https://orcid.org/0009-0009-7258-7174
Figures
Abstract
Dengue is a mosquito-borne viral disease affecting tropical and subtropical regions. In Bangladesh, dengue fever remains a rising public health threat driven by meteorological factors. This study aimed to assess the temporal trends and how meteorological factors influence dengue incidence in Bangladesh from 2008 to 2024. Monthly reported dengue cases were analyzed using time series forecasting techniques and multivariate Poisson regression models. Seasonal Autoregressive Integrated Moving Average (SARIMA) and Extreme Gradient Boosting (XGBoost) models were used for forecasting. Correlation analysis and Poisson regression assessed meteorological effects with one- and two-month lags. The result indicates that the highest number of dengue cases was found in September 2023 (79,598 cases). Autocorrelation revealed a strong positive correlation at 1-month and 2-month lags. Forecasts from 2024–2027 predict that dengue cases will fluctuate between 10,000 and 20,000 annually from the predictive models. Spearman’s rank correlation indicated significant positive associations between dengue cases and precipitation, temperature, wind speed, and humidity. Multivariable Poisson regression revealed that temperature (°C) (IRR = 1.02), Humidity (%) (IRR = 1.25), and Wind speed (m/s) (IRR = 1.10) significantly increased dengue incidence. Between multivariate SARIMA, XGBoost, and Poisson regression, the best-performing model was ARIMA (RMSE: 5058.066). In conclusion, the study highlights the substantial influence of climatic factors on dengue dynamics in Bangladesh, emphasizing the need to integrate meteorological data into early warning systems and develop adaptive, climate-informed control and surveillance strategies.
Citation: Alam KE, Ahmed MJ, Chalise R, Rahman MA, Mathin TT, Bhuiyan MIH, et al. (2025) Time series analysis of dengue incidence and its association with meteorological risk factors in Bangladesh. PLoS One 20(8): e0323238. https://doi.org/10.1371/journal.pone.0323238
Editor: Md. Kamrujjaman, University of Dhaka, BANGLADESH
Received: April 6, 2025; Accepted: July 13, 2025; Published: August 18, 2025
Copyright: © 2025 Alam et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: DF, dengue fever; SARIMA, seasonal autoregressive integrated moving average; RMSE, root mean squared error; IRR, incidence rate ratio; CI, confidence interval; ADE, antibody-dependent enhancement; ARIMA, autoregressive integrated moving average; SVM, support vector machines; ANN, artificial neural networks; HMM, Hidden Markov models; IEDCR, Institute of Epidemiology, Disease Control and Research; BMD, Bangladesh Meteorological Department; AR, autoregression; MA, moving average; MPR, multivariate Poisson regression; VIF, VARIANCE inflation factors; ACF, autocorrelation function; PACF, partial Autocorrelation function; MAE, mean absolute error; SD, standard deviation; MAPE, mean absolute percentage error
Introduction
Dengue is the most significant mosquito-borne viral disease and is found mainly in tropical and subtropical areas. However, due to globalization and climate change, it has spread to countries where it was not common before [1,2]. This disease is caused by four closely related viruses but antigenically distinct virus serotypes (DEN 1–4) of the genus Flavivirus and is transmitted primarily by the Aedes (A.) aegypti, the primary urban vector, and A. albopictus, the secondary vector [3]. Almost 100 nations in Africa, the Americas, Southeast Asia, the Eastern Mediterranean, and the Western Pacific now have dengue as an endemic disease [4]. More than 50% of the global population is at risk of contracting dengue, with an estimated 50–200 million cases annually [5]. Severe dengue results in approximately 12,500 fatalities each year, primarily in endemic regions where nearly 2.5 billion people are affected [6].
Numerous interconnected elements, including viral serotypes and climatic, hydrological, and societal factors, influence the spread of dengue [7]. Rainfall, temperature, and humidity are examples of meteorological parameters that are closely linked to dengue outbreaks. For instance, it has been demonstrated that higher temperatures and rainfall increase the likelihood of dengue infection by increasing the number of mosquito-breeding sites, improving the mosquito breeding cycle, encouraging vector development (shortening the maturation period), increasing vector survival, and promoting blood-feeding patterns. As a result, there are more vectors in the population, which increases the likelihood that a vector and a human would interact and spread the virus [8,9]. The Aedes mosquito has four stages in its life cycle: the eggs, larvae, and pupae are aquatic, and the adult mosquito is terrestrial. Climate variables such as temperature and rainfall affect each stage of the life cycle [10,11]. Research has shown that A. albopictus cannot develop at all at a constant temperature of 10°C and that eggs cannot develop at 15°C. However, when the temperature increased to 30°C, the rate at which each stage developed increased steadily, with the fastest development occurring for the larvae and pupae at approximately 30°C [12]. In addition, relatively high temperatures make it possible for mosquito larvae or adults to hibernate [13–16]. The endemic range of Dengue Fever (DF) is expected to spread regionally as long as global temperatures rise [17–19]. Increased reproduction and activity, and a shorter larval incubation period are all made possible by warmer temperatures, which increase the ability to produce progeny. Accordingly, it appears likely that the prevalence and potential for dengue transmission will increase [8,20]. By lengthening the season during which transmission occurs, rising temperatures may also contribute to an increase in dengue transmission [21]. Extended dry spells in endemic regions lacking a reliable source of drinking water could promote water storage, which would increase the number of mosquito breeding grounds used by the A. aegypti vector [22].
Consequently, Bangladesh has experienced dengue outbreaks in recent years that have never been severe. Furthermore, Bangladesh is close to nations where dengue is endemic, it is constantly in danger of importing novel strains of the disease [23]. The deadlier secondary dengue infection may have resulted from the introduction of a new serotype through antibody-dependent enhancement (ADE) [24,25]. The first DF was officially identified in 1964 in Bangladesh [26]. The country experienced its first dengue outbreak during the monsoons of 2000, which resulted in 5551 confirmed cases and 93 deaths [23]. A sudden rise in dengue outbreaks was found in 2018 (10,148 cases and 26 deaths), 2019 (101,354 cases and 179 deaths), and 2023 (321,179 cases with 1,705 deaths) [27,28]. Multiple factors contribute to the proliferation of dengue vector habitats in Bangladesh, such as high densities of infected mosquitoes, limited population immunity to circulating dengue serotypes, and poor housing conditions characterized by inadequate waste management, sanitation, drainage, and access to clean water [29–31]. The use of unsafe water storage containers further exacerbates the problem. Additionally, the country’s tropical monsoon climate, marked by heavy rainfall (record highest rainfall in 2023) and favorable temperatures, creates ideal conditions for mosquito breeding, thereby increasing the frequency of dengue outbreaks.
Over the past ten years, a growing amount of research has been conducted to anticipate infectious diseases, specifically dengue and malaria. Traditional techniques, which look for a linear association between underlying causes and disease occurrence, such as regression [30] or autoregressive integrated moving average (ARIMA) [32] Still, achieves a limited level of accuracy. Moreover, contemporary methods such as Support vector machines (SVMs) [33] Artificial neural networks (ANNs) [34,35], and Hidden Markov Models (HMMs) [36] have been developed. The absence of a novel or more robust modeling approach in some studies limits potential advancements in predictive [37–40]. Furthermore, few studies have done predictive modeling like XGBoost for comparing the models and model accuracy, but those studies did not predict the long-term cases, also some of the studies are from other regions, not in Bangladesh [41–45].
Previous studies examining the relationship between climate factors and dengue incidence in Bangladesh have focused on specific regions, particularly Dhaka, and have relied on outdated datasets. This limits their ability to capture recent national trends in dengue transmission. Moreover, most of these studies have not employed predictive modeling techniques, and only a few have utilized basic count regression approaches. The absence of machine learning-based forecasting and comprehensive statistical modeling underscores the need for updated, nationwide research that integrates recent data and robust predictive tools. Such studies are essential to better understand the evolving dynamics of dengue outbreaks under changing climatic conditions across Bangladesh [46–51]. Furthermore, the unusual weather patterns due to climate change, particularly the record rainfall in 2023, have significantly exacerbated the dengue situation in Bangladesh. Thus, this study aims to analyze temporal trends in dengue incidence in Bangladesh and forecast future outbreaks until 2027 using the SARIMA model and XGBoost. Additionally, it examines the influence of dengue-related meteorological factors: temperature, rainfall, humidity, and wind speed, through multivariate Poisson regression. We hypothesize that dengue incidence in Bangladesh is significantly influenced by meteorological factors. The study’s findings will aid public health authorities in developing climate-informed early warning systems and targeted dengue control policies.
