rohanp2051/MalariaPredictionFromSatellite2020

Linear Regression, Lasso Regression, DT, RF, and Neural Nets were used to see the relationship between spatiotemporal factors and malaria caseload in Ibadan, Nigeria. # Machine learning models to predict malaria caseload using remote sensing data ## Research Plan Malaria harms human health and devastates socioeconomic conditions in low-income countries. There were 230 million infections and 420,000 deaths last year. Though there has been enormous progress since the 2000s, progress has plateaued in recent years, and innovative methods are needed to address the challenge. Remote sensing temporal variables such as temperature, humidity, rainfall, etc., were used to predict malaria caseload for continents or countries using mathematical models. The accuracy of said models varies significantly for local areas because of the inconsistent local temporal data and parameters that impact malaria transmission. I believe malaria caseloads can be predicted more accurately using a machine learning model that uses the spatiotemporal variable vegetation index in conjunction with temporal variables for small local areas such as towns, districts, or cities. I plan to build a machine learning model using remote sensing parameters such as satellite observed vegetation index, local weather, and historical malaria caseload to improve caseload prediction accuracy. I have divided the process into five steps. Data Download - Download data from Google earth engine/NASA's earth data API, historical malaria caseload data from the public domain/malaria atlas database. Clean and prepare various datasets- Pre-process the data into a machine-consumable format, calculate vegetation index, etc. Prepare different datasets from training, testing and validation. Build the model - The Neural net model written using Python and TensorFlow. Train the model - Training the model using training dataset and model algorithm written in python. Analyze results - Analyze results and formulate plots and visual maps based on results. ## Abstract It has been estimated that malaria has been responsible for more deaths than any other disease in human history. Despite progress …