Aug 2026· International journal of biometeorology· Vol 70· 0 citations· 44 references
Medicine
Abstract
Dengue is an arboviral disease of high public health relevance, characterized by pronounced temporal variability, nonlinearity, and recurrent outbreaks, which pose challenges to epidemiological surveillance and decision-making. This study evaluated the performance of machine learning methods for short-term forecasting of the weekly dengue morbidity rate in the 27 Brazilian capital cities, comprising the 26 state capitals and Brasília, Federal District, with horizons up to 4 weeks. Epidemiological, climatic, and socioeconomic data were compiled for these capital cities and used to compare a Gated Recurrent Unit (GRU) neural network, formulated as a Multi-Input Multi-Output (MIMO) model, and a Gradient Boosting model (CatBoost), implemented using a Direct forecasting strategy with horizon-specific models. Validation was conducted using a walk-forward approach, with evaluation based on absolute error metrics and the coefficient of determination. The results indicated that the GRU architecture presented recurring limitations, including underfitting, temporal lag, and low capacity to anticipate epidemic peaks. In contrast, the CatBoost model demonstrated greater robustness and better adaptation to the variability of epidemiological time series, showing superior performance in most of the analyzed capitals. The findings reinforce that greater architectural complexity does not necessarily imply better operational performance and highlight the potential of ensemble-based methods for short-term epidemiological surveillance applications. These findings contribute to dengue forecasting by showing that, under a common validation framework, ensemble-based strategies may provide greater operational robustness than recurrent MIMO architectures for short-term prediction in heterogeneous epidemiological settings.
Dengue Fever (DF) remains one of the most significant vector-borne diseases in Indonesia, including Sukabumi City, where the number of reported cases exhibits strong seasonal fluctuations influenced by climatic and demographic factors. Accurate forecasting is therefore essential to support early prevention and public health decision-making. This study compares the predictive performance of Long Short-Term Memory (LSTM) and Seasonal Autoregressive Integrated Moving Average with Exogenous Variables (SARIMAX) for forecasting monthly dengue cases in Sukabumi City. The forecasting models incorporate rainfall, temperature, humidity, and population density as predictor variables. The study follows the Cross Industry Standard Process for Data Mining (CRISP-DM) framework, consisting of business understanding, data understanding, data preparation, modeling, evaluation, and deployment. The dataset comprises 96 monthly observations collected between 2018 and 2025 and is divided into training and testing sets using an 80:20 ratio. Model performance is evaluated using Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE). Experimental results demonstrate that the SARIMAX model outperforms LSTM, achieving an MAE of 19.50 and a MAPE of 19.54%, whereas LSTM records an MAE of 20.96 and a MAPE of 21.13%. These findings indicate that SARIMAX is more suitable for forecasting dengue incidence characterized by strong seasonal patterns and relatively limited observations. The proposed comparison provides practical insights for selecting appropriate forecasting models to support evidence-based dengue prevention and public health planning in Sukabumi City.
George Elmar, Asriyanik, Winda Apriandari· Kontribusia (Research Dissem...· 0 citations
Dengue fever is a significant public health issue in tropical and subtropical areas, and predicting its spread is challenging due to the complex interactions between climate factors, mosquito behavior, and case reporting. This study develops a high-resolution forecasting framework that integrates daily dengue case data with meteorological drivers (including mean, maximum, and minimum temperature; precipitation; specific and relative humidity; surface pressure; mean, minimum, and maximum wind speed; wind speed range; wind direction; and sunshine duration) to improve predictions in Bangladesh. Aggregated case counts were disaggregated using a Stochastic Bayesian Downscaling (SBD) algorithm, followed by systematic feature engineering. A wide range of deep learning models, including Artificial Neural Networks (ANN), recurrent networks (LSTM, GRU, BiLSTM, BiGRU), attention-based models, and hybrid convolution neural network (CNN) based ensembles, were optimized through Bayesian hyperparameter tuning and evaluated under a unified process. Results demonstrated that a simple ANN achieved the highest performance, with an accuracy of 97.05%, RMSE of 145.02, and MAPE of 0.51%, surpassing more complex recurrent and attention-based models. Feature importance analysis revealed that weather variables accounted for 76.2% of predictive accuracy, with lagged climate features contributing 59.4%. Short-term lags of three to seven days proved especially influential, underscoring the importance of near-real-time weather monitoring. The strongest predictors included surface pressure with a 3-day lag, precipitation, and maximum wind speed with 7-day lags, while measures of variability such as rolling standard deviations also contributed. The study highlights three key contributions: the benefit of combining downscaling, lagged climate features, and systematic feature selection; evidence that ANN models can outperform more complex architectures in dengue forecasting; and the identification of short-lag weather factors that can support early warning systems. The framework provides an efficient, reproducible, and scalable approach that can strengthen dengue preparedness in public health systems with limited resources.