Materials and methods
Data collection
For this research study, data on monthly dengue cases from January 2008- November 2024 were obtained from the Institute of Epidemiology, Disease Control and Research (IEDCR) website (https://iedcr.portal.gov.bd/site/page/45aea1fa-5756-4feb-8d09-f0da895a3baa/-), Dhaka, Bangladesh [52,53]. The IEDCR provides regularly updated public health data, which ensures the accuracy and relevance of the dengue case statistics used in the study. Similarly, meteorological variable measurement data from January 2008- November 2024 were collected from the Bangladesh Meteorological Department (BMD) (https://bmd.gov.bd/), a reliable platform offering access to a range of environmental and meteorological data of Bangladesh. The data we collected were recorded via the following units: temperature measured in degrees Celsius (°C), rainfall recorded in millimeters (mm), relative humidity expressed as a percentage (%), and wind speed measured in meters per second (m/s). The dataset used in this study was complete, with no missing months or data gaps during the study period. Although the model incorporates key climatic and epidemiological predictors, socio-economic variables such as income, education, and urbanization levels were not included due to data unavailability. This is acknowledged as a limitation, as these factors are known to influence disease dynamics and healthcare access.
Statistical analysis
Descriptive.
A general descriptive summary and box-and-whisker plots were used to analyze the spread time and determine the dengue case distribution in Bangladesh. A box plot provides a visual representation of the distribution. The box extends from the bottom hinge, which is the 25th percentile, to the top hinge, which is the 75th percentile. The line across the box represents the median. To make sure there were no zero values in any month, we added 1 to the total number of cases each month. Since the dengue cases data was not evenly spread out, we used a natural log transformation to make it more suitable for analysis. Later, we changed the log-transformed values back to their original form to make the results easier to understand [54].
Decomposition.
The seasonal decomposition process based on loess (STL) is used in a variety of study fields, such as the natural sciences, environmental science, and public health, to analyze time series data. SLT separates the long-term, low-frequency variation in the data into three components: trend (the variation in the data over the same period), seasonal (the variation within the same period), and random or remainder (the remaining variation in the data after extracting the trend and seasonal component). The benefits of SLT include its ease of use, reliable outcomes, and potent data visualization. where ,
, and Rt represents the time series data, trend, seasonal component, and random component, respectively. The following is a description of the equation.
The number of dengue cases in this study is denoted by . It is measured on a monthly period. Every year, the number of cases of dengue varies greatly. The numbers during the epidemic year could triple those of the period’s average. As a result, the pattern may be mistranslated. It is imperative to maintain a consistent annual number of dengue cases. As a result, we define a new parameter.
, the corrected dengue data are as follows.
where is a dengue case of the peak month of the year. We calculated that the dengue endemic would last six months, from January to December. By lessening the impact of outlier cases, the adjusted value enables us to examine the pattern of dengue incidence.
The variance of Yt can be described as follows:
The ratio of the variance of the component and the variance of the dataset was calculated as follows:
SARIMA.
An analysis model known as an ARIMA model forecasts potential outcomes via time series data. ARIMA (p, d, q) refers to an ARIMA model that is not seasonal. The number of autoregressive terms, p, is given by the nonnegative integers. The number of times the raw observations are differenced is denoted by d. In the prediction equation, q is the number of lags in the forecast errors. SARIMA (p, d, q)(P, D, Q)m, where m is the number of periods in each season and P, D, Q correspond to the autoregressive, differencing, and moving average terms for the seasonal component, respectively, is an extension of the ARIMA model that incorporates seasonality [55]. In our study, we chose the m value of a 6-month seasonal lag in the SARIMA model, indicating a potential semi-annual autocorrelation structure. This pattern suggests the possibility of sub-annual cycles in dengue incidence, a phenomenon reported in dengue-endemic regions with bimodal climate influences, where dengue peaks have been observed approximately six months apart due to dual rainy seasons [56,57]. Therefore, although SARIMA models with both 6- and 12-month seasonal components were considered, the final model selection prioritizes the 6-month lag to reflect the biologically and climatologically consistent annual dengue transmission cycle, aligning with findings from other tropical regions [58,59].
With the assumption that the causes of these events continue to exhibit the same behavior, these models employ nonseasonal differences, autoregressions, and moving average data from prior samples in addition to seasonal differences, autoregressions, and moving averages from prior periods to accurately predict the subsequent steps of a time series.
A seasonal ARIMA (SARIMA) model is particularly suitable for modeling dengue incidence due to its capacity to capture seasonal components that a conventional ARIMA model may overlook. Based on its structural complexity, the seasonal model can be categorized into two types: a simple additive model and a multiplicative seasonal model. The mathematical form of the simple additive seasonal model is expressed as:
Where, St, Tt, and It denote seasonal information, trend information, and random fluctuation information in the data, respectively.
To construct a SARIMA model for dengue case data, the non-seasonal components are identified first, followed by the seasonal components. A visual inspection of the time series plot of monthly dengue cases is useful to reveal any apparent seasonality pattern. A Box-Cox transformation may be applied to stabilize variance. To remove long-term trends and seasonal effects, first-order differencing and seasonal differencing are employed, respectively. An Augmented Dickey-Fuller (ADF) test is used to check for stationarity in the time series [60,61].
The mathematical formulation of SARIMA models can be generalized as described in Eq. (6).
In this equation, p, d, and q represent the nonseasonal order of autoregression (AR), differentiation, and moving average (MA), respectively. P, D, and Q represent the seasonal order of AR, differentiation, and MA, respectively. Moreover, represents the time series data in period t.
represents the Gaussian white noise process (random walk) in period t. B represents the backward shift operator
.
represents the seasonal difference, and
represents the nonseasonal difference. S represents the seasonal order.
Extreme Gradient Boosting (XGBoost).
Extreme Gradient Boosting (XGBoost) is an optimized and scalable implementation of the gradient boosting framework, designed for high performance in both accuracy and computational efficiency. It is particularly useful for handling complex and large-scale datasets, such as those involved in forecasting dengue cases. XGBoost efficiently evaluates the importance of input features and has demonstrated strong predictive power in various machine learning tasks [62,63]. XGBoost has since undergone significant refinement by the research community [61,64]. The underlying principle of boosting involves combining a series of weak learners (typically decision trees) to form a strong, high-accuracy predictive model. When dealing with complex or high-dimensional datasets, multiple iterations are often required to achieve acceptable predictive accuracy, an area where XGBoost excels [61]. XGBoost uses a regularized objective function that balances model fit and complexity, making it robust against overfitting. The objective function for XGBoost is defined as follows:
Where yi is the observed value, is the predicted value of the last iteration, xi is the feature vector, n is the sample size, ft is a new function that the model learns, Ω(ft) is the regularization term, which saves the model from complexity. l denotes the loss function, which calculates the difference between the label and the prediction in the previous phase, the new tree’s output [60,65].
Multivariate Poisson regression (MPR).
Since there is usually a lag of months between changes in meteorological and associated dengue, the goal of this analysis was to determine the best-lagged influence that different meteorological conditions had on dengue occurrence [66]. The interval between a single weather observation and a dengue case was used to define the time lag [67]. The monthly meteorological factors and DF cases were correlated for time lags ranging from zero to two months via Spearman’s rank correlation test. The interval between a single weather observation and a dengue case was used to define the time lag [67].
The following is an example of a basic multivariable Poisson regression model:
The model adjusted for first-order autocorrelation was as follows:
Here, the variables XTmean, XPrecipitation, XRHumidity, and XWspeed represent the mean temperature, amount of rainfall, relative humidity, and wind speed, respectively. All the meteorological variables were tested for multicollinearity with variance inflation factors (VIFs).
Although nonlinear methods can better capture complex relationships in climate-health data, we employed multivariate Poisson regression due to its interpretability and suitability for modeling count data like monthly dengue cases. This approach allowed us to estimate incidence rate ratios (IRRs) for specific meteorological variables, providing actionable epidemiological insights. We acknowledge that Poisson models assume a linear-log link and may not fully capture nonlinear interactions; future studies could integrate machine learning models or generalized additive models (GAMs) to better address these complexities. Thus, the choice of Poisson regression is often a practical and methodologically sound decision when the focus is on understanding variable relationships rather than prediction [50,51].
