An Efficient Disease Prediction in Big Data using Neuralnetwork based Optimization Method

Abstract

Generally, with Big Data concept large complex data are handled easily and in healthcare it helps to access data rapidly. Previously, Big Data in healthcare reached a complexity level but sometimes it is impossible to obtain the required data for use; thereby the growth in healthcare sector is slow. Big Data is much interesting when used to analyze healthcare data. Classifiers with cost-sensitive factors increases the stability of classification and reduces its computational costs when dealing with large scale, imbalanced and redundantdatasets like medical data. Moreover, the nature and growth of disease are unknown; thus, making prediction complex. This work predicts various diseases where the parameters of neural networkare optimized by using big data which includes data from social media. The efficiency of the TRAPezoidal Neural Network (TRAP-NN) and Improved Whale Optimization Algorithm (WOA) learning models named as (IWOA-TRAP-NN) were compared with the two standard methods such as Genetic Algorithm optimized Convolutional Neural Network (GA-CNN) and Ant ColonyOptimized Convolutional Neural Network (ACO-CNN). The proposed TRAP-NN achieves 86% accuracy, 74%of F1 score and 56.2% of kappa static. The results show that the TRAP-NN performs better than GA-CNN and ACO-CNN.