSO Kmeans for finding optimal centers of the hidden neuron are represented in Algorithm 2. The
The proposed IRBFNN was executed in python and its accomplishment was measured using three Parkinson’s disease datasets taken from the UCI repository [
In order to evaluate the efficiency of the proposed IRBFNN, several investigations were performed. The analysis was conducted on 3 benchmarking Parkinson’s datasets taken from the UCI repository. The datasets include Dataset 1Parkinson’s dataset, Dataset 2Parkinson’s disease classification dataset, Dataset 3Parkinson’s speech dataset with multiple types of sound recordings data set. Researchers widely used these datasets for classifying the Parkinson’s disease.
A metaheuristic algorithm PSO integrated with Kmeans with the defined fitness represented in
Dataset  Dataset 1  Dataset 2  Dataset 3 

#instances  197  756  1040 
#features  23  754  26 
#classes  2  2  2 
Initialize input neurons 

Initialize hidden neurons 

Initialize summation neurons 

Trained model IRBFNN  




Compute variance 







Compute 





































compute error 









Compute change in weight using 

Compute weight 






Dataset 

Cluster centres 



Initialize particle 

Initialize Velocity 

Initialize Personal Best 

Initialize Global Best 

Initialize Personal Best Position 

Initialize GBest Position 



















Compute fitness 

















Compute Velocity 

Compute Position 





return 
Maximum number of iterations is set as 100 for all Radial Basis Function except KMeans where the maximum iteration is set as 500. The number of neurons in the input layer is set as number of features in the respective datasets. The number of neurons in the input layer is 23, 754 and 26 for dataset 1, dataset 2 and dataset 3 respectively. The number of neurons in hidden layer is 6 for dataset 1 and dataset 3 and 29 for dataset 2 respectively. The parameter settings for the variations of radial basis functions and the neural network are listed in
Algorithms  Population size  Dimension  Values for other parameters  

For training neural network in RBFNN  #Instances  –  
RBFNN 
#Instances  
RBFNN1 
20  
RBFNN2 
20  
RBFNN3 
20  
IRBFNN 
20 
The results acquired for IRBFNN are elaborated in this section. The proposed IRBFNN is compared to assess the outcome of using PSO Kmeans as radial basis function instead of using Kmeans in RBFNN, GA Kmeans in RBFNN1, SCA Kmeans in RBFNN2 and WOA Kmeans in RBFNN3. The metrics used to evaluate the proposed mechanism includes:
Experimentation results carried out to measure the Fscore of IRBFNN, and other variants of RBFNN are shown in
Next experiment is carried out to measure the positive predictive value, which is shown in
Dataset \method  RBFNN  RBFNN1  RBFNN2  RBFNN3  IRBFNN 

Dataset 1  0.4292  0.8275  0.8466  0.9069  
Dataset 2  0.6335  0.5260  0.7327  0.8349  
Dataset 3  0.6380  0.7499  0.7995  0.8598 
Dataset \method  RBFNN  RBFNN1  RBFNN2  RBFNN3  IRBFNN 

Dataset 1  0.7843  0.8056  0.8140  0.8571  
Dataset 2  0.2777  0.3589  0.4634  0.6512  
Dataset 3  0.8725  0.8981  0.9797  0.9896 
Dataset \method  RBFNN  RBFNN1  RBFNN2  RBFNN3  IRBFNN 

Dataset 1  0.8812  0.9034  0.9259  0.9523  
Dataset 2  0.7719  0.7741  0.7967  0.8086  
Dataset 3  0.8522  0.8589  0.9818  0.9898 
Dataset \method  RBFNN  RBFNN1  RBFNN2  RBFNN3  IRBFNN 

Dataset 1  0.8272  0.8460  0.8552  0.9053  
Dataset 2  0.6144  0.5254  0.7450  0.8227  
Dataset 3  0.7479  0.7519  0.8431  0.8754 
The fitness value of various algorithms for all datasets is represented in
The accuracy of IRBFNN is compared with other existing Parkinson’s Prediction Algorithm. For dataset 1, the accuracy was measured as: neural network [
The computational complexity of the proposed IRBFNN is measured using BigOh
Operations  Cost 

Computing initial centers using PSO Kmeans  
Computing variance  
Computing initial weight  
Computing gaussian radial basis function  
Convergence of error rate  
Computing error  
Change in weight  
Total computational cost 
Dataset  RBFNN  RBFNN1  RBFNN2  RBFNN3  IRBFNN 

Dataset 1  32.13  24.67  19.10  12.31  9.30 
Dataset 2  58.95  54.60  52.87  52.07  51.84 
Dataset 3  57.35  52.30  44.63  40.20  39.23 
From
Dataset  Kmeans (K = 2)  Decision Tree  Random Forest  SVM  IRBFNN 

Dataset 1  0.8010  0.8860  0.8430  0.8957  
Dataset 2  0.8451  0.8135  0.8532  0.8928  
Dataset 3  0.8373  0.8570  0.8730  0.8212 
Dataset  Kmeans (K = 2)  Decision Tree  Random Forest  SVM  IRBFNN 

Dataset 1  14.13  0.10  0.40  0.23  9.30 
Dataset 2  19.38  14.1  22.37  15.53  51.84 
Dataset 3  14.52  7.49  18.09  17.90  39.23 
The inferences made from the experiment results were listed as:
Improved radial basis function neural network maximizes accuracy together with minimizing root mean square error
The use of PSO Kmeans with the fitness of maximizing Intercluster distance and minimizing intracluster distance finds optimal cluster centers, which is used in hidden neurons of IRBFNN
Experiments performed to measure the positive predictive value, and negative predictive value also signifies the introduction of PSO Kmeans radial basis function improves the performance in identifying the positive and negative instances
The execution time of the proposed IRBFNN is higher than conventional machine learning algorithms but with the increase in accuracy
The introduction of PSO Kmeans improves the accuracy of IRBFNN by 3.83%, 14.85%, 19.57% and 25.15% than RBFNN3, RBFNN2, RBFNN1 and RBFNN, respectively.
Finally, through rigorous analysis, it has been inferred that the IRBFNN was designed and experimented successfully to predict Parkinson’s disease. Besides, the proposed network reveals that finding the efficient radial basis function is essential for accurate prediction. The presented IRBFNN best solves the given problem of predicting Parkinson’s disease by efficiently finding the centers of hidden neurons for designing the radial basis function of IRBFNN. Thus, to obtain the good performance, metaheuristic algorithms are used to find optimal values of these parameters, leading to minimizing error and maximizing accuracy. PSO Kmeans’ performance is compared with other metaheuristic way of finding centers in designing radial basis function neural network and the proposed IRBFNN shows that PSO Kmeans choose the optimal center by doing good level of exploration and exploitation by avoiding struck in local optima when predicting Parkinson’s disease. The key findings of the paper are listed as:
The problem of finding the centers of the hidden neurons is solved by using PSO with Kmeans, which maximizes the accuracy of the presented IRBFN
The integration of PSO with Kmeans diminishes the problems caused by the initial random centroid of conventional Kmeans by doing a good level of exploration and exploitation
The fitness value of PSO takes Intracluster distance, Intercluster distance, which produces optimal cluster centers
The use of PSO Kmeans in finding the hidden neurons’ centers maximize the accuracy, Fscore, positive predictive value, negative predictive value, recall and minimizes the root mean square error. In future work, a novel feature selector algorithm will be integrated before the analytics process for further enhancing the accuracy of prediction.