Statistical tools and packages
In this study, the R programming language (version 4.4.2) was employed to handle various statistical tasks. Descriptive statistics and regression model summaries were generated via the ‘gtsummary’ package, which provides easy-to-interpret, publication-ready tables of results. [68]. The ‘MASS’ package was used to conduct Poisson regression, fitting models for count data [69]. Time series analysis and forecasting were performed via the forecast and ‘dlnm’ packages, enabling decomposition and distributed lag models to understand temporal patterns [70]. For data visualization, the ‘ggplot2’ package was used to create sophisticated plots [71], while the ‘corrplot’ package helps in visualizing the correlation matrix effectively, uncovering hidden relationships among variables [72].
Results
Dengue cases and meteorological factors
There are large variations in the number of dengue cases in Bangladesh, with an average of 3,093 and a standard deviation (SD) of 10,960 from January 2008 to November 2024. Meteorological parameters showed a maximum temperature of 30.47°C (SD ± 3.05°C) and a low of 21.6°C (SD ± 5.2°C), with a wider range of 1.6–27.5°C. The mean temperature showed moderate variability, averaging 25.7°C (SD ± 4.1°C). Precipitation and humidity also vary considerably, while wind speed remains relatively stable (Table 1).
Temporal distribution of dengue cases
The highest peak number of dengue cases between January 2008 and November 2024 was found in August 2019 (52,636 cases) and in September 2023 (79,598 cases), along with 71,976 cases in October same year, both years experiencing significant peaks surpassing 40,000 cases (Fig 1). The boxplot shows that the maximum number of dengue cases is found from July to November, with September standing out as the peak month for most dengue cases in Bangladesh, and these months also contain outliers of the cases, which were later handled for adjusting the SARIMA model (Fig 2).
Decomposition
We created the adjusted data from the raw data by using Eqs. (1) and (2). Fig 3 shows the STL plot of the raw data. There are prominent high peaks in the data, reflecting significant events, whereas the seasonal component captures strong annual cycles with regular peaks, likely indicating periodic factors. The trend component shows a gradual increase followed by a decline, suggesting greater underlying influences on the data. The remainder of the component displays random fluctuations and spikes, which could represent unexpected events or noise after removing the seasonal and trend effects.
Autocorrelation and SARIMA
The autocorrelation (ACF) and partial autocorrelation (PACF) analyses for dengue cases from January 2008 to November 2024 revealed a strong seasonal pattern (Fig 4). The ACF plot shows significant positive autocorrelation at lag 1 (~1 month) and lag 2 (~2 months), with values around 0.8 and 0.5, respectively, indicating that dengue cases are highly correlated with cases from the preceding months. Additionally, moderate autocorrelation is observed around lags 10–12, suggesting possible annual seasonality. The PACF plot exhibits a large spike at lag 1 (~1 month) with a value above 0.5, followed by no significant spikes beyond lag 2, implying that the short-term correlation is primarily driven by the most recent monthly cases. The blue dashed lines in both plots indicate significance thresholds, reinforcing the presence of strong autocorrelations within the first few months.
SARIMA (Seasonal Autoregressive Integrated Moving Average) model identification involves determining the appropriate orders of autoregression (AR), differencing (I), and moving average (MA) terms for both seasonal and non-seasonal components of the time series. Based on the ACF and PACF plots for dengue cases from January 2008 to November 2024, a first-order differencing (d = 1) is needed due to the slow decay of the ACF, indicating non-stationarity. The strong spike at lag 1 in the PACF suggests an AR (1) process (p = 1 or 2), while the ACF pattern indicates a moving average component (q = 1 or 2). The first-order seasonal differencing (D = 1). The seasonal ACF and PACF patterns indicate the need for a seasonal autoregressive term (P = 1) and possibly a seasonal moving average term (Q = 0 or 1). The 6-month lag was constructed and determined based on autocorrelation (Fig 4) and the monthly distribution pattern of dengue cases (Fig 2) and based on the cumulative effect of environmental conditions on case occurrence. From the ACF and PACF result 14 SARIMA models can be constructed and the best model was selected by checking the best model selection criteria like mean absolute error (MAE), Root Mean Squared Error (RMSE), and L Jung (p-value) (p > 0.05), and the model selection criteria revealed the following results: Root Mean Squared Error (RMSE): 5203.34, and Mean Absolute Error (MAE) = 1649.56 and determined the best model which is SARIMA (2,1,2) (1,1,1) [6] (Table 2).
The prediction from 2024 to 2027 using the SARIMA (2,1,2) (1,1,1) [6] and XG-boost model, where both show periodic oscillations in dengue case numbers every six months (Fig 5a and 5b), indicating a clear seasonal pattern. The forecasted cases range between 10,000 and 20,000, with consistent peaks that suggest recurring outbreaks during specific times of the year. This seasonal regularity highlights the model’s ability to capture the cyclical nature of dengue transmission.
Correlation analysis of meteorological factors with dengue cases
Fig 6 shows the cross-correlation between pre-whitened meteorological variables and dengue cases at lags 0–12 months. Significant positive correlations were observed for temperature at lags 1–5 months, humidity at lags 1–4 months, and wind speed at lags 1–6 months, indicating that these factors may influence dengue incidence after a short delay. As the positive correlations mainly occur between 1–4 months and the dengue transmission period typically spans around 2 months, we selected 1- and 2-month lags for the Spearman correlation analysis.
Spearman’s rank correlation analysis (Fig 7) revealed that dengue cases had the strongest positive correlation with Precipitation (r = 0.37, p < 0.05), indicating that as rainfall increased, so did dengue cases. Temperature (r = 0.28, p < 0.05) and wind speed (r = 0.25, p < 0.05) also exhibited positive correlations with dengue cases, which also indicated the increase in dengue. The relationship between dengue cases and humidity (r = 0.18, p < 0.05) is weaker, but it is significant.
The temperature, precipitation, and humidity exhibited significant positive correlations with dengue cases at a one-month lag (Fig 8). Specifically, temperature (°C) lag1 showed a moderate positive correlation with dengue cases (r = 0.49, p < 0.001). Precipitation (mm) lag1 displayed a stronger correlation (r = 0.56, p < 0.001), while humidity (%) lag1 had a weaker positive association (r = 0.17, p < 0.001). Wind speed (m/s) lag1 also showed a moderate positive correlation (r = 0.47, p < 0.001). With a two-month lag (Fig 8), the associations strengthened. Temperature (°C) lag2 demonstrated a higher correlation with dengue cases (r = 0.60, p < 0.001), and precipitation (mm) lag2 maintained a strong positive relationship (r = 0.60, p < 0.001). Humidity (%) lag2 remained weakly associated (r = 0.13, p < 0.05), while wind speed (m/s) lag2 showed a stronger correlation (r = 0.65, p < 0.001). These findings indicate that temperature and precipitation play a crucial role in influencing dengue outbreaks with a time-lag effect.
Multivariable Poisson regression
The Poisson regression results indicate that various meteorological factors have significant effects on the cases of dengue (Table 3). An increase in temperature (°C) (IRR = 1.02, 95% CI: 1.02–1.02, p < 0.001) is associated with an increase of 2% cases of dengue, likely due to the acceleration of mosquito development and increased virus replication under warmer conditions. Precipitation (mm) (IRR = 1.00, 95% CI: 1.00–1.00, p < 0.001) showed a statistically significant but marginal effect at the per-millimeter level. This suggests that while very small changes in rainfall may have limited immediate impact, larger cumulative increases, such as over 10–100 mm or more, may still influence mosquito breeding and dengue transmission patterns. Additionally, the observed significance may reflect non-linear or threshold effects where moderate rainfall fosters breeding, but excessive rain can reduce larval survival. Humidity (%) (IRR = 1.25, 95% CI: 1.24–1.25, p < 0.001) increases dengue cases by 25%, likely because higher humidity enhances mosquito survival, feeding behavior, and overall vector activity. Wind speed (m/s) (IRR = 1.10, 95% CI: 1.09–1.10, p < 0.001) is associated with an increase of 10% dengue cases, indicating that moderate wind speeds may facilitate mosquito dispersal, potentially increasing the reach of infected vectors.
