Neural networks for dengue forecasting: a systematic review

Neural networks for dengue forecasting: a systematic review | medRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-P4HH5NV'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search Neural networks for dengue forecasting: a systematic review View ORCID Profile L. Lober , View ORCID Profile F. A. Rodrigues , View ORCID Profile K. O. Roster doi: https://doi.org/10.1101/2025.10.14.25338016 L. Lober 1 Departamento de Matemática Aplicada e Estatística, Instituto de Ciências Matemáticas e de Computação, Universidade de São Paulo—Campus de São Carlos , Caixa Postal 668, 13560-970 São Carlos, São Paulo, Brazil Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for L. Lober For correspondence: luiza.lober{at}usp.br F. A. Rodrigues 1 Departamento de Matemática Aplicada e Estatística, Instituto de Ciências Matemáticas e de Computação, Universidade de São Paulo—Campus de São Carlos , Caixa Postal 668, 13560-970 São Carlos, São Paulo, Brazil Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for F. A. Rodrigues K. O. Roster 1 Departamento de Matemática Aplicada e Estatística, Instituto de Ciências Matemáticas e de Computação, Universidade de São Paulo—Campus de São Carlos , Caixa Postal 668, 13560-970 São Carlos, São Paulo, Brazil 2 Harvard T. H. Chan School of Public Health , Boston, MA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for K. O. Roster Abstract Full Text Info/History Metrics Supplementary material Data/Code Preview PDF Abstract Background Early forecasts of dengue are an important tool for disease mitigation. Neural networks are powerful predictive models that have made contributions to many areas of public health. In this study, we reviewed the application of neural networks in the dengue forecasting literature, with the objective of informing model design for future work. Methods Following PRISMA guidelines, we conducted a systematic search of studies that use neural networks to forecast dengue in human populations. We summarized the relative performance of neural networks and comparator models, architectures and hyper-parameters, choices of input features, geographic spread, and model transparency. Results Sixty two papers were included. Most studies implemented shallow feed-forward neural networks, using historical dengue incidence and climate variables. Prediction horizons varied greatly, as did the model selection and evaluation approach. Building on the strengths of neural networks, most studies used granular observations at the city level, or on its subdivisions, while also commonly employing weekly data. Performance of neural networks relative to comparators, such as tree-based supervised models, varied across study contexts, and we found that 63% of all studies do include at least one such model as a baseline, and in those cases about half of the studies report neural networks as the best performing model. Conclusions The studies suggest that neural networks can provide competitive forecasts for dengue, and can reliably be included in the set of candidate models for future dengue prediction efforts. The use of deep networks is relatively unexplored but offers promising avenues for further research, as does the use of a broader set of input features and prediction in light of structural changes in the data generation mechanism. 1 Introduction Dengue fever is a mosquito-borne disease with a significant global burden. More than 100 countries are at risk of infection [ 1 ], with a substantial increase registered in the last five years, including European and eastern Mediterranean regions [ 2 ]. 2025, up to July, accounted for 3.6 million symptomatic dengue fever cases worldwide, a number that is possibly under-reported considering evidence from previous years [ 1 ]. Dengue fever presents as a flu-like illness with symptoms ranging from mild to severe, with potentially life-threatening complications arising from severe dengue. The virus occurs in four serotypes, meaning up to four infections are possible for each person over their lifetime, and no specific treatment currently exists. Preventing the circulation of the arboviruses and early warnings to public health agencies, including vector control methods and pinpointing hotspots of the disease, are essential methods for dengue mitigation, even considering recently available vaccination strategies against all dengue serotypes, which are already available in multiple countries [ 1 , 3 , 4 ]. To that end, predictive models can help efficiently allocate public health resources to combat dengue, with the goal of reducing the overall disease burden. Dengue incidence is associated with many different risk factors. These include climate conditions such as rainfall, extreme weather events, and temperature [ 5 , 6 , 7 ], land use [ 8 ], and poverty [ 9 ]. These factors, and others related to dengue, may have nonlinear, context-specific, and time-varying effects on disease incidence, which poses a challenge to disease modeling. The literature on dengue forecasting is multi-disciplinary, uniting expertise from areas such as epidemiology, environmental science, computer science, and mathematics. Modeling frameworks include both theoretical and data-driven approaches, spanning dynamic transmission models, statistical time series forecasting, and machine learning methods. Machine learning models benefit from the increasing data availability on risk factors on the spreading of diseases, and offer non-parametric approaches that require less detailed knowledge of the disease and its context [ 10 ]. Neural networks are a subset of these algorithms, which have made significant contributions to medicine and public health, including applications such as medical image analysis for disease diagnosis [ 11 ], identifying abnormalities in signals such as electrocardiographs (ECG) [ 12 ], and optimizing decisions of health care providers, hospitals, and policy-makers [ 13 ]. Neural networks are also used to forecast diseases, including malaria [ 14 ], influenza [ 15 ], and COVID-19 [ 16 ]. This review examined the use of neural networks for dengue fever prediction, focusing specifically on: the technical decisions made in the literature, including architecture selection, hyper-parameter tuning and model proposals; input data, such as climate and demographic variables; the relative performance of different neural network architectures and comparator models, such as other machine learning techniques; geographic and temporal evolution of publications. 