Malaria remains a major public health problem worldwide, and the World Health Organization predicts that the number of malaria cases will reach 249 million and the number of deaths will reach 608,000 by 2022. Children under five are the most vulnerable group, accounting for approximately 80% of all malaria deaths in the region. Infectious diseases such as malaria are greatly affected by climate change in the region. In this work, we present a deep learning model with a supervised technique called Long short-term model (LSTM). We performed a comparative study using a number of time series and machine learning models, such as Support Vector Regression (Polynomial Kernel), XGBoost, LightGBM, SARIMA, SARIMAX, Random Forest, Linear Regression, and Logistic Regression, in order to verify the effectiveness of our suggested model. Out of all the models, the LSTM model has the lowest Root Mean Squared Error (RMSE). We also developed a Bi-LSTM model, which confirmed the superior performance of our method with an even lower RMSE. The efficacy of the feature attention-based LSTM model in forecasting malaria incidence from meteorological data is demonstrated by its lower validation loss and error metrics (RMSE and MAE) as compared to the temporal attention-based LSTM, as indicated by the performance metrics.