Multivariable Poisson regression with lagged
Table 4 presents the Poisson regression results with one-month and two-month lags to assess the influence of meteorological variables on monthly dengue cases in Bangladesh. In the one-month lag model, temperature showed a negative and statistically significant effect on dengue cases (IRR = 0.319, 95% CI: 0.316–0.322, p < 0.001), indicating that higher temperatures in the previous month were associated with lower cases. Precipitation also had a significant negative effect (IRR = 0.002, 95% CI: 0.002–0.002, p < 0.001), suggesting that increased rainfall may reduce dengue transmission, possibly by flushing mosquito breeding sites. Similarly, humidity (IRR = 0.193, 95% CI: 0.192–0.193, p < 0.001) and wind speed (IRR = 0.601, 95% CI: 0.596–0.606, p < 0.001) negatively impacted cases. In the two-month lag model, temperature was still negatively associated with dengue cases but with a weaker effect (IRR = 0.935, 95% CI: 0.930–0.940, p < 0.001). Precipitation (IRR = 0.001, 95% CI: 0.001–0.001, p < 0.001) and humidity (IRR = 0.106, 95% CI: 0.105–0.106, p < 0.001) continued to show significant negative effects. Wind speed also exhibited a negative association with cases (IRR = 0.933, 95% CI: 0.928–0.939, p < 0.001), indicating that stronger winds may disrupt mosquito activity and reduce transmission risk.
Model Comparisons with Predictors
Table 5 summarizes the predictive performance of three multivariate models, ARIMA, Poisson Regression, and XGBoost, using climatic variables to model dengue incidence in Bangladesh from 2008 to 2024, where RMSE is considered as the comparison factor. The ARIMA model demonstrated the best predictive performance with the lowest RMSE (5058.066) and was ranked 1st. It identified precipitation and wind speed as key climatic factors. The Poisson Regression model ranked 2nd with a moderate RMSE (8089.84), highlighting temperature, humidity, and wind speed as influential predictors. Conversely, XGBoost performed the poorest (RMSE: 23075.05), ranking 3rd, and identified temperature and wind speed as significant variables.
Discussion
Dengue development, transmission, and incidence are closely tied to climate. As a vector-borne disease, its spread heavily relies on the presence and abundance of its vector. Numerous studies have shown that ecological and climatic factors significantly affect the seasonal occurrence of the dengue virus [55,73–77]. In tropical and subtropical regions, climate change poses a severe threat to global health by enabling the spread of DF. Bangladesh’s total number of dengue cases has risen steadily since 2010, from 101,354 in 2019–321,179 confirmed cases in 2023. Hsan, et al. [78] reported that dengue outbreaks in 2019 were intensified by climate change, rapid unplanned urbanization, high population density, and a strained healthcare system. Inadequate preparedness and weak vector control further exacerbate the severity of epidemics. Furthermore, the record-breaking surge in 2023 was driven primarily by climate anomalies and systemic vulnerabilities. Unusually heavy monsoon rains and higher-than-normal temperatures increased mosquito survival and reproduction, while prolonged humidity extended the transmission season [79].
The dengue outbreak in 2023 was unusually severe, with incidence rates more than ten times higher than the average reported during the study period. This spike could lead to significant errors in predictive models, as it deviates considerably from the typical trend. The link between DF and climatic patterns remains poorly understood because of the complexity of the vector and host life cycles. The life cycle of the A. aegypti mosquito, the primary vector for dengue, is highly sensitive to climate conditions, which adds to the challenge of fully understanding this relationship [73,80]. However, previous research conducted in South Asia and Southeast Asia has examined the role of climate change in the spread of DF [55,73,81–83]. These studies identified temperature, precipitation, wind speed, and humidity as key climatic factors that may contribute to dengue outbreaks. They also demonstrated the consistent impact of certain climate conditions over time. In this research, we assess the effects of climate variables on dengue incidence, providing valuable insights for mitigating severe future outbreaks and enhancing preparedness efforts in Bangladesh.
The fluctuating trends of dengue fever demonstrate that meteorological factors substantially affect dengue occurrence in southern Thailand [84]. Our research indicates that dengue transmission in Bangladesh transpired from January to December throughout the years 2008–2024. The longest period of dengue transmission in Bangladesh occurred from July to November during the study period. These findings correspond with other studies conducted in southern Thailand [84], which indicated that the most extended dengue transmission season in the Gulf of Thailand occurred from June to September. This aligns with previous findings in Singapore [85], which indicated a rise in dengue incidence from June to October, perhaps because of the geographic and climatic similarities between Thailand and Singapore. Dengue incidence in Delhi, India, peaks in April and from June to October, which is similar to Bangladesh, likely because of the varying monsoon periods between Thailand and India [30]. Moreover, India’s peak temperatures may reach 45°C during the summer months of April, May, and June, hence influencing dengue transmission dynamics [30].
Temperature has been identified as a key predictor for the increase in dengue cases. In this study, a significant correlation was found between dengue incidence from 2008--2024 and factors such as temperature, precipitation, windspeed, and humidity. These findings align with those of a previous study by Pinto et al. [82], which showed that minimum and maximum temperatures are strong indicators for predicting dengue case increases on the basis of data from Singapore between 2000 and 2007.
Temperature is the most crucial climatic factor influencing the growth and spread of mosquito vectors, making it a potential predictor of dengue outbreaks [84]. It significantly impacts the entire life cycle and behavior of mosquitoes, including population density, biting rates, gonotrophic cycle durations, and vector size [86,87]. These findings align with earlier research [55,88] confirming that extreme temperatures negatively affect dengue vector development. As a result, temperature influences vector efficiency and the potential for an epidemic [89]. Given the clear link between temperature and dengue incidence, projected temperature changes could intensify dengue transmission in Bangladesh. A possible explanation is that the optimal breeding temperature range for A. aegypti mosquitoes is 20–35°C [74]. In our study, the minimum and mean temperatures ranged between 21°C and 30°C, providing ideal conditions for mosquito breeding and facilitating the transmission of dengue viruses.
Another key factor is humidity, which increases the number of dengue patients in Bangladesh. The correlation between dengue incidence and humidity was strongly positive. Similar results were also reported in previous studies [90,91]. This may be attributable to the influence of relative humidity on rainfall, as it indicates the degree of moisture saturation in the atmosphere. A reduction in temperature while the air is saturated can induce condensation, ultimately producing rain [75].
This study revealed that precipitation was associated with dengue cases. In our study, the lag 1 and lag 2 correlations were strongly correlated and were crucial influencing factors in dengue endemics. This outcome is supported by previous studies [92,93]. In Bangladesh, the rainy season lasts from April to September [94]. Rainfall is known to have both positive and negative effects on mosquito growth. On the one hand, light rainfall provides standing water, which is essential for mosquito breeding. On the other hand, excessive or excessive rainfall can disrupt mosquito development by destroying mosquito breeding sites or washing away larvae, limiting their growth potential [95]. In this research, we found that there was a significant increase from July to November, just before the dry season started. Moreover, the peak season for rain is April to June, when the rate is quite low because of extreme rainfall compared with that in later months [73]. This study also revealed that wind speed was weakly positively correlated with dengue incidence in Bangladesh, which is similar to the findings of a previous study [73,96].
In our study, temperature, precipitation, humidity, and wind speed had significant effects on dengue cases between 2008 and 2024. Meteorological factors mostly affect the increase in dengue cases worldwide. The relationships between weather factors and dengue cases may vary across different periods, especially with increasing global warming and urbanization [97]. [98] reported that minimum temperature, humidity, and rainfall were key factors influencing dengue cases, whereas maximum temperature and maximum humidity were not significant in the Mekong Delta area in Vietnam.
In our study, the autocorrelation plot revealed significant positive autocorrelation at the 1-month and 2-month intervals for dengue cases. The virus typically has an incubation period of seven to fourteen days, during which it progresses from an infectious state to an active viral disease. The stability of the dengue virus often remains in endemic equilibrium, influenced by these delay times. Consequently, vector-borne illnesses may persist in carriers for up to one to two weeks [99–101]. The autocorrelation function (ACF) plot was truncated at lag 4, which was parameterized to a moving average (MA) model, with no seasonal lag observed in Nakhon Si Thammarat [102]. In contrast, Islam et al. [50] reported that after 12 weeks, the autocorrelation of DF cases gradually decreased to negligible levels. Peaks in the partial autocorrelation were noted at weeks 1 and 4. Chumpu et al. [103] also reported that 1-week lag cases had the highest correlation among 65 provinces in Thailand.