2 Methodology This review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement guidelines [ 17 ]. The PRISMA flow chart for this study is represented in figure 9 . A systematic search was conducted using Web of Science/Knowledge, Scopus (abstract, title, keywords), PubMed, and Science Direct (abstract, title) databases. References of papers appearing in the search results were also examined for relevant works. The searches were conducted in September 10 2024, and used the following search string: The papers were examined for inclusion in the review in three stages: by title, by abstract, and finally by full text. In each phase, the following inclusion criteria were applied conservatively. If the information in the title and/or abstract was inconclusive, the paper was included in the full-text review. These inclusion criteria were as follows: Studies must implement a neural network or deep learning technique, either as the main method or as a comparator model. Reviews of the literature on neural networks for dengue forecasting were also considered; Studies must predict dengue fever incidence or risk. There were no restrictions on the type of target variable used. For example, the number of dengue cases, dengue incidence rate, or a binary dengue risk variable were all accepted as target features; Studies must examine dengue in a human population. Models for disease diagnosis of individuals were excluded. Studies modeling the location of vectors without relation to dengue incidence were excluded. Studies examining animal hosts were excluded; As the search query is in English, only English language articles were identified and included in the review. Alongside these criteria, two other standards were used when screening the full texts of selected papers: Data sources and processing steps must have been clearly stated; Methodology must be clear and well structured, and there was no apparent data leakage in its proceedings. Metrics and models, along with all relevant parameters to the chosen NN architecture, when available, were specified. We also excluded papers that did not allow institutional access to the authors for reviewing, which was the case for three studies that fulfilled the eligibility criteria up to the full-text screening stage. The results of each stage of the screening process can be seen in Figure 9 on Appendix A. 3 Results 3.1 Data Sources Meteorologic features are the most common data source, being featured in 43 of the 61 papers included in this study. This data comprises temperature, humidity, wind speed, solar radiation, pressure, rainfall, sea surface temperature measurements and El Niño [ 18 ] data. Out of all papers that did not include meteorologic data, only four [ 19 , 20 , 21 , 22 ] add any extra features to train the neural networks beyond the epidemiologic information on dengue fever. In [ 19 , 20 ], both authors used environmental data coming from satellite or street imaging in conjunction to epidemiologic data for their forecasts; [ 21 ] instead uses social media data to complement their models. [ 22 ] on the other hand employs data on bus lines to understand how mobility restrictions in the pandemic years impacted dengue transmission in urban areas. The second most frequent data source used for dengue fever forecasting was environmental data, found both as satellite images [ 23 ], vegetation indexes [ 24 , 25 , 26 , 27 ] and street imaging from the Google StreetView program [ 19 , 20 ]. Following closely the number of uses of environment information, data on mosquitoes for fore-casting dengue consisted on mosquito density [ 26 ] and density per person in a given month [ 28 ], larvae free factor [ 29 ], biting rate [ 30 ], entomological data in the form of raw number of Ae. Aegypti and Ae. Albopictus mosquitoes [ 31 ], and also the infection rate of larvae, male and female mosquitoes [ 32 ]. Moreover, [ 27 ] used species distribution model for dengue transmitting mosquitoes and Ae. Aegypt temperature suitability. Other, rarely-occurring features, found throughout the studies were human behavior, social media, geographic and socioeconomic data, which includes: Behavior : number of holidays in a given week, used as a boolean variable in [ 33 ], human mobility through call data records (CDR) processing in [ 24 ] bus lines in [ 22 ] and time spent in transit stations from Google Regional Mobility Reports [ 34 ]. Social media : short messages from X (formerly Twitter) containing dengue-related information, selected through preprocessing methodologies to ascertain the relevance of the social media data to dengue circulation [ 35 , 21 , 36 ]. Geography : [ 37 ] included this information from data on the place of residence of infected individuals, labeled as township or village. We also listed population density from [ 32 , 38 ] considering land use, explicitly used in [ 39 ], to dengue forecast; Socioeconomic data : [ 31 ] used several extra information to improve their dengue forecasts: literacy, access to health services, and housing and sanitation information, which included houses with dirt floors, toilet facility, water pipelines, sewage system and electricity. Lastly, epidemiologic data, e.g. case counts or incidence rates, are necessarily included in all studies due to the nature of the prediction problem. Studies focused both on macro regions or smaller divisions of a given area for their forecasts, covering forecasts at district-, city-, province-, and country-level ( Figure 2 ). “Province” stands for a subdivision of countries, such as states, and “district” are subdivisions of cities, which included both neighborhoods and the health center coverage area within a city. Download figure Open in new tab Figure 1: Auxiliary variables to dengue epidemiological data included in the selected studies, displaying the frequency with which studies included various categories of predictor variables and their combinations. The top bars measure their intersection size in the selected papers, and the central table shows the interplay between multiple variables. ‘Meteorologic’ data is present in most (69%) studies. Download figure Open in new tab Figure 2: Geographic and temporal distribution, with green bars highlighting the most frequent category in each case. The left panel shows the frequency with which studies predict dengue cases at the city, district, province, and country level. The right panel shows the frequency with which studies predict dengue cases at daily, weekly, monthly, seasonal, and yearly level. The “entire period” of dengue data accounts for total number of cases in a certain period, and was used by studies that predicted dengue fever risk in specific regions [ 19 , 20 ]. In Figure 3 , Brazil was the most frequent country for studies forecasting dengue fever, with 12 papers using Brazilian data in their analyzes, followed by Indonesia (8 papers) and India (5 papers). There was an increase in publication of studies using neural networks to forecast dengue starting in 2017, with 2019, 2025 and 2023, in this order, featuring the highest number of papers using this methodology. Download figure Open in new tab Figure 3: Geographic coverage. a) Number of studies predicting dengue in each country from 2008 to 2025 and b) number of studies predicting dengue in different geographic regions, by year of publication. We also analyzed the accessibility criteria of papers, verifying data and code availability in the full text, public accessibility of the paper, and publication venue, with results condensed in Figure 4 . Download figure Open in new tab Figure 4: Accessibility of studies. The vertical bars show the frequency of open and restricted access, publication venue, data availability, and