Performance Assessment of Health Decision-Making Using Various Artificial Intelligence Techniques and Evolutionary Algorithms

Abstract

Presently, the latest technology is used for health management and diagnostic strategy in the well-being area. Artificial intelligence usually helps in medical problems utilizing various models, machine learning typically helps to make decisions about health problems. A plethora of help is provided in the prediction of diseases. Different machine learning techniques are available to classify and identify diseases but what comes first is the optimization of the techniques. Nature-based optimization techniques are developed, which works on the way things interact with each other in nature. In this paper, particle swarm optimization (PSO) is used to optimize the dataset on diseases such as cardiovascular disease, liver disease and cancer. Along with PSO, principal component analysis (PCA) is used to optimize as well as decrease the computational complexity of the model. The key aim is to find the most important attributes contributing to disease such that early diagnosis and hence treatment of the disease can be started. The proposed hybrid model is trained using the data sets for different diseases with different classification algorithms viz random forest, support vector machine (SVM) and K-nearest neighbour (KNN). Experimental results show that all the irrelevant features that were not necessary were removed by particle swarm optimization and principal component analysis. The highest classification accuracy of 83.52% was obtained for Heart Disease dataset with random forest classifier. The highest classification accuracy of 81.32% was obtained for Liver Disease dataset with Random Forest classifier. Highest classification accuracy of 100% was obtained for Cancer dataset with random forest classifier as well as support vector machine classifier.