In our study, the best-fitting ARIMA model was identified as SARIMA (2,1,2) (1,1,1) [6]. Similarly, Gui et al. [97] determined that ARIMA (1,1,0) with first-order differencing and one autoregressive term was the optimal model for capturing the trend and autoregressive properties of the dengue time series. In Bangkok, Polwiang [55] reported that the best-fitting model was SARIMA (1,0,2)(1,1,2) [12], which has a strong seasonal component (1,1,2) combined with a nonseasonal component (1,0,2), indicating a mixed model. For overall dengue incidence in northeastern Thailand, Silawan et al. [91] fitted a seasonal ARIMA (2,1,0) (0,1,1) [12] model to data spanning 1996--2003. Additionally, Earnest et al. [104] reported that ARIMA (3,1,0) provided the lowest mean absolute percentage error (MAPE) of 19.86, indicating its effectiveness in their study. The validity of our selected SARIMA (2,1,2) (1,1,1) [6] model was assessed through standard model diagnostic checks, including residual analysis and evaluation metrics such as the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Ljung–Box Q-statistics. These diagnostics confirmed that the model adequately captured the temporal structure of the data without significant autocorrelation in the residuals. However, it is important to note that the model’s predictive validity may be limited by the exclusion of non-climatic covariates such as socio-economic status and urbanization, which are known to influence disease transmission dynamics. Although the SARIMA framework effectively models seasonal and autoregressive patterns, the absence of these explanatory variables may reduce the generalizability of forecasts, particularly at the national level. Future work should aim to incorporate such covariates and explore hybrid models that can better account for multifactorial drivers of disease incidence.
Research has shown that temperature, relative humidity, and rainfall all affect the spread of DF, either directly or indirectly [30,59]. Temperature affects the extrinsic incubation period, reproduction rates, and vector population growth [105,106]. Research has indicated that the confluence of temperature and humidity has a major effect on the quantity of blood consumed and increases the vector survival rate [107]. Numerous studies have demonstrated that the connection between climatic conditions and dengue transmission varies over time and across the Asia-Pacific region [108,109]. The significance of the lag time of climate factors has also been emphasized by numerous studies [110,111]. For example, there was a substantial positive correlation with daily rainfall at a lag of 10 weeks, the minimum temperature at a lag of 1–3 months, the maximum temperature at a lag of 1–4 months, and the relative humidity at a lag of 1–3 months in Taiwan [112].
According to our research, the incidence rates of minimum temperature, relative humidity, and rainfall in Bangladesh were significantly related to one- and two-month lags, respectively. This is also in line with a study conducted in Taiwan [112]. Understanding how climate factors affect Aedes mosquito development, maturation, and survival (approximately 7–9 days from egg to adult), as well as how long DENV takes to incubate intrinsically in humans (4–6 days) [113] and extrinsically in the vector (10 days) [114,115] is important. The duration and humidity of dormancy in the environment can impact larval survival, developmental rates, and the production of relatively small adults. Eggs are resistant to desiccation and can endure months of dormancy [116].
In our analysis, we observed a contrasting pattern between non-lagged and lagged Poisson models. While non-lagged models showed positive associations between climatic variables such as temperature and disease incidence (e.g., IRR = 1.02), the lagged models particularly at lag1 and lag2, exhibited IRRs below 1 (e.g., temperature lag1 IRR = 0.319). To investigate whether multicollinearity might explain this inversion, we conducted a Variance Inflation Factor (VIF) analysis. All VIF values were below 1.3 for both lag1 and lag2 models (e.g., Mean_temp_lag1 = 1.259, Precipitation_lag1 = 1.120; Mean_temp_lag2 = 1.201, Precipitation_lag2 = 1.134), indicating no significant multicollinearity. Therefore, this inversion is unlikely due to collinearity or model mis-specification.
Instead, the contradictory IRRs likely reflect true delayed effects of environmental variables. Climatic influences on disease outcomes often operate through complex, time-dependent mechanisms involving vector biology, pathogen incubation, or delayed behavioral responses. For instance, while elevated temperature may enhance immediate transmission risk, it could also lead to subsequent behavioral adaptations (e.g., increased hydration, staying indoors) or changes in vector dynamics that reduce cases after a delay. Similar inverted lag effects have been reported in earlier time series studies. Imai and Hashizume [117] emphasized the importance of considering such temporal dynamics in weather-disease models. Kakarla et al. [118] also showed that climatic variables influence dengue burden differently across lag periods. Likewise, Islam et al. [119] found that childhood diarrhea risks in Bangladesh varied with temperature across different lags, supporting our findings.
One limitation of this study is the exclusion of important socioeconomic and urbanization-related factors that could significantly influence dengue transmission. Variables such as population density, income levels, housing conditions, access to healthcare, and sanitation infrastructure play a crucial role in shaping disease dynamics. Additionally, factors like land use patterns, urban expansion, and climate variations interact with mosquito breeding habitats, further affecting dengue incidence. Incorporating these parameters in future studies could enhance the accuracy and comprehensiveness of dengue risk assessments and predictive models. These covariates could significantly enhance model accuracy and generalizability, particularly for a national-level forecasting framework.
This study has significant ramifications, although it also has considerable shortcomings. The dataset, intended to evaluate the influence of climate variables on dengue incidence in Bangladesh, contains several missing or incomplete records, reflecting inadequate data collection and reporting practices. This may arise from insufficient monitoring mechanisms or a deficiency in awareness among reporting authorities. Consequently, deficiencies in population data may result in biases in estimates due to underreporting or overreporting, given that the dataset encompasses both confirmed and suspected dengue cases.
Conclusion
This study aims to find the relationship between the seasonal dynamics and dengue in Bangladesh from January 2008 to November 2024. Meteorological variables, temperature, humidity, and wind speed, were found to influence dengue outbreaks. Our findings emphasize the need to integrate real-time meteorological data and consider urbanization and socioeconomic variables in future models. Early warning systems informed by seasonal climate forecasts could guide timely vector control interventions and improve public health response. Strengthening collaboration among meteorological services, health departments, and vector control programs will be essential to operationalize these strategies in the context of a changing climate.
Supporting information
S1 Table. Meteorological data and Dengue cases according to date.
https://doi.org/10.1371/journal.pone.0323238.s001
(XLSX)
Acknowledgments
The authors also thank all the members of the Association of Coding, Technology, and Genomics (ACTG), Sher-e-Bangla Agricultural University (SAU), Dhaka, Bangladesh, who helped to conduct the research. We also appreciate the Directorate General of Public Health and Services, Government of Bangladesh, for providing the required data and access to other necessary information.
References
- 1. Lee SH, Nam KW, Jeong JY, Yoo SJ, Koh Y-S, Lee S, et al. The effects of climate change and globalization on mosquito vectors: evidence from Jeju Island, South Korea on the potential for Asian tiger mosquito (Aedes albopictus) influxes and survival from Vietnam rather than Japan. PLoS One. 2013;8(7):e68512. pmid:23894312
- 2. Simo FBN, Bigna JJ, Kenmoe S, Ndangang MS, Temfack E, Moundipa PF, et al. Dengue virus infection in people residing in Africa: a systematic review and meta-analysis of prevalence studies. Sci Rep. 2019;9(1):13626. pmid:31541167
- 3. Ong J, Aik J, Ng LC. Short report: adult aedes abundance and risk of dengue transmission. PLoS Negl Trop Dis. 2021;15(6):e0009475. pmid:34081695
- 4. Jayaraj VJ, Avoi R, Gopalakrishnan N, Raja DB, Umasa Y. Developing a dengue prediction model based on climate in Tawau, Malaysia. Acta Trop. 2019;197:105055. pmid:31185224
- 5. Murray NEA, Quam MB, Wilder-Smith A. Epidemiology of dengue: past, present and future prospects. Clin Epidemiol. 2013;5:299–309. pmid:23990732
- 6.
WHO. Global strategy for dengue prevention and control, 2012–2020. Available from https://www.who.int/publications/i/item/9789241504034
- 7. Cheong YL, Burkart K, Leitão PJ, Lakes T. Assessing weather effects on dengue disease in Malaysia. Int J Environ Res Public Health. 2013;10(12):6319–34. pmid:24287855
- 8. Barbazan P, Guiserix M, Boonyuan W, Tuntaprasart W, Pontier D, Gonzalez J-P. Modelling the effect of temperature on transmission of dengue. Med Vet Entomol. 2010;24(1):66–73. pmid:20377733
- 9. Stewart Ibarra AM, Ryan SJ, Beltrán E, Mejía R, Silva M, Muñoz A. Dengue vector dynamics (Aedes aegypti) influenced by climate and social factors in Ecuador: implications for targeted control. PLoS One. 2013;8(11):e78263. pmid:24324542
- 10. Liu QY. Impact of climate change on vector-borne diseases and related response strategies in China: Major research findings and recommendations for future research. Chin J Vector Biol Control. 2021;32:1–11.
