Time series forecasting of plasmodium falciparum malaria epidemic: A comparative analysis of ARIMA and Integrated Artificial Neural Networks - A case study of the Ministry of Health AND Child Care, Mt Darwin District.

Author
Dzapasi, Nyasha
Title
Time series forecasting of plasmodium falciparum malaria epidemic: A comparative analysis of ARIMA and Integrated Artificial Neural Networks - A case study of the Ministry of Health AND Child Care, Mt Darwin District.
Abstract
Malaria caused mainly by Plasmodium falciparum is still a significant public health problem in rural areas of Zimbabwe. This study develops and contrasts time series forecasting models to predict monthly malaria incidence and mortality in Mt Darwin District, using historical data from January 2013 to December 2023. Two model approaches were employed, a standard Autoregressive Integrated Moving Average (ARIMA) model and an integrated computational intelligence strategy incorporating Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and Feedforward Neural Networks (FFNN). ARIMA model selection was consistent with the Box-Jenkins method, while the hybrid neural network was trained with a 12-month sliding input window. Model performance was assessed with a held-out 2024 test set with Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and the coefficient of determination (R²). Optimal ARIMA specifications (2, 1, 3) for incidence and (1, 1, 1) for mortality achieved moderate accuracy (R² = 0.83 for incidence), but unacceptability for mortality (R² = –0.16). The CNN+LSTM+FFNN hybrid model performed the best among all models with an MAE of 5.43, RMSE of 96.85, and R² = 0.94 for mortality, and an MAE of 2, RMSE of 2.16, and R² = 0.91 for incidence. 2025-2030 projections of malaria case declines from 1,824 in 2025 to 1,703 in 2030 and of deaths from 46 to 33 over the same period, with seasonal highs in February to April. These findings illustrate the strength of hybrid neural networks in modeling nonlinear, intricate patterns of disease in under researched environments. The study recommends that Mt Darwin District Health officials and the Ministry of Health and Child Care coordinate antimalarial procurement with NatPharm, augment bed-net and diagnostic kit distribution in high months, intensify targeted indoor
Date
June 2025
Publisher
BUSE
Keywords
Plasmodium Falciparum Malaria Epidemic
Supervisor
Mr. K. Basira
Item sets
Department of Statistics and Financial Mathematics
Media
Dzapasi, Nyasha.pdf

Part of Time series forecasting of plasmodium falciparum malaria epidemic: A comparative analysis of ARIMA and Integrated Artificial Neural Networks - A case study of the Ministry of Health AND Child Care, Mt Darwin District.