Early Warning of Malaria Disease Based on Satellite Imagery Environmental and Climatic Data Using Gradient Boost Algorithm

Death rate due to malaria was increased in 2020. This is expected to increase in the forthcoming year globally accordingly to World Health Organization (WHO). This research area aims at a study of the risk of malaria incidence in Tamil Nadu in India as per environmental and climatic parameters. Considering the malaria as highly correlated with temperature, the author used Land Surface Temperature (LST), Air Temperature (AT), and Earth Skin Temperature (EST) along with Precipitation, Surface Pressure (SP) and Relative Humidity (RH) as a predictor. All these parameters were estimated or derived from the Landsat 8/NASA satellite images during the year 2013-2021. These predictors were applied into the most significant ensemble modeling called Gradient Boost (GB) along with the clinical data during the year 2013-2021.The GB modeling was trained to predict malaria incidence during the year 2013-2017 and it was validated during 2018-2021. Achievement of 97% of accuracy was seen in validation. The GB modeling results was compared with those of another significant ensemble modeling called Random Forest (RF). The GB modeling produced 97% of accuracy whereas RF modeling produced only 79% of accuracy. Hence GB outperformed in prediction accuracy with 18% increase compared with RF modeling. So, the use of GB modeling helped prediction of the malaria incidence rate for the near future 2022-2024 an exponential decrease in the risk of malaria vector borne disease for the region of Tamil Nadu in India was noted. Thus this malaria early warning system using GB modeling communicates, though the threat of malaria vector borne disease is increasing world-wide showed a decreasing incidence rate in the near future which is caused by the spatial and temporal pattern of environmental and climatic parameters in the region of Tamil Nadu in India.