- 11. Gao J, Zhang H-D, Guo X-X, Xing D, Dong Y-D, Lan C-J, et al. Dispersal patterns and population genetic structure of Aedes albopictus (Diptera: Culicidae) in three different climatic regions of China. Parasit Vectors. 2021;14(1):12. pmid:33407824
- 12. Li J, Zhu G, Zhou H, Tang J, Cao J. Effect of different temperatures on development of Aedes albopictus. Zhongguo Xue Xi Chong Bing Fang Zhi Za Zhi. 2015;27(1):59–61. pmid:26094417
- 13. Kuhlisch C, Kampen H, Walther D. The Asian tiger mosquito Aedes albopictus (Diptera: Culicidae) in Central Germany: surveillance in its northernmost distribution area. Acta Trop. 2018;188:78–85. pmid:30145257
- 14. Little E, Bajwa W, Shaman J. Local environmental and meteorological conditions influencing the invasive mosquito Ae. albopictus and arbovirus transmission risk in New York City. PLoS Negl Trop Dis. 2017;11(8):e0005828. pmid:28832586
- 15. Armstrong PM, Andreadis TG, Shepard JJ, Thomas MC. Northern range expansion of the Asian tiger mosquito (Aedes albopictus): Analysis of mosquito data from Connecticut, USA. PLoS Negl Trop Dis. 2017;11(5):e0005623. pmid:28545111
- 16. Jia P, Chen X, Chen J, Lu L, Liu Q, Tan X. How does the dengue vector mosquito Aedes albopictus respond to global warming? Parasit Vectors. 2017;10(1):140. pmid:28284225
- 17. Githeko AK, Lindsay SW, Confalonieri UE, Patz JA. Climate change and vector-borne diseases: a regional analysis. Bull World Health Organ. 2000;78(9):1136–47. pmid:11019462
- 18. Hopp MJ, Foley JA. Global-scale relationships between climate and the dengue fever Vector, Aedes Aegypti. Climatic Change. 2001;48(2–3):441–63.
- 19. Woodruff R, McMichael T. Climate change and human health: all affected, but some more than others. Soc Alt 2004;23(4):17.
- 20. Jetten TH, Focks DA. Potential changes in the distribution of dengue transmission under climate warming. Am J Trop Med Hyg. 1997;57(3):285–97. pmid:9311638
- 21. Patz JA, Reisen WK. Immunology, climate change and vector-borne diseases. Trends Immunol. 2001;22(4):171–2. pmid:11274908
- 22. Beebe NW, Cooper RD, Mottram P, Sweeney AW. Australia’s dengue risk driven by human adaptation to climate change. PLoS Negl Trop Dis. 2009;3(5):e429. pmid:19415109
- 23. Sharmin S, Viennet E, Glass K, Harley D. The emergence of dengue in Bangladesh: epidemiology, challenges and future disease risk. Trans R Soc Trop Med Hyg. 2015;109(10):619–27. pmid:26333430
- 24. Dejnirattisai W, Jumnainsong A, Onsirisakul N, Fitton P, Vasanawathana S, Limpitikul W, et al. Cross-reacting antibodies enhance dengue virus infection in humans. Science. 2010;328(5979):745–8. pmid:20448183
- 25. Katzelnick LC, Gresh L, Halloran ME, Mercado JC, Kuan G, Gordon A, et al. Antibody-dependent enhancement of severe dengue disease in humans. Science. 2017;358(6365):929–32. pmid:29097492
- 26. Russell PK, Buescher EL, McCown JM, Ordõnez J. Recovery of dengue viruses from patients during epidemics in Puerto Rico and East Pakistan. Am J Trop Med Hyg. 1966;15(4):573–9. pmid:4957424
- 27. Hossain K, Chowdhury S, Shanta IS, Hossain MS, Ghosh PK, Alam MS. Spatio-temporal patterns of dengue in Bangladesh during 2019 to 2023: Implications for targeted control strategies. PLoS Negl Trop Dis. 2024;18(9):e0012503. pmid:39302980
- 28. Naher S, Rabbi F, Hossain MM, Banik R, Pervez S, Boitchi AB. Forecasting the incidence of dengue in Bangladesh-Application of time series model. Health Sci Rep. 2022;5(4):e666. pmid:35702512
- 29. Shahid S, Harun SB, Katimon A. Changes in diurnal temperature range in Bangladesh during the time period 1961–2008. Atmospheric Res. 2012;118:260–70.
- 30. Hii YL, Zhu H, Ng N, Ng LC, Rocklöv J. Forecast of dengue incidence using temperature and rainfall. PLoS Negl Trop Dis. 2012;6(11):e1908. pmid:23209852
- 31. Shirin T, Muraduzzaman AKM, Alam AN, Sultana S, Siddiqua M, Khan MH, et al. Largest dengue outbreak of the decade with high fatality may be due to reemergence of DEN-3 serotype in Dhaka, Bangladesh, necessitating immediate public health attention. New Microbes New Infect. 2019;29:100511. pmid:30937172
- 32. Briët OJT, Vounatsou P, Gunawardena DM, Galappaththy GNL, Amerasinghe PH. Models for short term malaria prediction in Sri Lanka. Malar J. 2008;7:76. pmid:18460204
- 33. Ch S, Sohani SK, Kumar D, Malik A, Chahar BR, Nema AK, et al. A Support Vector Machine-Firefly Algorithm based forecasting model to determine malaria transmission. Neurocomputing. 2014;129:279–88.
- 34. Wongkoon S, Jaroensutasinee M, Jaroensutasinee K. Development of temporal modeling for prediction of dengue infection in northeastern Thailand. Asian Pac J Trop Med. 2012;5(3):249–52. pmid:22305794
- 35. Shi B, Liu J, Zhou X-N, Yang G-J. Inferring Plasmodium vivax transmission networks from tempo-spatial surveillance data. PLoS Negl Trop Dis. 2014;8(2):e2682. pmid:24516684
- 36.
Liang F, Shi B, Liu J, Chen L. Forecasting disease incidences using state space model. Paper presented at 2nd International Workshop on Pattern Recognition for Healthcare Analytics collocated with 22nd International Conference on Pattern Recognition. 2014.
- 37. Adebiyi AA, Adewumi AO, Ayo CK. Comparison of ARIMA and Artificial Neural Networks Models for Stock Price Prediction. J Appl Math. 2014;2014:1–7.
- 38. Kontopoulou VI, Panagopoulos AD, Kakkos I, Matsopoulos GK. A review of ARIMA vs. machine learning approaches for time series forecasting in data driven networks. Future Internet. 2023;15(8):255.
- 39. Shahriar SA, Kayes I, Hasan K, Hasan M, Islam R, Awang NR, et al. Potential of arima-ann, arima-svm, DT and catboost for atmospheric PM2.5 forecasting in Bangladesh. Atmosphere. 2021;12(1):100.
- 40. Wang L, Zou H, Su J, Li L, Chaudhry S. An ARIMA‐ANN Hybrid Model for Time Series Forecasting. Syst Res Behav Sci. 2013;30(3):244–59.
- 41. Chowdhury AH. Comparison of deep learning and gradient boosting: ANN Versus XGBoost for climate‐based dengue prediction in Bangladesh. Health Sci Rep. 2025;8(4).
- 42. Ong SQ, Isawasan P, Ngesom AMM, Shahar H, Lasim AM, Nair G. Predicting dengue transmission rates by comparing different machine learning models with vector indices and meteorological data. Sci Rep. 2023;13(1):19129. pmid:37926755
- 43.
Agarwala S, Ali MS, Quanita AT, Sajid ASM, Srizon AY, Esha NT, et al. Precision death forecasting of dengue outbreaks in bangladesh by blending statistical and machine learning models. IEEE International Conference on Power, Electrical, Electronics and Industrial Applications (PEEIACON). IEEE; 2024. pp. 981–6.
- 44. Maulana H, Sari AS. Analysis of dengue fever spread prediction using ensemble learning approach with xgboost and random. Instal: Jurnal Komputer. 2024;16(03):388–99.
- 45.
Syfullah MK, Ali MS, Oishy AM, Hossain MS. Towards early dengue diagnosis in Bangladesh: A non-invasive prediction model based on symptoms and local trends. IEEE International Conference on Power, Electrical, Electronics and Industrial Applications (PEEIACON), 12-13 Sept. 2024. 2024. pp. 833–8.
- 46. Akter T, Islam MdT, Hossain MdF, Ullah MS. A comparative study between time series and machine learning technique to predict dengue fever in Dhaka city. Discrete Dynamics Nat Soc. 2024;2024(1).