source code availability of the selected studies. The gray flows indicate how frequently combinations of characteristics appear together. Download figure Open in new tab Figure 5: Types of neural network architectures employed by the selected studies and their respective publication year, grouped by structural similarity. The green bars indicate the most commonly used architecture, which was the artificial neural network (ANN). Download figure Open in new tab Figure 6: All implemented models found in this review. The horizontal bars accounts for model appearances in the selected studies, with colors grouping structural similarity of these models (green: neural networks, blue: supervised models and yellow: statistical models). Vertical bars show the frequency with which models were used together in a given study. We split papers not published as “open access” in two categories, to understand whether the authors made the full text available as a pre-print before publication. With that in mind, 71% of the selected studies allows for free access to their methodology and results. As for the data being readily available from the listed references, some cases included sensitive data on a population (such as call data records), and as such had only partially published data sources alongside the papers, being then labeled as “partial access to data” in the figure above. As for papers with restrictions on the data sources, most of them either referenced sources no longer available, or that needed some form of authorization or request to the organization responsible for maintaining the data to be accessed. Figure 4 . : Accessibility of studies. The vertical bars show the frequency of open and restricted access, publication venue, data availability, and source code availability of the selected studies. The gray flows indicate how frequently combinations of characteristics appear together. Lastly, only one fifth of all papers were published alongside their source code, with almost all that do also coming from open access journals. From those, the policies of most of those journals were to encourage authors to also make their code available, being only a requirement at Nature Communications: medicine (for [ 31 ]), PLOS One (for [ 40 ]) and PLOS Neglected tropical diseases (for [ 22 , 39 ]). 3.2 Architectures We reviewed the implemented architectures of neural network algorithms from the selected studies, distinguishing among similarly structured neural networks: recurrent neural networks (RNN), convolutional neural networks (CNN), feed-forward artificial neural networks (ANN), and other architectures. We also analyzed hyper-parameters and other modeling choices, such as network size (hidden layers and units), regularization (dropout), and learning rate. We also compared the optimal hyper-parameters reported in the selected studies for the implemented NNs in Table 1 , describing the parameters through the same architecture grouping criteria. View this table: View inline View popup Download powerpoint Table 1: Hyper-parameter values by neural network (NN) architecture type. Appearances refers to the number of papers using a neural network architecture that also report a given parameter. As some models define the size of layers or number of hidden units according to the number of lags, namely the number of steps delaying a time series, this information is included here for clarity. Most studies that implemented ANNs do so with the usual fully connected structure, with a few exceptions being given by the NNAR in [ 41 , 42 , 43 , 44 , 45 ], an autoregressive neural network algorithm that relies on statistical criteria for parameter tuning and data processing; and the RBFNN in [ 46 ], which uses radial basis functions as activation functions in the output layer. One study [ 47 ] uniquely implemented a Fuzzy system in order to use clustering for dengue forecast, while also comparing their method to other clustering algorithms. Two studies featuring CNNs [ 37 , 23 ] implemented a U-Net, which is a convolutional neural network architecture widely used for image segmentation [ 48 ], to combine satellite images and epidemiologic data to forecast dengue infections in a given region. All others employed the usual convolutional layer stacking of CNNs, with varying number of layers and parameters (see 1). As for the RNNs, the most common architecture used is the long short-term memory (LSTM) NN, appearing in 20 studies. Other less frequent architectures are given by the gated recurrent units (GRU) in [ 49 , 50 , 51 , 21 ] and the BlockRNN [ 45 ]. A Hopfield network is used in [ 52 ] to infer the reproduction number R 0 in a Susceptible-Infectious-Recovered (SIR) compartmental dynamic transmission model, with the resulting equation then being integrated and used as a comparison to the epidemiologic data. Other studies [ 23 , 30 ] also employed neural networks and compartmental models for forecasting. All mentions of optimizer usage were found in this category, with six studies opting for the Adam optimizer [ 49 , 51 , 53 , 25 , 54 , 40 , 55 , 38 ], and one each for Adamax [ 54 ] and RMSProp [ 21 ]. From the 13 papers that discuss activation functions, it was found that: rectifier linear units (ReLu) was the most commonly used activation function, found in six studies [ 35 , 50 , 37 , 56 , 40 , 20 ], followed by the sigmoid activation function [ 57 , 58 , 28 , 29 ]. Studies also used the logistic sigmoid [ 59 ], logistic [ 60 ], hyperbolic tangent (tanh) [ 35 ], linear [ 28 ] activation functions, with the last two used in a layer before the dense layer. Uniquely mentioned parameters are the Akaike information criterion (AIC) in [ 45 ], used to select the size of the input layer; the spread factor σ for a particle swarm optimization (PSO) algorithm found on [ 26 ]; momentum on [ 57 ], and [ 47 ] made use of fuzzy if-then rules to create input-output pairs for a ANN+fuzzy layer hybrid. Pre-trained embeddings are also used in [ 21 ]. 3.3 Model selection and evaluation Most studies implemented multiple algorithms, comparing different neural network architectures or machine learning models. ARIMA, being by far the most frequent statistical-based comparator model, was set apart from other similar models, while supervised models were split into tree-based methods, such as XGBoost and Random Forests, and others algorithms. Figure 10 shows the variety of NN and comparator models found in this review, alongside their number of appearances. Out of the 10 studies that used CNNs, half of them do not employ another comparator model in their report, which is also the highest rate out of all neural network categories (11 out of 34 for ANNs and 6 out of 28 for RNNs). This frequency does not depend on the year of publication of the studies. When studies did include comparator models, ANNs tended to feature statistical models more often (14 out of 34 entries), with ARIMA being particularly prominent in this case, with 9 papers, while other papers used supervised models for comparisons. RNNs have the opposite distribution, with studies having used supervised models more often and only 