Forecasting infectious disease outbreak using support vector regression (SVR) case study: measles (rubeola)

Abstract

Disease outbreak forecasting, provides warning that a certain amount of disease may occur at a particular time in the future. This research work uses measles, which is a highly contagious disease caused by the measles virus Morbillivirus as a case study. It has been problematic detecting the outbreaks of measles, which leads to high childhood mortality rate with either little or no response from the public health workers. Therefore, there is the need to forecast measles outbreaks to assist the public health workers facilitate preventive measures in Oyo state. By training a machine learning algorithm, Support Vector Regression (SVR), using the past measles outbreak records (2008-2015), obtained from the ministry of health, Oyo state, Ibadan; thirty-three models representing the thirty-three local governments in Oyo state were generated. Three features were extracted which are, Moving Average (MA), Statistic, and Relative Strength Indicaor. The result of this research project returned a Boolean value which depends on the set outbreak threshold. Mean Squared Error (MSE) and Mean Relative Error (MRE) were the metrics used to measure the performance of the algorithm. Another parameter of significance is the window size, which represents the number of previous data selected in order to estimate each feature from the measles record data. Therefore, it can be concluded that the window size value affected the training time of the algorithm and the efficiency of the models generated. The results of this research can be used as a tool to facilitate the preparedness against Measles outbreak ahead of time.