- 47. Hasan MN, Khalil I, Chowdhury MAB, Rahman M, Asaduzzaman M, Billah M, et al. Two decades of endemic dengue in Bangladesh (2000-2022): trends, seasonality, and impact of temperature and rainfall patterns on transmission dynamics. J Med Entomol. 2024;61(2):345–53. pmid:38253990
- 48.
Mohammed N, Rahman MZ. Forecasting dengue incidence in bangladesh using seasonal arima model, a time series analysis. In: International Conference on Machine Intelligence and Emerging Technologies. Springer: 2022. pp. 589–98.
- 49.
Meem SM, Hossain MT, Chowdhury JK, Ullah Miah MS, Monir MF. Understanding the Dynamics of Dengue in Bangladesh: EDA, Climate Correlation, and Predictive Modeling. TENCON 2023 - 2023 IEEE Region 10 Conference (TENCON). IEEE; 2023. pp. 1309–14. doi: https://doi.org/10.1109/tencon58879.2023.10322490
- 50. Islam MA, Hasan MN, Tiwari A, Raju MAW, Jannat F, Sangkham S, et al. Correlation of Dengue and Meteorological Factors in Bangladesh: A Public Health Concern. Int J Environ Res Public Health. 2023;20(6):5152. pmid:36982061
- 51. Hossain S, Islam MM, Hasan MA, Chowdhury PB, Easty IA, Tusar MK, et al. Association of climate factors with dengue incidence in Bangladesh, Dhaka City: A count regression approach. Heliyon. 2023;9(5):e16053. pmid:37215791
- 52.
IECDR. Dengue Situation Update. Institute of epidemiology, disease control and research (iedcr). 2020. Available from: https://www.old.iedcr.gov.bd/
- 53.
IECDR. Institute of epidemiology, disease control and research. 2024. Available from: https://iedcr.portal.gov.bd/site/page/45aea1fa-5756-4feb-8d09-f0da895a3baa/
- 54. Haider N, Hasan MN, Khalil I, Tonge D, Hegde S, Chowdhury MAB, et al. The 2022 dengue outbreak in Bangladesh: hypotheses for the late resurgence of cases and fatalities. J Med Entomol. 2023;60(4):847–52. pmid:37202843
- 55. Polwiang S. The time series seasonal patterns of dengue fever and associated weather variables in Bangkok (2003-2017). BMC Infect Dis. 2020;20(1):208. pmid:32164548
- 56. Do TTT, Martens P, Luu NH, Wright P, Choisy M. Climatic-driven seasonality of emerging dengue fever in Hanoi, Vietnam. BMC Public Health. 2014;14:1078. pmid:25323458
- 57. Othman M, Indawati R, Suleiman AA, Qomaruddin MB, Sokkalingam R. Model forecasting development for dengue fever incidence in surabaya city using time series analysis. Processes. 2022;10(11):2454.
- 58. Descloux E, Mangeas M, Menkes CE, Lengaigne M, Leroy A, Tehei T, et al. Climate-based models for understanding and forecasting dengue epidemics. PLoS Negl Trop Dis. 2012;6(2):e1470. pmid:22348154
- 59. Gharbi M, Quenel P, Gustave J, Cassadou S, La Ruche G, Girdary L, et al. Time series analysis of dengue incidence in Guadeloupe, French West Indies: forecasting models using climate variables as predictors. BMC Infect Dis. 2011;11:166. pmid:21658238
- 60. Lv C-X, An S-Y, Qiao B-J, Wu W. Time series analysis of hemorrhagic fever with renal syndrome in mainland China by using an XGBoost forecasting model. BMC Infect Dis. 2021;21(1):839. pmid:34412581
- 61. Wang T, Liu J, Zhou Y, Cui F, Huang Z, Wang L, et al. Prevalence of hemorrhagic fever with renal syndrome in yiyuan county, China, 2005–2014. BMC Infect Dis. 2015;16:1–7.
- 62. Zheng Y, Zhu Y, Ji M, Wang R, Liu X, Zhang M, et al. A Learning-Based Model to Evaluate Hospitalization Priority in COVID-19 Pandemics. Patterns (N Y). 2020;1(6):100092. pmid:32838344
- 63. Hu CA, Chen CM, Fang YC, Liang SJ, Wang HC, Fang WF, et al. Using a machine learning approach to predict mortality in critically ill influenza patients: a cross-sectional retrospective multicentre study in Taiwan. BMJ Open. 2020;10(2):e033898.
- 64. Li W, Yin Y, Quan X, Zhang H. Gene Expression Value Prediction Based on XGBoost Algorithm. Front Genet. 2019;10:1077. pmid:31781160
- 65. Nishio M, Nishizawa M, Sugiyama O, Kojima R, Yakami M, Kuroda T, et al. Computer-aided diagnosis of lung nodule using gradient tree boosting and Bayesian optimization. PLoS One. 2018;13(4):e0195875. pmid:29672639
- 66. Choi Y, Tang CS, McIver L, Hashizume M, Chan V, Abeyasinghe RR, et al. Effects of weather factors on dengue fever incidence and implications for interventions in Cambodia. BMC Public Health. 2016;16:241. pmid:26955944
- 67.
Chatfield C. The analysis of time series: Theory and practice. Springer; 2013.
- 68. Sjoberg DD, Whiting K, Curry M, Lavery JA, Larmarange J. Reproducible summary tables with the gtsummary package. The R J. 2021;13(1):570–80.
- 69.
Venables W, Ripley B. Modern applied statistics with S. New York: Springer; 2002.
- 70. Gasparrini A. Distributed lag linear and non-linear models in R: The Package dlnm. J Stat Softw. 2011;43(8):1–20. pmid:22003319
- 71. Wickham H. Ggplot2. Wiley Interdiscip Rev: Comput Stat. 2011;3(2):180–5.
- 72. Hahsler M, Hornik K, Buchta C. Getting things in order: an introduction to theRPackageseriation. J Stat Soft. 2008;25(3).
- 73. Islam S, Haque CE, Hossain S, Hanesiak J. Climate variability, dengue vector abundance and dengue fever cases in Dhaka, Bangladesh: a time-series study. Atmosphere. 2021;12(7):905.
- 74. Navarro Valencia V, Díaz Y, Pascale JM, Boni MF, Sanchez-Galan JE. Assessing the effect of climate variables on the incidence of dengue cases in the metropolitan region of Panama City. Int J Environ Res Public Health. 2021;18(22):12108. pmid:34831862
- 75. Sriklin T, Kajornkasirat S, Puttinaovarat S. Dengue transmission mapping with weather-based predictive model in three southernmost provinces of Thailand. Sustainability. 2021;13(12):6754.
- 76. Lu L, Lin H, Tian L, Yang W, Sun J, Liu Q. Time series analysis of dengue fever and weather in Guangzhou, China. BMC Public Health. 2009;9:395. pmid:19860867
- 77. Gui H, Gwee S, Koh J, Pang J. Weather factors associated with reduced risk of dengue transmission in an urbanized tropical city. Int J Environ Res Public Health. 2021;19(1):339. pmid:35010600
- 78. Hsan K, Hossain MM, Sarwar MS, Wilder-Smith A, Gozal D. Unprecedented rise in dengue outbreaks in Bangladesh. Lancet Infect Dis. 2019;19(12):1287. pmid:31782396
- 79. Hasan MN, Rahman M, Uddin M, Ashrafi SAA, Rahman KM, Paul KK, et al. The 2023 fatal dengue outbreak in Bangladesh highlights a paradigm shift of geographical distribution of cases. Epidemiol Infect. 2025;153:e3. pmid:39763239
- 80. Gubler DJ, Reiter P, Ebi KL, Yap W, Nasci R, Patz JA. Climate variability and change in the United States: potential impacts on vector- and rodent-borne diseases. Environ Health Perspect. 2001;109 Suppl 2(Suppl 2):223–33. pmid:11359689
- 81. Karim MN, Munshi SU, Anwar N, Alam MS. Climatic factors influencing dengue cases in Dhaka city: a model for dengue prediction. Indian J Med Res. 2012;136(1):32–9. pmid:22885261
- 82. Pinto E, Coelho M, Oliver L, Massad E. The influence of climate variables on dengue in Singapore. Int J Environ Health Res. 2011;21(6):415–26. pmid:21557124
- 83. Aswi A, Cramb S, Duncan E, Hu W, White G, Mengersen K. Climate variability and dengue fever in Makassar, Indonesia: Bayesian spatio-temporal modelling. Spat Spatiotemporal Epidemiol. 2020;33:100335. pmid:32370940
- 84. Wongkoon S, Jaroensutasinee M, Jaroensutasinee K. Spatio-temporal climate-based model of dengue infection in Southern, Thailand. Trop Biomed. 2016;33(1):55–70. pmid:33579141
- 85. Sahay S. Climatic variability and dengue risk in urban environment of Delhi (India). Urban Climate. 2018;24:863–74.