10 applying statistical models. It is important to note that including one type of comparator model does not impede the study of also reporting others. Finally, two studies included CNNs with it as a comparator itself: one for an original wavelet neural network architecture [ 45 ], and another using multiple CNN and RNN architectures [ 50 ]. Others compared the CNNs to other neural networks [ 61 ], statistical models and tree-based algorithms [ 62 ] and an ensemble of all three model types in [ 36 ]. Notably, 21 out of all studies included lags, which are delays in a time series set according to a window size defined by the user. They were used in these studies for both epidemic records and the auxiliary variables. Moreover, the 11 studies that also use any type of statistical models as comparators, such as the various autorregressive algorithms shown in Figure 10 , favor employing lagged time series to the modeling process, with 65% of them having this type of feature in the training dataset. Within the criteria of study selection, there was significant variation in the studies’ specifications, including the formulation of the target feature. Most studies (88%) predicted the future number of dengue cases based on historical time series. Others made forecasts on the risk of outbreak or dengue severity and its hotspots in a given region. The metrics found throughout this review were grouped accordingly in Figure 7 . Download figure Open in new tab Figure 7: All performance metrics found in the select papers, grouped by target variable: dengue cases forecast, risk of outbreak and dengue severity or hotspots. Another important distinction between studies is that the forecast window varies depending on the predictive objective and not all studies reported prediction windows: most of the studies (44%) made predictions for the whole testing set, which varies in size depending on the data split chosen by authors. When reported, forecasts were mostly made from 1 to 12 week ahead, with the most common forecasting windows employed by these studies ranging from one up to four weeks. From Figure 7 , the most common metric by type of forecast are: Root Mean Squared Error (RMSE) for forecasting dengue fever case counts, Mean Average Error (MAE) for risk of outbreak, and accuracy for detecting dengue severity, or hotspots, in a given area. More than half (58%) of these studies also employed more than one evaluation metric. For niche metrics, only two studies [ 63 , 64 , 22 ] do not include more frequently used ones when evaluating models through multiple measurements. It was also found that there was not an standardized way of acquiring and treating datasets to train neural networks for dengue forecast, as shown by the different choices in comparison to the type of metric used (see left panel of Figure 8 ). Cross-validation is also rare, and used specifically in [ 65 , 60 , 66 , 52 , 28 , 64 , 67 , 62 , 30 ]. Download figure Open in new tab Figure 8: Model evaluation. The left panel shows the averages over the observed splits for each metric, with color gradients indicating the total length of data employed. The right panel shows whether neural networks performed best, and whether the study included comparator models. It was also noted that some papers only discussed a single architecture and at most compared their results to other similar applications from the literature, without further comparator models implemented, as it can be verified in the right panel of Figure 8 . 4 Discussion We reviewed the literature on dengue forecasting using neural networks to analyze trends in methodologies and use of input features, identify research gaps and methodology transparency, and also provide a practical guide for researchers aiming to implement neural networks in their own forecasting work. Most of the studies included in this systematic review reported that their proposed neural network architectures outperformed other machine learning or statistical approaches, suggesting neural networks as a useful method that should be considered alongside other candidate models to forecast dengue incidence, and consequently create strategies that can be later used to guide the decisions of public health agencies. Though many risk factors of dengue have been identified, most neural network models limit inputs to meteorological and epidemiological data. Alternative sources of predictor variables leveraged by studies in this review include data from from X (formerly Twitter) [ 35 , 21 , 36 ], Baidu search data [ 10 ], Google street view and satellite images including derived features [ 20 , 24 , 23 , 27 , 39 ], mobile phone records [ 24 ], and mobility data [ 22 , 34 ]. Future studies may evaluate the value of alternative predictors, especially as neural networks can be implemented to handle high-dimensional problems well [ 68 ]. While datasets are expanding, there are still limits to dengue surveillance, including high asymptomatic rates and limits to diagnostic capacity [ 1 ]. One study in this review aimed to address data scarcity using transfer learning[ 65 ]. The authors trained a model on a city with high dengue incidence and later used it to predict the disease in lower-incidence geographies. This was the only study in this review that used transfer learning outside the context of pre-trained CNNs. This may present another avenue for further research, especially in locations where data is scarce, or to tackle under-reporting of dengue cases. Another paper aimed to improve dengue data by addressing the common problem of seasonal fluctuations in case counts, especially at fine temporal and spatial resolution, i.e. zero-inflation, which may lead to inaccurate predictions and performance evaluation [ 39 ]. The authors demonstrated that sparse data interfered with machine learning model accuracy. More research in this space can provide an avenue for a reliable use of data with higher granularity in the future. There is increasing interest in improving explainability of machine learning models [ 69 ], which is also reflected in this review. For example, SHapley Additive exPlanations (SHAP) were used alongside models to locate climatic drivers of dengue outbreaks [ 55 ]. Changes in climate are expected to greatly impact the global burden of infectious diseases, including dengue [ 70 ]. While much of the reviewed literature focuses on prediction in unchanged conditions, more work is needed to anticipate dengue outbreaks in places with profound structural changes in climate and socioeconomics, which may disrupt historic patterns of dengue seasonality, the scale and scope of outbreaks, and endemicity. These models may prove critical to anticipating and controlling dengue outbreaks under variable conditions. One study [ 27 ] used an ANN to provide global and local projections of dengue distribution and infection considering changes in temporal patterns of climate and socioeconomic data to support policy formulation, however the aforementioned gap in addressing climate change in dengue forecast remains, specially considering that the majority of the studies included in this review