Mahadee Al Mobin, Arju Manara Begum· PLOS Global Public Health· 0 citations
Vector borne disease like dengue continues to pose a significant climate-sensitive public health challenge in tropical regions such as Brazil, Peru, and India. This study examines the feasibility of predicting dengue outbreaks using weekly multivariate time-series data from San Juan (SJ), Puerto Rico and Iquitos (IQ), Peru. Dengue incidence was analyzed alongside meteorological, environmental, and vegetation-based variables to capture key climatic influences. Several machine learning and deep learning approaches were evaluated, including LightGBM. Model performance was assessed using root mean square error (RMSE) and mean absolute error (MAE). The results show that LightGBM achieved the low est RMSE/MAE, indicating strong short-term predictive accuracy and excellent interpretability. Feature importance analysis and principal component analysis (PCA) identified precipitation, dew point temperature, and humidity as the most influential predictors of dengue incidence. The study demonstrates that advanced machine learning models can serve as reliable early warning systems for vector-borne diseases. While this research focuses on dengue, the methodology is adaptable to other vector-bone datasets and diseases, offering a flexible tool for public health authorities to predict and mitigate outbreaks in diverse urban contexts.
Pratik S. Machchar, Purvi N. Ramanuj, Rajan Patel et al.· International Journal of Inf...· 0 citations
For effective public health preparedness, it is essential to accurately predict the number of dengue cases, especially in tropical and subtropical areas where climate change has a significant impact on how the disease spreads. Standard statistical time-series models, such as the Autoregressive Integrated Moving Average (ARIMA), are commonly utilized but have certain limitations because they assume linear interactions and cannot easily capture climate-disease relationships that are delayed or nonlinear. We examine the feasibility of a Long Short-Term Memory (LSTM)–based forecasting approach that incorporates meteorological factors with lagged dengue incidence to enhance the modelling of temporal dependencies. Monthly dengue surveillance data from Gujarat, India, spanning 2010 to 2019, were integrated with meteorological variables like temperature, relative humidity, and precipitation. Lagged climatic and dengue features were created to take into account short-term effects and seasonal persistence. The dataset was separated into three parts: training, validation, and testing. We trained a single-layer LSTM network with 16 hidden units with dropout regularization. The Adam optimizer was used to train the model, and the mean absolute error (MAE), root-mean-square error (RMSE), and the coefficient of determination (R²) were used to evaluate the model's performance. To test how well the proposed model works in general, it was also tested on dengue datasets from San Juan (Puerto Rico) and Iquitos (Peru). The experimental results show that LSTM model outperformed both the ARIMA and Random Forest baselines on both datasets. The LSTM got a test MAE of about 32 cases and a R² value of 0.82 on the Gujarat test set. This shows that it works well in the real-world dataset. The datasets from Kaggle showed similar gains. These results show that climate-aware LSTM models are good at predicting dengue cases and can help build strong early warning systems in areas where cases are dependent on climate change.
