Background: Dengue is a mosquito-borne viral disease that poses a significant public health threat in tropical and subtropical regions worldwide. Accurate forecasting of dengue outbreaks is crucial for effective public health planning and intervention. This study aims to assess the predictive performance and computational efficiency of a number of statistical models and machine learning techniques for dengue forecasting, both with and without the inclusion of climate factors, to inform the design of dengue surveillance systems. Methods: The dengue forecasting methods comparison in this study considers dengue cases in Rio de Janeiro, Brazil, as well as climate factors known to affect disease transmission. Employing a dynamic window approach, various statistical methods and machine learning techniques were used to generate weekly forecasts at several time horizons. Error measures, uncertainty intervals, and computational efficiency obtained with each method were compared. Statistical models considered were Autoregressive (AR), Moving Average (MA), Autoregressive Integrated Moving Average (ARIMA), and Exponential Smoothing State Space Model (ETS). In addition, models incorporating temperature and humidity as covariates, such as Vector Autoregression (VAR) and Seasonal ARIMAX (SARIMAX), were employed. Machine learning techniques evaluated were Random Forest, XGBoost, Support Vector Machine (SVM), Long–Short-Term Memory (LSTM) networks, and Prophet. Ensemble approaches that integrated the top performing methods were also considered. The evaluated methods also incorporated lagged climatic variables to account for delayed effects. Results: Among the statistical models, ARIMA demonstrated the best performance using only historical case data, while SARIMAX significantly improved predictive accuracy by incorporating climate covariates. In general, the LSTM model, particularly when combined with climate covariates, proved to be the most accurate machine learning model, despite being slower to train and predict. For long-term forecasts, Prophet with climate covariates was the most effective. Ensemble models, such as the combination of LSTM and ARIMA, showed substantial improvements over individual models. Conclusions: This study demonstrates the strengths and limitations of various methods for dengue forecasting across multiple timeframes. It highlights the best-performing statistical and machine learning methods, including their computational efficiency, underscoring the significance of machine learning techniques and the integration of climate covariates to improve forecasts. These findings offer valuable insights for public health officials, facilitating the development of dengue surveillance systems for more accurate forecasting and timely allocation of resources to mitigate dengue outbreaks.

Dengue fever is a mosquito-borne viral disease that poses significant health risks and socioeconomic challenges in Brazil, necessitating accurate forecasting across its 27 federal states. With the country’s diverse climate and geographical spread, effective dengue prediction requires models that can account for both climate variations and spatial dynamics. This study addresses these needs by using Long Short-Term Memory (LSTM) neural networks enhanced with SHapley Additive exPlanations (SHAP) integrating optimal lagged climate variables and spatial influence from neighboring states. An LSTM-based model was developed to forecast dengue cases across Brazil’s 27 federal states, incorporating a comprehensive set of climate and spatial variables. SHAP was used to identify and select the most important lagged climate predictors. Additionally, lagged dengue cases from neighboring states were included to capture spatial dependencies. Model performance was evaluated using MAE, MAPE, and CRPS, with comparisons to baseline models. The LSTM-Climate-Spatial model consistently demonstrated superior performance, effectively integrating temporal, climatic, and spatial information to capture the complex dynamics of dengue transmission. SHAP-enhanced variable selection improved accuracy by focusing on key drivers such as temperature, precipitation and humidity. The inclusion of spatial effects further strengthened forecasts in highly connected states showcasing the model’s adaptability and robustness. This study presents a scalable and robust framework for dengue forecasting across Brazil, effectively integrating temporal, climatic, and spatial information into an LSTM-based model. The model’s successful application across Brazil’s diverse regions demonstrates its generalizability to other dengue-endemic areas with varying climatic and epidemiological conditions. By integrating diverse data sources, the framework captures key transmission drivers, demonstrating the potential of LSTM neural networks for robust predictions. These findings provide valuable insights to enhance public health strategies and outbreak preparedness in Brazil.