- 86. Li Y, Dou Q, Lu Y, Xiang H, Yu X, Liu S. Effects of ambient temperature and precipitation on the risk of dengue fever: a systematic review and updated meta-analysis. Environ Res. 2020;191:110043. pmid:32810500
- 87. Rueda LM, Patel KJ, Axtell RC, Stinner RE. Temperature-dependent development and survival rates of Culex quinquefasciatus and Aedes aegypti (Diptera: Culicidae). J Med Entomol. 1990;27(5):892–8. pmid:2231624
- 88. Wongkoon S, Jaroensutasinee M, Jaroensutasinee K. Weather factors influencing the occurrence of dengue fever in Nakhon Si Thammarat, Thailand. Trop Biomed. 2013;30(4):631–41. pmid:24522133
- 89. Bangs MJ, Larasati RP, Corwin AL, Wuryadi S, Jakarta I. Climatic factors associated with epidemic dengue in Palembang, Indonesia: implications of short-term meteorological events on virus transmission. Southeast Asian J Trop Med Public Health. 2006;37(6):1103. pmid:17333762
- 90. Thammapalo S, Chongsuwiwatwong V, McNeil D, Geater A. The climatic factors influencing the occurrence of dengue hemorrhagic fever in Thailand. Southeast Asian J Trop Med Public Health. 2005;36(1):191–6. pmid:15906666
- 91. Silawan T, Singhasivanon P, Kaewkungwal J, Nimmanitya S, Suwonkerd W. Temporal patterns and forecast of dengue infection in Northeastern Thailand. Southeast Asian J Trop Med Public Health. 2008;39(1):90–8. pmid:18567447
- 92. Chen T-HK, Chen VY-J, Wen T-H. Revisiting the role of rainfall variability and its interactive effects with the built environment in urban dengue outbreaks. Appl Geogr. 2018;101:14–22.
- 93. Fakhruddin M, Putra PS, Wijaya KP, Sopaheluwakan A, Satyaningsih R, Komalasari KE, et al. Assessing the interplay between dengue incidence and weather in Jakarta via a clustering integrated multiple regression model. Ecol Complex. 2019;39:100768.
- 94.
CCAP. Climate change knowledge portal: Current Climate - Climatolog. Available from: https://climateknowledgeportal.worldbank.org/country/bangladesh/climate-data-historical#:~:text=The%20warmest%20months%20coincide%20with,mm)%20of%20rainfall%20per%20year
- 95. Li C, Lim T, Han L, Fang R. Rainfall, abundance of aedes aegypti and dengue infection in Selangor, Malaysia. The Southeast Asian J Trop Med Public Health. 1985;16(4):560–8. pmid:3835698
- 96.
Lemon SM, Sparling PF, Hamburg MA, Relman DA, Choffnes ER, Mack A. Vector-borne diseases: Understanding the environmental, human health, and ecological connections. National Academies Press; 2008. pp. 40.
- 97. Gui H, Gwee S, Koh J, Pang J. Weather factors associated with reduced risk of dengue transmission in an urbanized tropical city. Int J Environ Res Public Health. 2021;19(1):339. pmid:35010600
- 98. Phung D, Huang C, Rutherford S, Chu C, Wang X, Nguyen M, et al. Identification of the prediction model for dengue incidence in Can Tho city, a Mekong Delta area in Vietnam. Acta Trop. 2015;141(Pt A):88–96. pmid:25447266
- 99. Chen CS, Min LC, Pin CC, Chou CH, You SH, Cheng YH. Lagged temperature effect with mosquito transmission potential explains dengue variability in southern Taiwan: Insights from a statistical analysis. Sci Total Environ. 2010;408(19):4069–75. pmid:20542536
- 100. Wang Z, Chan HM, Hibberd ML, Lee GKK. Delayed Effects of Climate Variables on Incidence of Dengue in Singapore during 2000-2010. APCBEE Procedia. 2012;1:22–6.
- 101.
Cheong YL. Dengue disease in Malaysia. Vulnerability mapping and environmental risk assessment. Humboldt University of Berlin; 2015.
- 102. Sintupachee S, Onuma R, Manit P, Suppawan P. Development of the univariate time series model for forecasting dengue hemorrhagic fever cases in Nakhon Si Thammarat. MJPHM. 2023;23(2):63–72.
- 103. Chumpu R, Khamsemanan N, Nattee C. The association between dengue incidences and provincial-level weather variables in Thailand from 2001 to 2014. PLoS One. 2019;14(12):e0226945. pmid:31877191
- 104. Earnest A, Tan SB, Wilder-Smith A, Machin D. Comparing statistical models to predict dengue fever notifications. Comput Math Methods Med. 2012;2012:758674. pmid:22481978
- 105. Farjana T, Tuno N. Effect of body size on multiple blood feeding and egg retention of Aedes aegypti (L.) and Aedes albopictus (Skuse) (Diptera: Culicidae). Jap J Sanit Zool. 2012;63(2):123–31.
- 106. Xiao F-Z, Zhang Y, Deng Y-Q, He S, Xie H-G, Zhou X-N, et al. The effect of temperature on the extrinsic incubation period and infection rate of dengue virus serotype 2 infection in Aedes albopictus. Arch Virol. 2014;159(11):3053–7. pmid:24990415
- 107. Thu HM, Aye KM, Thein S. The effect of temperature and humidity on dengue virus propagation in Aedes aegypti mosquitos. Southeast Asian J Trop Med Public Health. 1998;29(2):280–4. pmid:9886113
- 108. Banu S, Hu W, Hurst C, Tong S. Dengue transmission in the Asia-Pacific region: impact of climate change and socio-environmental factors. Trop Med Int Health. 2011;16(5):598–607. pmid:21320241
- 109. Naish S, Dale P, Mackenzie JS, McBride J, Mengersen K, Tong S. Climate change and dengue: a critical and systematic review of quantitative modelling approaches. BMC Infect Dis. 2014;14:167. pmid:24669859
- 110. Sang S, Yin W, Bi P, Zhang H, Wang C, Liu X, et al. Predicting local dengue transmission in Guangzhou, China, through the influence of imported cases, mosquito density and climate variability. PLoS One. 2014;9(7):e102755. pmid:25019967
- 111. Sang S, Gu S, Bi P, Yang W, Yang Z, Xu L, et al. Predicting unprecedented dengue outbreak using imported cases and climatic factors in Guangzhou, 2014. PLoS Negl Trop Dis. 2015;9(5):e0003808. pmid:26020627
- 112. Chen M-J, Lin C-Y, Wu Y-T, Wu P-C, Lung S-C, Su H-J. Effects of extreme precipitation to the distribution of infectious diseases in Taiwan, 1994-2008. PLoS One. 2012;7(6):e34651. pmid:22737206
- 113. Guzman MG, Harris E. Dengue. The Lancet. 2015;385(9966):453–65.
- 114. Scott TW, Morrison AC. Vector dynamics and transmission of dengue virus: implications for dengue surveillance and prevention strategies: vector dynamics and dengue prevention. Dengue virus. 2010;338:115–28.
- 115.
CDC. Dengue and the aedes aegypti mosquito. 2012. Available from: https://stacks.cdc.gov/view/cdc/31032
- 116.
Perez MH. Aedes aegypti pharate first instar aseasonal quiescence cues anticipatory plasticity with implications for urban vector ecology and control. FIU Electronic Theses and Dissertations, 913. 2013.
- 117. Imai C, Armstrong B, Chalabi Z, Mangtani P, Hashizume M. Time series regression model for infectious disease and weather. Environ Res. 2015;142:319–27. pmid:26188633
- 118. Kakarla SG, Caminade C, Mutheneni SR, Morse AP, Upadhyayula SM, Kadiri MR, et al. Lag effect of climatic variables on dengue burden in India. Epidemiol Infect. 2019;147:e170. pmid:31063099
- 119. Rahaman MR, Dear K, Satter SM, Tong M, Milazzo A, Marshall H, et al. Short-term effects of climate variability on childhood diarrhoea in Bangladesh: multi-site time-series regression analysis. Int J Environ Res Public Health. 2023;20(13):6279. pmid:37444126