Mining environmental data for prediction of transmission patterns of communicable diseases

Climate changes, such as overall warmer temperatures, increased rainfall, or increased water scarcity and even more extreme weather events, such as flooding and acute precipitation events have measurable effects on the varied climates found across countries like India. The magnitude of these changes impacts the local conditions and the specific ecology and epidemiology of the different diseases. Climatic conditions strongly affect air-borne, water-borne and vector-borne diseases. They lengthen the transmission seasons of important vector-borne diseases and alter their geographic range. The transmission patterns of communicable diseases are influenced by many factors and diseases tend to get introduced to regions that have not previously encountered them. Therefore capacities need to be built to assess how climate change might alter the effectiveness of proposed programs or might affect future population health so that the cumulative or catastrophic events with large health impacts could be avoided. These evaluations should consider short-term rapid climate change as well as longer-term changes in means of meteorological variables. Mining of such scientific databases is often very different from traditional market-driven data mining applications. It involves datasets comprising vast amounts of precise and continuous data. In such scenarios accounting for underlying system nonlinearities can be extremely challenging from a machine learning point of view. We study the inferences that can be drawn by mining of environmental data and correlating it with the disease data to predict the transmission patterns of communicable diseases.