make use of meteorological variables. We were not able to conduct a meta-analysis of the performance of neural networks in the selected studies, due to limited overlap in datasets and geographies, variations in model architectures, including the train-test split, and lack of open access and transparency in model code and methods. This limits our ability to generalize on the performance of different modeling choices in different contexts, such as epidemic vs. endemic regions of dengue. Future research would benefit from comparing several different model types and architectures on multiple benchmark datasets to better understand how the ideal model varies by their geographic context or data availability. The field would also benefit from greater transparency and open source coding, as much of the code and data was restricted in the selected studies. One particular study proposed an online and modular platform to automate data retrieval and forecasting for dengue and other arboviruses of interest through several supervised approaches, including multiple possible neural network implementations, which can also be a valuable tool to guide health policies by allowing the benchmark these methods and datasets. [ 71 ]. Efforts like these can help improve usability of and trust in advanced neural network models by governments to inform dengue control policies. Neural networks are a promising and increasingly-used tool in dengue forecasting. While much of the field still relies on simple architectures and datasets, several studies have pioneered the inclusion of diverse digital data streams, novel methodologies, such as transfer learning or satellite data. Prediction in the context of changing data generation mechanisms, for example due to climate change, is another promising new avenue of inquiry. More work is needed to compare model performance across geographies and settings and improve the transparency and accessibility of models for policymakers. When implemented carefully, neural networks can aid in the control of dengue. Data Availability The present study did not produce additional data, and all databases screened for this review are available online at PubMed, Scopus, Web of Science and Science Direct. A The PRISMA flowchart for this study Download figure Open in new tab Figure 9: PRISMA [ 17 ] flow chart for this study. A All NN architectures Download figure Open in new tab Figure 10: All models found throughout the screening process and included in this review, grouped by structural similarities. Acknowledgments Luiza Lober thanks the support given by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) (grants number 2022/16065-3 and 2013/07375-0). Francisco A. Rodrigues acknowledges CNPq (grant 308162/2023-4) and FAPESP (grants 20/09835-1 and 13/07375-0) for the financial support given for this research. Kirstin Oliveira Roster gratefully acknowledges support from the São Paulo Research Foundation (FAPESP) (grant number 2019/26595-7). References 1. ↵ World Health Organization (WHO ). Dengue – Global Situation. [ Accessed 25. Sep. 2025 ]. 2024 May 30. Available from: https://www.who.int/news-room/fact-sheets/detail/dengue-and-severe-dengue 2. ↵ European Centre for Disease Prevention and Control . Dengue worldwide overview. [ Accessed 26. Aug. 2025 ]. Available from: https://www.ecdc.europa.eu/en/dengue-monthly 3. ↵ Takeda Pharmaceutical Company. Takeda and Biological E . Limited Collaborate to Accelerate Access to Dengue Vaccine in Endemic Areas . [Online; accessed 26. Aug. 2025 ]. 2025 Jun . Available from: https://www.takeda.com/newsroom/newsreleases/2024/collaboration-to-accelerate-access-to-dengue-vaccine 4. ↵ Ranzani OT , Neto FL , Mareto LK , Brumatti TS , Oliveira RD de , Silva PV da , Santos ER dos , Agostini TLD , De Paula RAC , Dean NE , Ko AI , Cummings DAT , Andrews JR , Hitchings MDT , and Croda J. Effectiveness of the TAK-003 dengue vaccine in adolescents during the 2024 outbreak in São Paulo, Brazil: a test-negative, case–control study . Lancet Infect. Dis . 2025 Aug ; 0 . doi: 10.1016/S1473-3099(25)00382-2 OpenUrl CrossRef 5. ↵ Chien LC and Yu HL . Impact of meteorological factors on the spatiotemporal patterns of dengue fever incidence . Environ. Int . 2014 Dec ; 73 : 46 – 56 . doi: 10.1016/j.envint.2014.06.018 OpenUrl CrossRef PubMed 6. ↵ Thu HM , Aye KM , and Thein S. The effect of temperature and humidity on dengue virus propagation in Aedes aegypti mosquitos . Southeast Asian J. Trop. Med. Public Health 1998 Jun ; 29 : 280 – 4 . eprint: 9886113 OpenUrl PubMed 7. ↵ Fan J , Lin H , Wang C , Bai L , Yang S , Chu C , Yang W , and Liu Q. Identifying the high-risk areas and associated meteorological factors of dengue transmission in Guangdong Province, China from 2005 to 2011 . Epidemiol. Infect . 2014 Mar ; 142 : 634 – 43 . doi: 10.1017/S0950268813001519 OpenUrl CrossRef 8. ↵ Kilpatrick AM and Randolph SE . Drivers, dynamics, and control of emerging vector-borne zoonotic diseases . Lancet 2012 Dec ; 380 : 1946 – 55 . doi: 10.1016/S0140-6736(12)61151-9 OpenUrl CrossRef PubMed Web of Science 9. ↵ Mulligan K , Dixon J , Sinn CLJ , and Elliott SJ . Is dengue a disease of poverty? A systematic review. Pathogens and Global Health 2015 Feb 10. ↵ Guo P , Liu T , Zhang Q , Wang L , Xiao J , Zhang Q , Luo G , Li Z , He J , Zhang Y , and Ma W. Developing a dengue forecast model using machine learning: A case study in China . PLoS Negl.Trop. Dis . 2017 ; 11 : e0005973 . doi: 10.1371/journal.pntd.0005973 OpenUrl CrossRef 11. ↵ Mienye ID , Swart TG , Obaido G , Jordan M , and Ilono P. Deep Convolutional Neural Networks in Medical Image Analysis: A Review . Information 2025 Mar ; 16 : 195 . doi: 10.3390/info16030195 OpenUrl CrossRef 12. ↵ Haglin JM , Jimenez G , and Eltorai AEM . Artificial neural networks in medicine . Health Technol . 2019 Jan ; 9 : 1 – 6 . doi: 10.1007/s12553-018-0244-4 OpenUrl CrossRef 13. ↵ Shahid N , Rappon T , and Berta W. Applications of artificial neural networks in health care organizational decision-making: A scoping review . PLoS One 2019 Feb ; 14 : e0212356 . doi: 10.1371/journal.pone.0212356 OpenUrl CrossRef PubMed 14. ↵ Zinszer K , Verma AD , Charland K , Brewer TF , Brownstein JS , Sun Z , and Buckeridge DL . A scoping review of malaria forecasting: past work and future directions . BMJ Open 2012 ; 2 . doi: 10.1136/bmjopen-2012-001992 . eprint: https://bmjopen.bmj.com/content/2/6/e001992.full.pdf OpenUrl Abstract / FREE Full Text 15. ↵ Alessa A and Faezipour M. A review of influenza detection and prediction through social networking sites . Theor. Biol. Med. Model . 2018 Dec ; 15 : 1 – 27 . doi: 10.1186/s12976-017-0074-5 OpenUrl CrossRef PubMed 16. ↵ Bullock J , Luccioni A , Pham KH , Lam CSN , and Luengo-Oroz M. Mapping the landscape of Artificial Intelligence applications against COVID-19 . J. Artif. Intell. Res . 2020 Nov ; 69 : 807 – 45 . doi: 10.1613/jair.1.12162 OpenUrl CrossRef 17. ↵ Page MJ , McKenzie JE , Bossuyt PM , Boutron I , Hoffmann TC , Mulrow CD , Shamseer L , Tetzlaff JM , Akl EA , Brennan SE , Chou R , Glanville J , Grimshaw JM , HrÓbjartsson A , Lalu MM , Li T , Loder EW , Mayo-Wilson E , McDonald S , McGuinness LA , Stewart LA , Thomas J , Tricco AC , Welch VA , Whiting P , and Moher D. