Computational Methods for Predicting Chronic Disease in Healthcare Communities

摘要

A computational model designed on the basis of big data analytics has a vast application in medical field. The chronic disease like cerebral stroke results in demises of large number of human lives in an unpredicted way. Hence it is very important to have a prediction model to reduce the impact of such issues. The accuracy of the prediction model is purely based on the potential to extract the unique valid features from the dataset used in prediction. In this analytical study, we collected both structured and unstructured data from National Stroke Mortality dataset. Ten-fold cross validation was performed to both training and test sets. We proposed three classifiers such as Naive Bayes, K-Nearest Neighbor and Decision tree to predict the risk of stroke. As it is a life saving problem, the outcome of the model was evaluated on the basis of various performance and performance error measures. In order to prove the accuracy of the models, we made comparison with existing works and reached to a result that the decision tree shows a better performance than other models. With the selection of real life dataset, we could achieve an accuracy of 99% and it is experimentally proved that decision tree is best for the prediction of risk in cerebral stroke.