Amiben Mehta, Kajal Patel· ITEGAM- Journal of Engineeri...· 0 citations
Background Dengue fever remains a critical public health challenge in Thailand, with transmission dynamics driven by complex interactions between environmental and socioeconomic factors. Understanding these predictive factors is essential for developing robust forecasting systems. Methods We developed a machine learning framework to classify spatiotemporal dengue risk and identify key predictive factors across Thailand. We analyzed 20 years of monthly dengue hemorrhagic fever surveillance data (2003–2022) from 77 provinces, integrating 54 environmental, climatic, and socioeconomic features. We benchmarked four candidate classifiers — logistic regression, support vector machines, random forests, and eXtreme Gradient Boosting (XGBoost) — and selected XGBoost on the basis of performance across six metrics. SHapley Additive exPlanations (SHAP) were used to interpret feature contributions. The dataset was stratified into training (2003–2016) and testing periods, with the latter subdivided into pre-COVID-19 (2017–2019), COVID-19 (2020–2021), and post-COVID-19 (2022) phases. Results The XGBoost model achieved an AUC of 0.80 in pre-pandemic testing and 0.74 across the combined during- and post-pandemic period. Temperature dominated the feature-importance ranking, comprising seven of the top ten features, with non-linear thresholds near 21°C for 1-month lagged minimum temperature and near 32°C for 3-month lagged maximum temperature — values that align with established biological constraints on Aedes aegypti–mediated transmission. Precipitation features contributed minimally to model predictions, while a higher Gross Provincial Product was associated with increased dengue risk, consistent with predominantly urban transmission patterns. Model performance deteriorated significantly during the COVID-19 pandemic (AUC = 0.62 in 2021), with systematic overprediction indicating that non-environmental factors operating outside the model dominated dengue dynamics during this period. Conclusions Temperature is the dominant predictor of dengue risk in Thailand, and the thresholds we recover correspond closely to known biological constraints on vector competence. Environmentally driven prediction is reliable under stationary conditions but degrades substantially during periods of major societal disruption, underscoring the need to integrate behavioral and surveillance-coverage indicators alongside environmental predictors when applying such models in real time.
P. Suttirat, S. Chadsuthi, S. Aekthong et al.· PLoS Neglected Tropical Dise...· 0 citations
Accurate forecasting of infectious disease dynamics is essential for effective public health planning and early outbreak response. Traditional statistical models and modern deep learning approaches have both been widely used for epidemiological time–series forecasting. Dengue fever remains a major public health concern in Bangladesh, where seasonal outbreaks often produce sharp and unpredictable infection spikes. Reliable forecasting models are therefore important for anticipating infection trends and supporting early intervention strategies. This study investigates the performance of Long Short-Term Memory (LSTM) networks for predicting daily dengue infection counts in Bangladesh using surveillance data from 2010 to 2022. Multiple LSTM architectures were developed by varying the number of LSTM units, lookback window size, and data preprocessing strategies. A fixed time-based data split was used, with training data up to 30 September 2021 and testing on subsequent observations to avoid information leakage. The proposed LSTM models were compared with several baseline forecasting approaches, including Naive, Seasonal-Naive, and ARIMA models. Model performance was evaluated using standard forecasting metrics including Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Symmetric Mean Absolute Percentage Error (SMAPE). Experimental results indicate that classical statistical models remain strong benchmarks for dengue forecasting. Among all evaluated models, the ARIMA(2,1,2) model achieved the best predictive performance on the test dataset with MAE of 8.40 and RMSE of 14.72. The best-performing LSTM configuration produced competitive results with MAE of 8.92 and RMSE of 15.87, outperforming simple baseline approaches such as the Naive and Seasonal-Naive models. These findings suggest that while deep learning models are capable of capturing complex temporal patterns, their advantage over well-tuned statistical models is not guaranteed for this dataset. The results highlight the importance of benchmarking deep learning models against classical time-series approaches when forecasting epidemiological data. While LSTM networks demonstrate promising performance in modeling nonlinear temporal dynamics, the ARIMA model provided the most accurate predictions for the studied dengue dataset.
Tahmid Nowsher, Shaiful Islam Arafat, Md. Kamrujjaman· Discover Public Health· 0 citations