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews . BMJ 2021 ; 372 . doi: 10.1136/bmj.n71 . eprint: https://www.bmj.com/content/372/bmj.n71.full.pdf OpenUrl FREE Full Text 18. ↵ Tran NT , Xin J , and Zhou G. FWin Transformer for Dengue Prediction Under Climate and Ocean Influence . Machine Learning, Optimization, and Data Science . Cham, Switzerland : Springer , 2025 Mar : 160 – 75 . doi: 10.1007/978-3-031-82484-5_12 OpenUrl CrossRef 19. ↵ Andersson VO , Ferreira Birck MA , and Araujo RM . Towards Predicting Dengue Fever Rates Using Convolutional Neural Networks and Street-Level Images . 2018 International Joint Conference on Neural Networks (IJCNN) . 2018 : 1 – 8 . doi: 10.1109/IJCNN.2018.8489567 OpenUrl CrossRef 20. ↵ Andersson VO , Cechinel C , and Araujo RM . Combining Street-level and Aerial Images for Dengue Incidence Rate Estimation . 2019 International Joint Conference on Neural Networks (IJCNN) . 2019 : 1 – 8 . doi: 10.1109/IJCNN.2019.8851749 OpenUrl CrossRef 21. ↵ Livelo ED and Cheng C. Intelligent Dengue Infoveillance Using Gated Recurrent Neural Learning and Cross-Label Frequencies . 2018 IEEE International Conference on Agents (ICA) . 2018 : 2 – 7 . doi: 10.1109/AGENTS.2018.8459963 OpenUrl CrossRef 22. ↵ Araújo JLB , Bomfim R , Filho CINS , Cavalcanti LPG , Neto ASL , Andrade Jr . JS , and Furtado V. The impact of COVID-19 mobility restrictions on dengue transmission in urban areas . PLoS Negl.Trop. Dis . 2024 Nov ; 18 : e0012644 . doi: 10.1371/journal.pntd.0012644 OpenUrl CrossRef 23. ↵ Abdur Rehman N , Saif U , and Chunara R. Deep Landscape Features for Improving Vectorborne Disease Prediction . Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops . 2019 Jun 24. ↵ Dharmawardana KGS , Lokuge JN , Dassanayake PSB , Sirisena ML , Fernando ML , Perera AS , and Lokanathan S. Predictive model for the dengue incidences in Sri Lanka using mobile network big data . 2017 IEEE International Conference on Industrial and Information Systems (ICIIS). Peradeniya : IEEE , 2017 Dec . doi: 10.1109/ICIINFS.2017.8300381 OpenUrl CrossRef 25. ↵ Li Z , Gurgel H , Xu L , Yang L , and Dong J. Improving dengue forecasts by using geospatial big data analysis in Google Earth Engine and the historical dengue information-aided long short term memory modeling . en. Biology (Basel) 2022 Jan ; 11 : 169 . doi: 10.3390/biology11020169 OpenUrl CrossRef 26. ↵ Anggraeni W , Yuniarno EM , Rachmadi RF , Sumpeno S , Pujiadi P , Sugiyanto S , Santoso J , and Purnomo MH . A hybrid EMD-GRNN-PSO in intermittent time-series data for dengue fever forecasting . en. Expert Syst. Appl . 2024 Mar ; 237 : 121438 . doi: 10.1016/j.eswa.2023.121438 OpenUrl CrossRef 27. ↵ Mosquito and global dengue cases in a warming world . [Online; accessed 15. Sep. 2025 ]. 2025 May . doi: 10.1136/bmjgh-2023-014688 OpenUrl Abstract / FREE Full Text 28. ↵ Dinh TQ , L. HV , Cao TH , Luong QC , and Diep HT . Forecasting the magnitude of dengue in southern Vietnam. en . Intelligent Information and Database Systems. Lecture notes in computer science . Berlin, Heidelberg : Springer Berlin Heidelberg , 2016 : 554 – 63 . doi: 10.1007/978-3-662-49381-6_53 OpenUrl CrossRef 29. ↵ Anggraeni W , Wicaksono AA , Yuniarno EM , Rachmadi RF , Sumpeno S , and Purnomo MH . Multilevel analysis of temporal-based spatial factors impact in dengue fever forecasting us-ing RReliefF - deep learning . 2022 IEEE International Conference on Imaging Systems and Techniques (IST) . Kaohsiung, Taiwan : IEEE , 2022 Jun . doi: 10.1109/IST55454.2022.9827717 OpenUrl CrossRef 30. ↵ Lu X , Teh SY , Tay CJ , Abu Kassim NF , Fam PS , and Soewono E. Application of multiple linear regression model and long short-term memory with compartmental model to forecast dengue cases in Selangor, Malaysia based on climate variables . Infectious Disease Modelling 2025 Mar ; 10 : 240 – 56 . doi: 10.1016/j.idm.2024.10.007 OpenUrl CrossRef PubMed 31. ↵ Dong B , Khan L , Smith M , Trevino J , Zhao B , Hamer GL , Lopez-Lemus UA , Molina AA , Lubinda J , Nguyen USDT , and Haque U. Spatio-temporal dynamics of three diseases caused by Aedes-borne arboviruses in Mexico . en. Commun. Med. (Lond .) 2022 Oct ; 2 : 134 . doi: 10.1038/s43856-022-00192-7 OpenUrl CrossRef 32. ↵ Kesorn K , Ongruk P , Chompoosri J , Phumee A , Thavara U , Tawatsin A , and Siriyasatien P. Morbidity rate prediction of dengue hemorrhagic fever (DHF) using the support vector machine and the Aedes aegypti infection rate in similar climates and geographical areas . en. PLoS One 2015 May ; 10 : e0125049 . doi: 10.1371/journal.pone.0125049 OpenUrl CrossRef 33. ↵ Bergero P , Schaposnik LP , and Wang G. Correlations between COVID-19 and dengue obtained via the study of South America, Africa and Southeast Asia during the 2020s . Sci. Rep . 2023 Jan ; 13 : 1 – 17 . doi: 10.1038/s41598-023-27983-9 OpenUrl CrossRef PubMed 34. ↵ Roster KO , Martinelli T , Connaughton C , Santillana M , and Rodrigues FA . Impact of the COVID-19 pandemic on dengue in Brazil: Interrupted time series analysis of changes in surveillance and transmission . PLoS Negl.Trop. Dis . 2024 Dec ; 18 : e0012726 . doi: 10.1371/journal.pntd.0012726 OpenUrl CrossRef 35. ↵ Mussumeci E and Codeço Coelho F. Large-scale multivariate forecasting models for Dengue - LSTM versus random forest regression . Spatial and Spatio-temporal Epidemiology 2020 ; 35 : 100372 . doi: 10.1016/j.sste.2020.100372 OpenUrl CrossRef 36. ↵ Salsabiila NJ and Anggraeni W. CNN+BiGRU-Attention Classification and TiDE-PSO Forecasting Approach for Social Media-based Predictive Analysis of Dengue . International Journal of Intelligent Engineering and System 2025 37. ↵ Anno S , Tsubasa H , Sugita S , Yasumoto S , Lee MA , Sasaki Y , and and KO . Challenges and implications of predicting the spatiotemporal distribution of dengue fever outbreak in Chinese Taiwan using remote sensing data and deep learning . Geo-spatial Information Science 2024 ; 27 : 1155 – 61 . doi: 10.1080/10095020.2022.2144770 . eprint: 10.1080/10095020.2022.2144770 OpenUrl CrossRef 38. ↵ Das D , Maiti S , and Sarma DK . A high-resolution GIS and machine learning approach for targeted disease management and localized risk assessment in an urban setup: A case study from Bhopal, Central India . Acta Trop . 2025 Jul ; 267 : 107662 . doi: 10.1016/j.actatropica.2025.107662 OpenUrl CrossRef PubMed 39. ↵ Francisco ME , Carvajal TM , and Watanabe K. Hybrid Machine Learning Approach to Zero-Inflated Data Improves Accuracy of Dengue Prediction . PLoS Negl.Trop. Dis . 