Modeling Malaria Outbreaks Utilizing Weather Factors

Anonymous

Keywords: machine learning, transfer learning, epidemiology, malaria, dengue, weather, disease predication, recurrent neural networks, long short term memory network

TL;DR: A look into modelling mosquito-borne disease outbreaks based on weather conditions

Abstract:

Using meteorological data, time series forecasts of disease outbreaks can better
capture the true epidemiological profile of tropical diseases such as malaria. In this
study, several methods of time series analysis are employed to study the disease
patterns of malaria in the Indian state of Odisha. Weather information, including
temperature and precipitation data, is incorporated alongside monthly case numbers
in SARIMA and LSTM models. The viability of transferring the model trained on
malaria in Odisha to dengue in Bangkok, Thailand, is also explored. The methods
outlined in this paper can serve as the basis for forecasting mosquito-borne disease
outbreaks in settings with a poor data-collection infrastructure.