Maleria Prevention Using Social Media and Text Mining

Abstract

The battle with malaria especially in the African continent still exists and has been taking the lives of many in the area, so there is a need to keep fighting the battle, monitor progress and challenges. One way is the usage of social media particularly twitter as a tool to fight malaria. Research has been done on malaria twitter data to classify tweets as malaria and non-malaria cases using support vector machine (SVM) which is used in the predictions of future tweets to avoid outbreaks. Malaria twitter data has also been used to find trends and patterns on public opinions regarding malaria topics which is used by health sectors in managing funds allocation and making informed decisions. The objective of this study is to tap into Nigerian Malaria twitter data to understand public opinions of tweets relating to malaria, gain insight into the data to find trends and patterns and compare results with WHO battle against malaria. We describe a combine approach of sentiment analysis, word cloud and topic modelling using LDA. The sentiment analysis is for assessing public opinion about malaria in Nigeria. Word cloud for data visualization and LDA to find hidden topics which is compared to WHO fight against malaria. Despite the small size of the data set, the word cloud visualized topics with the highest frequency and this could be labelled as topics creating awareness on malaria, malaria treatment, testing before treating malaria and the goal of having a malaria free Nigeria. The LDA result correlated well with WHO’s battle against malaria and issues the battle is still facing like adverse effect of malaria on pregnant women and young children under age 5. The sentiment analysis provided us sentiment and public opinion of tweets with 42.6% positive, 15.6% negative and 41.8% neutral