2024 Oct ; 18 : e0012599 . doi: 10.1371/journal.pntd.0012599 OpenUrl CrossRef 40. ↵ Baquero OS , Santana LMR , and Chiaravalloti-Neto F. Dengue forecasting in São Paulo city with generalized additive models, artificial neural networks and seasonal autoregressive integrated moving average models . en. PLoS One 2018 Apr ; 13 : e0195065 . doi: 10.1371/journal.pone.0195065 OpenUrl CrossRef 41. ↵ Mustaffa NA , Zahari SM , Farhana NA , Nasir N , and Azil AH . Forecasting the incidence of dengue fever in Malaysia: A comparative analysis of seasonal ARIMA, dynamic harmonic regression, and neural network models . International Journal of Advanced and Applied Sciences 2024 Feb ; 11 : 20 – 31 . doi: 10.21833/ijaas.2024.01.003 OpenUrl CrossRef 42. ↵ Chakraborty T , Chattopadhyay S , and Ghosh I. Forecasting dengue epidemics using a hybrid methodology . Physica A: Statistical Mechanics and its Applications 2019 ; 527 : 121266 . doi: 10.1016/j.physa.2019.121266 OpenUrl CrossRef 43. ↵ Koh YM , Spindler R , Sandgren M , and and JJ . A model comparison algorithm for increased forecast accuracy of dengue fever incidence in Singapore and the auxiliary role of total precipitation information . International Journal of Environmental Health Research 2018 ; 28 . PMID: 30016117 : 535 – 52 . doi: 10.1080/09603123.2018.1496234 . eprint: 10.1080/09603123.2018. 1496234 OpenUrl CrossRef PubMed 44. ↵ Mahajan A , Sharma N , Aparicio-Obregon S , Alyami H , Alharbi A , Anand D , Sharma M , and Goyal N. A novel stacking-based deterministic ensemble model for infectious Disease Prediction . en. Mathematics 2022 May ; 10 : 1714 . doi: 10.3390/math10101714 OpenUrl CrossRef 45. ↵ Panja M , Chakraborty T , Nadim SS , Ghosh I , Kumar U , and Liu N. An ensemble neural network approach to forecast Dengue outbreak based on climatic condition . Chaos, Solitons Fractals 2023 Feb ; 167 : 113124 . doi: 10.1016/j.chaos.2023.113124 OpenUrl CrossRef 46. ↵ Muhammad F , Hertono GF , and Handari BD . Comparison of activation functions on radial basis function neural network in predicting dengue hemorrhagic fever incidents in DKI Jakarta . AIP Conference Proceedings 2020 Nov ; 2296 : 020017 . doi: 10.1063/5.0030453 . eprint: https://pubs.aip.org/aip/acp/article-pdf/doi/10.1063/5.0030453/14222583/020017\_1\_online.pdf OpenUrl CrossRef 47. ↵ Hasanah H , Hertono GF , and Sarwinda D. Prediction of dengue incidence in DKI Jakarta using adaptive neuro-fuzzy inference system . INTERNATIONAL CONFERENCE ON SCI-ENCE AND APPLIED SCIENCE (ICSAS2020) . Surakarta, Indonesia : AIP Publishing , 2020 . doi: 10.1063/5.0030455 OpenUrl CrossRef 48. ↵ Ronneberger O , Fischer P , and Brox T. U-Net: Convolutional Networks for Biomedical Image Segmentation . arXiv 2015 May . doi: 10.48550/arXiv.1505.04597 . eprint: 1505.04597 OpenUrl CrossRef 49. ↵ Ullah MA , Mim AS , Hasan MN , and Sadik MR . Deep Learning Based Forecasting Models of Dengue Outbreak in Bangladesh: Comparative Analysis of LSTM, RNN, and GRU Models Using Multivariate Variables with a Two-Decade Dataset . 2024 International Conference on Smart Systems for applications in Electrical Sciences (ICSSES) . 2024 : 1 – 6 . doi: 10.1109/ICSSES62373.2024.10561382 OpenUrl CrossRef 50. ↵ Zhao X , Li K , Ang CKE , and Cheong KH . A deep learning based hybrid architecture for weekly dengue incidences forecasting . Chaos, Solitons Fractals 2023 Mar ; 168 : 113170 . doi: 10.1016/j.chaos.2023.113170 OpenUrl CrossRef 51. ↵ Ferdousi T , Cohnstaedt LW , and Scoglio CM . A Windowed Correlation-Based Feature Selection Method to Improve Time Series Prediction of Dengue Fever Cases . IEEE Access 2021 ; 9 : 141210 – 22 . doi: 10.1109/ACCESS.2021.3120309 OpenUrl CrossRef 52. ↵ Garcáa-Garaluz E , Atencia M , Joya G , Garcáa-Lagos F , and Sandoval F. Hopfield networks for identification of delay differential equations with an application to dengue fever epidemics in Cuba . Neurocomputing 2011; 74. Advances in Extreme Learning Machine: Theory and Applications Biological Inspired Systems. Computational and Ambient Intelligence : 2691 – 7 . doi: 10.1016/j.neucom.2011.03.022 OpenUrl CrossRef 53. ↵ Li Z. Forecasting Weekly Dengue Cases by Integrating Google Earth Engine-Based Risk Predictor Generation and Google Colab-Based Deep Learning Modeling in Fortaleza and the Federal District, Brazil . International Journal of Environmental Research and Public Health 2022 ; 19 . doi: 10.3390/ijerph192013555 OpenUrl CrossRef 54. ↵ Kakarla SG , Kondeti PK , Vavilala HP , Boddeda GSB , Mopuri R , Kumaraswamy S , Kadiri MR , and Mutheneni SR . Weather integrated multiple machine learning models for prediction of dengue prevalence in India . en. Int. J. Biometeorol . 2023 Feb ; 67 : 285 – 97 . doi: 10.1007/s00484-022-02405-z OpenUrl CrossRef 55. ↵ Chen X and Moraga P. Forecasting dengue across Brazil with LSTM neural networks and SHAP-driven lagged climate and spatial effects . BMC Public Health 2025 Dec ; 25 : 1 – 22 . doi: 10.1186/s12889-025-22106-7 OpenUrl CrossRef PubMed 56. ↵ Brieva J , Romero E , and Lepore N. Medelin Ceballos-Arroyo AM , Maldonado-Perez D , Mesa-Yepes H , Perez L , Ciuoderis K , Comach G , Osorio-Benitez JE , Hernandez-Ortiz JP , and Branch-Bedoya JW . Towards a machine learning-based approach to forecasting Dengue virus outbreaks in Colombian cities: a casestudy: Medellin, Antioquia . 15th International Symposium on Medical Information Processing and Analysis . Ed. by Brieva J , Romero E , and Lepore N. Medelin , Colombia : SPIE , 2020 Jan . doi: 10.1117/12.2547001 OpenUrl CrossRef 57. ↵ Anggraeni W , Pramudita G , Riksakomara E P.W. R , Samopa F , Pujiadi , and Dewi RS . Artificial Neural Network for Health Data Forecasting, Case Study: Number of Dengue Hem-orrhagic Fever Cases in Malang Regency, Indonesia . 2018 International Conference on Electrical Engineering and Computer Science (ICECOS) . 2018 : 207 – 12 . doi: 10.1109/ICECOS.2018.8605254 OpenUrl CrossRef 58. ↵ Sukama Y , Hertono GF , Handari BD , and Aldila D. Comparing activation functions in predicting dengue hemorrhagic fever cases in DKI Jakarta using recurrent neural networks . AIP Conference Proceedings 2020 Nov ; 2296 : 020059 . doi: 10.1063/5.0030456 . eprint: https://pubs.aip.org/aip/acp/article-pdf/doi/10.1063/5.0030456/14220614/020059\_1\_online.pdf OpenUrl CrossRef 59. ↵ P.h.m. Nishanthi Herath A.a.i. Perera HW . Prediction of Dengue Outbreaks in Sri Lanka using Artificial Neural Networks . International Journal of Computer Applications 2014 Sep ; 101 : 1 – 5 . doi: 10.5120/17760-8862 OpenUrl CrossRef 60. ↵ Rachata N , Charoenkwan P , Yooyativong T , Chamnongthal K , Lursinsap C , and Higuchi K. Automatic Prediction System of Dengue Haemorrhagic-Fever Outbreak Risk by Using Entropy and Artificial Neural Network . 2008 International Symposium on Communications and Information Technologies . 