Dengue fever is a major global health concern, with Brazil experiencing recurrent and severe outbreaks due to its favorable climate factors, socio-environmental conditions, and increasing human mobility. Accurate forecasting of dengue cases and outbreak risk is essential for early warning systems and effective public health interventions. Traditional forecasting models primarily rely on historical case data and climate variables, often neglecting the role of human movement in virus transmission. This study addresses this gap by incorporating human mobility data into a deep learning-based dengue forecasting framework. An LSTM-based model was developed to forecast weekly dengue cases and detect outbreaks across selected Brazilian cities. The model integrates historical dengue cases, lagged climate variables (temperature and humidity), and human mobility-adjusted imported cases to capture both temporal trends and spatial transmission dynamics. Its performance was evaluated against three alternative models: (1) an LSTM using only dengue case data, (2) an LSTM incorporating climate variables, and (3) an LSTM integrating climate and geographic neighborhood effects. Forecasting accuracy was assessed using Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Continuous Ranked Probability Score (CRPS), while outbreak classification was evaluated using accuracy, sensitivity, specificity, and the F1 score. The proposed mobility-enhanced LSTM model consistently outperformed all baselines in both dengue case forecasting and outbreak detection. Across all cities, it achieved lower MAE and MAPE values, indicating improved accuracy, while also demonstrating superior CRPS performance, reflecting well-calibrated uncertainty estimates. In outbreak classification, the model achieved the highest sensitivity and F1 scores, highlighting its effectiveness in detecting outbreak periods compared to models that relied solely on case trends, climate variables, or geographic proximity. The results underscore the importance of integrating mobility data in dengue forecasting, particularly in urban centers with high population movement. By incorporating human mobility dynamics into deep learning-based forecasting, this study presents a scalable and adaptable framework for enhancing dengue early warning systems. The proposed model provides more accurate case predictions and outbreak classifications, offering actionable insights for public health planning and resource allocation. Beyond dengue, this approach can be extended to other vector-borne diseases influenced by mobility and climate factors, supporting more effective epidemic preparedness strategies worldwide.

Background: Dengue fever is a major global health concern, with Brazil experiencing recurrent and severe outbreaks due to its favorable climate factors, socio-environmental conditions, and increasing human mobility. Accurate fore-casting of dengue cases and outbreak risk is essential for early warning systems and effective public health interventions. Traditional forecasting models primarily rely on historical case data and climate variables, often neglecting the role of human movement in virus transmission. This study addresses this gap by incorporating human mobility data into a deep learning-based dengue forecasting framework. Method: An LSTM-based model was developed to forecast weekly dengue cases and detect outbreaks across selected Brazilian cities. The model integrates historical dengue cases, lagged climate variables (temperature and humidity), and human mobility-adjusted imported cases to capture both temporal trends and spatial transmission dynamics. Its performance was evaluated against three alternative models: (1) an LSTM using only dengue case data, (2) an LSTM incorporating climate variables, and (3) an LSTM integrating climate and geo-graphic neighborhood effects. Forecasting accuracy was assessed using Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Continuous Ranked Probability Score (CRPS), while outbreak classification was evaluated using accuracy, sensitivity, specificity, and the F1 score. Results: The proposed mobility-enhanced LSTM model consistently outperformed all baselines in both dengue case forecasting and outbreak detection. Across all cities, it achieved lower MAE and MAPE values, indicating improved accuracy, while also demonstrating superior CRPS performance, reflecting well-calibrated uncertainty estimates. In outbreak classification, the model achieved the highest sensitivity and F1 scores, highlighting its effectiveness in detecting outbreak periods compared to models that relied solely on case trends, climate variables, or geographic proximity. The results underscore the importance of integrating mobility data in dengue forecasting, particularly in urban centers with high population movement. Conclusion: By incorporating human mobility dynamics into deep learning-based forecasting, this study presents a scalable and adaptable framework for enhancing dengue early warning systems. The proposed model provides more accurate case predictions and outbreak classifications, offering actionable insights for public health planning and resource allocation. Beyond dengue, this approach can be extended to other vector-borne diseases influenced by mobility and climate factors, supporting more effective epidemic preparedness strategies worldwide.

Recent several years have witnessed the rapid explosion of artificial intelligence applied in various domains with the surpassing human-level performance. Despite the success, these models’ underlying mechanisms remain a mystery, as their complicated representations make human understanding impossible. This mystery may cause discrimination and non-robustness in prediction. Making deep learning models more transparent and understandable is gaining popularity, but most of interpretation approaches provide spurious correlations leading to suboptimal, incorrect or even biased interpretations, which could be reduced by causal explanations. Motivated by this, we attempt to study the generation of causal explanations and propose CUBE , a causal intervention-based counterfactual interpretation method. To ensure that the generation process of counterfactual explanation conforms to causality, we model the counterfactual generation process as a causal graph and construct a counterfactual generation model based on the causal intervention; to generate counterfactuals that adhere to the causality, we introduce a causal director to capture the causal relationships in the distribution and guide the generation of counterfactuals; to improve the efficiency of the counterfactual generation when facing a large number of explanation queries, we model it as a sample generation problem and propose an explainable framework based on adversarial generation. The experimental results validate that CUBE outperforms other approaches in terms of both lower time costs and higher explanation quality.