Time-series forecasting of mRDT-positive malaria counts in Tanzania’s lake zone: a comparative evaluation of statistical and machine-learning models using routine surveillance data, 2018–2025
- 1Department of Science and Management, Dar es Salaam Maritime Institute, P.O. Box 6727, Dar es Salaam, Tanzania
- 2Department of Mathematics, University of Dar es Salaam, P.O. Box 35091, Dar es Salaam, Tanzania
- 3Global Water Parternership Tanzania, P.O. Box 32334, Dar es Salaam, Tanzania
Abstract
Background: Monthly malaria forecasts can support diagnostic and treatment planning when the surveillance outcome and evaluation protocol are clearly defined.
Methods: We analysed 4,128 council-month records from 43 councils in six Lake Zone regions during January 2018–December 2025. Complete mRDT-positive counts were aggregated to 576 region-month observations and one 96-month Lake Zone series. Six fitted models, an equal-weight ensemble and a seasonal-naive benchmark were evaluated on a fixed-origin 2025 holdout and 37 expanding-window origins with 1–12-month horizons.
Results: The dataset contained 11,874,519 mRDT-positive counts. Annual counts declined from 1.63 million in 2018 to 1.06 million in 2025, with the highest calendar-month averages in June–July. ARIMA (2, 1, 2) had the lowest 2025 holdout error (MAPE 12.1%; RMSE 11,709), reducing RMSE by 69.2% relative to seasonal naive. Across all rolling origin–horizon pairs, Gradient Boosting had the lowest RMSE (22,007), while the ensemble had the lowest MAE (16,687) and MAPE (16.1%).
Conclusions: Model rankings varied with forecast origin, horizon and metric. The findings favour routinely updated, benchmarked forecasts over reliance on a single terminal-year ranking. The outcome represents positive diagnostic counts rather than population-standardized incidence or confirmed unique outpatient episodes.
Supplementary Information: The online version contains supplementary material available at https://doi.org/10.1186/s41182-026-01105-3.
Related
- Adaptive forecasting and alert evaluation for pediatric RSV and influenza hospitalization burden using public U.S. surveillance data: a time-series study
- SDS screen-positive prediction from structured questionnaires in university students: a comparative evaluation of large language model prompting strategies
- Forecasting Culex pipiens/restuans abundance in Ontario, Canada using surveillance and weather data
- Forecasting Ethiopia's Neonatal Mortality Rate Toward the Sustainable Development Goal Target 3.2 Using Statistical and Deep Learning Models: A Time-Series Forecasting Study.
- Patient Experiences With ctHPVDNA-Guided Surveillance for HPV-Positive Oropharyngeal Cancer
BibTeX
@article{lugoye2026timeseries,
title = {{Time-series forecasting of mRDT-positive malaria counts in Tanzania’s lake zone: a comparative evaluation of statistical and machine-learning models using routine surveillance data, 2018–2025}},
author = {Jonnes James Lugoye and Theresia Crispin Marijani and Idrissa Said Amour and Frank Enock Anderson},
journal = {Tropical Medicine and Health},
year = {2026},
month = nov,
doi = {10.1186/s41182-026-01105-3},
url = {https://doi.org/10.1186/s41182-026-01105-3},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - Time-series forecasting of mRDT-positive malaria counts in Tanzania’s lake zone: a comparative evaluation of statistical and machine-learning models using routine surveillance data, 2018–2025
AU - Jonnes James Lugoye
AU - Theresia Crispin Marijani
AU - Idrissa Said Amour
AU - Frank Enock Anderson
PY - 2026
DA - 2026/11/06
JO - Tropical Medicine and Health
DO - 10.1186/s41182-026-01105-3
UR - https://doi.org/10.1186/s41182-026-01105-3
AB - Background: Monthly malaria forecasts can support diagnostic and treatment planning when the surveillance outcome and evaluation protocol are clearly defined. Methods: We analysed 4,128 council-month records from 43 councils in six Lake Zone regions during January 2018–December 2025. Complete mRDT-positive counts were aggregated to 576 region-month observations and one 96-month Lake Zone series. Six fitted models, an equal-weight ensemble and a seasonal-naive benchmark were evaluated on a fixed-origin 2025 holdout and 37 expanding-window origins with 1–12-month horizons. Results: The dataset contained 11,874,519 mRDT-positive counts. Annual counts declined from 1.63 million in 2018 to 1.06 million in 2025, with the highest calendar-month averages in June–July. ARIMA (2, 1, 2) had the lowest 2025 holdout error (MAPE 12.1%; RMSE 11,709), reducing RMSE by 69.2% relative to seasonal naive. Across all rolling origin–horizon pairs, Gradient Boosting had the lowest RMSE (22,007), while the ensemble had the lowest MAE (16,687) and MAPE (16.1%). Conclusions: Model rankings varied with forecast origin, horizon and metric. The findings favour routinely updated, benchmarked forecasts over reliance on a single terminal-year ranking. The outcome represents positive diagnostic counts rather than population-standardized incidence or confirmed unique outpatient episodes. Supplementary Information: The online version contains supplementary material available at https://doi.org/10.1186/s41182-026-01105-3.
SN - 1348-8945
SN - 1349-4147
ER -