2008 : 210 – 4 . doi: 10.1109/ISCIT.2008.4700184 OpenUrl CrossRef 61. ↵ Dhaked DK , Sharma O , Gopal Y , and Gopal R. Prediction of dengue patients using deep learning methods amid complex weather conditions in Jaipur, India . Discover Public Health 2025 Dec ; 22 : 1 – 13 . doi: 10.1186/s12982-025-00448-2 OpenUrl CrossRef 62. ↵ Jaya IGNM , Andriyana Y , Tantular B , Pangastuti SS , and Kristiani F. Spatiotemporal Dengue Forecasting for Sustainable Public Health in Bandung, Indonesia: A Comparative Study of Classical, Machine Learning, and Bayesian Models . Sustainability 2025 Jul ; 17 : 6777 . doi: 10.3390/su17156777 OpenUrl CrossRef 63. ↵ Akter T , Islam MT , Hossain MF , and Ullah MS . A Comparative Study between Time Series and Machine Learning Technique to Predict Dengue Fever in Dhaka City . Discrete Dynamics in Nature and Society 2024 ; 2024 : 2757381 . doi: 10.1155/2024/2757381 . eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1155/2024/2757381 OpenUrl CrossRef 64. ↵ Wotawa F , Friedrich G , Pill I , Koitz-Hristov R , and Ali M. Abeyrathna KD , Granmo OC , Zhang X , and Goodwin M. A Scheme for Continuous Input to the Tsetlin Machine with Applications to Forecasting Disease Outbreaks . Advances and Trends in Artificial Intelligence. From Theory to Practice . Ed. by Wotawa F , Friedrich G , Pill I , Koitz-Hristov R , and Ali M. Cham: Springer International Publishing , 2019 : 564 – 78 . doi: 10.1007/978-3-030-22999-3_49 OpenUrl CrossRef 65. ↵ Xu J , Xu K , Li Z , Meng F , Tu T , Xu L , and Liu Q. Forecast of Dengue Cases in 20 Chinese Cities Based on the Deep Learning Method . Int. J. Environ. Res. Public Health 2020 Jan ; 17 : 453 . doi: 10.3390/ijerph17020453 OpenUrl CrossRef PubMed 66. ↵ Sérgio de Souza L , Alves-Souza SN , Filgueiras LVL , Velloso LMR , Fontes de Carvalho M , Garcia LA , Ito M , Jarske JM , Santos TL dos , Fernandes HM , Araújo GM , and Barbosa WL . Forecast of Dengue Cases based on the Deep Learning Approach: A Case Study for a Brazilian City . Proceedings of the 3rd International Conference on Deep Learning Theory and Applications - DeLTA. INSTICC. SciTePress , 2022 : 71 – 6 . doi: 10.5220/0011135500003277 OpenUrl CrossRef 67. ↵ Sousa LB , Von Zuben CJ , and Von Zuben FJ . Regularized linear and nonlinear autoregressive models for dengue confirmed-cases prediction. en . Unconventional Computation and Natural Computation. Lecture notes in computer science. Cham: Springer International Publishing , 2015 : 121 – 31 . doi: 10.1007/978-3-319-21819-9_8 OpenUrl CrossRef 68. ↵ Sekmen A and Bilgin B. Manifold-based approach for neural network robustness analysis . Commun. Eng . 2024 Aug ; 3 : 1 – 13 . doi: 10.1038/s44172-024-00263-8 OpenUrl CrossRef 69. ↵ Linardatos P , Papastefanopoulos V , and Kotsiantis S. Explainable AI: A Review of Machine Learning Interpretability Methods . Entropy 2020 Dec ; 23 : 18 . doi: 10.3390/e23010018 OpenUrl CrossRef PubMed 70. ↵ Messina JP , Brady OJ , Golding N , Kraemer MUG , Wint GRW , Ray SE , Pigott DM , Shearer FM , Johnson K , Earl L , Marczak LB , Shirude S , Davis Weaver N , Gilbert M , Velayudhan R , Jones P , Jaenisch T , Scott TW , Reiner RC , and Hay SI . The current and future global distribution and population at risk of dengue . Nat. Microbiol . 2019 Sep ; 4 : 1508 – 15 . doi: 10.1038/s41564-019-0476-8 OpenUrl CrossRef PubMed 71. ↵ Ganem F , Vacaro LB , Araujo EC , Alves LD , Bastos L , Carvalho LM , Almeida I , Sá AM de , and Coelho FC . Mosqlimate: a platform to providing automatable access to data and forecasting models for arbovirus disease . 2024 . arXiv: 2410.18945 [stat.AP]. Available from: https://arxiv.org/abs/2410.18945 72. Necesito I , Velasco JM , Kwak J , Lee J , Lee M , Kim J , and Kim H. COMBINATION OF UNIVARIATE LONG-SHORT TERM MEMORY NETWORK AND WAVELET TRANS-FORM FOR PREDICTING DENGUE CASE DENSITY IN THE NATIONAL CAPITAL REGION, THE PHILIPPINES . The Southeast Asian journal of tropical medicine and public health 2021 Jul ; 52 : 479 – 94 OpenUrl 73. Nguyen VH , Tuyet-Hanh TT , Mulhall J , Van Minh H , Duong TQ , Van Chien N , Nhung NTT , Lan VH , Minh HB , Cuong D , Bich NN , Quyen NH , Linh TNQ , Tho NT , Nghia ND , Van Quoc Anh L , Phan DTM , Hung NQV , and Son MT . Deep learning models for forecasting dengue fever based on climate data in Vietnam . PLoS Negl.Trop. Dis . 2022 Jun ; 16 : e0010509 . doi: 10.1371/journal.pntd.0010509 OpenUrl CrossRef 74. Rahayu I , Nurhasanah N , and Adriat R. Prediction of Dengue Hemorrhagic Fever Cases Based on Weather Parameters Using Back Propagation Neural Networks (Case Study in Pontianak City) . Jurnal Pendidikan Fisika Indonesia 2019 ; 15 : 114 – 21 . doi: 10.15294/jpfi.v15i2.19633 OpenUrl CrossRef 75. Bogado JV , Stalder DH , Schaerer CE , and GÓmez-Guerrero S. Time Series Clustering to Improve Dengue Cases Forecasting with Deep Learning . 2021 XLVII Latin American Computing Conference (CLEI) . 2021 : 1 – 10 . doi: 10.1109/CLEI53233.2021.9640130 OpenUrl CrossRef 76. Soemsap T , Wongthanavasu S , and Satimai W. Forecasting number of dengue patients using cellular automata model . 2014 International Electrical Engineering Congress (iEECON) . 2014 : 1 – 4 . doi: 10.1109/iEECON.2014.6925876 OpenUrl CrossRef 77. Conde-Gutiérrez RA , Colorado D , Márquez-Nolasco A , and Gonzalez-Flores PB . Parallel prediction of dengue cases with different risks in Mexico using an artificial neural network model considering meteorological data . Int. J. Biometeorol . 2024 Jun ; 68 : 1043 – 60 . doi: 10.1007/s00484-024-02643-3 OpenUrl CrossRef PubMed 78. Anno S , Hara T , Kai H , Lee MA , Chang Y , Oyoshi K , Mizukami Y , and Tadono T. Spatiotemporal dengue fever hotspots associated with climatic factors in Taiwan including outbreak predictions based on machine-learning . en. Geospat. Health 2019 Nov ; 14 . doi: 10.4081/gh.2019.771 OpenUrl CrossRef 79. Navarro Valencia V , Dáaz Y , Pascale JM , Boni MF , and Sanchez-Galan JE . Assessing the effect of climate variables on the incidence of dengue cases in the Metropolitan Region of Panama City . en. Int. J. Environ. Res. Public Health 2021 Nov ; 18 : 12108 . doi: 10.3390/ijerph182212108 OpenUrl CrossRef PubMed 80. Aburas HM , Gultekin Cetiner B , and Sari M. Erratum to “Dengue confirmed-cases prediction: A neural network model” [Expert Systems with Applications 37 (2010) 4256–4260] . en. Expert Syst. Appl . 2011 Sep ; 38 : 13495 – 6 . doi: 10.1016/j.eswa.2011.04.120 OpenUrl CrossRef 81. Ranganathan G , Papakostas GA , and Rocha Á. Rajendran NM , Karthikeyan M , Karthik Raja B , Pragadishwaran K , Gopalakrishnan EA , and Sowmya V. Communicable Disease Prediction Using Machine Learning and Deep Learning Algorithms . Inventive Communication and Computational Technologies . Ed. by Ranganathan G , Papakostas GA , and Rocha Á. Singapore : Springer Nature Singapore , 2023 : 979 – 92 . doi: 10.1007/978-981-99-5166-6_66 OpenUrl CrossRef 82. Datoc HI , Caparas R , and Caro J. Forecasting and data visualization of dengue spread in the Philippine Visayas island group . 2016 7th International Conference on Information, Intelligence, Systems & Applications (IISA). Chalkidiki, Greece : IEEE , 2016 Jul . doi: 10.1109/IISA.2016.7785420 OpenUrl CrossRef 83. Yusof Y and Mustaffa Z. Dengue outbreak prediction: A least squares support vector machines approach . Int. J. Comput. Theory Eng . 2011 : 489 – 93 . doi: 10.7763/IJCTE.2011.V3.355 OpenUrl CrossRef 84. Isidoro C , Fachada N , Barata F , and Rosa A. Agent-Based Model of Aedes aegypti Population Dynamics . Progress in Artificial Intelligence . Berlin, Germany : Springer , 2009 : 53 – 64 . doi: 10.1007/978-3-642-04686-5_5 OpenUrl CrossRef 85. Johansson MA , Reich NG , Hota A , Brownstein JS , and Santillana M. Evaluating the performance of infectious disease forecasts: A comparison of climate-driven and seasonal dengue forecasts for Mexico . Sci. Rep . 2016 Sep ; 6 : 1 – 11 . doi: 10.1038/srep33707 OpenUrl CrossRef PubMed 86. Riley P , Ben-Nun M , Turtle J , Bacon D , and Riley S. SARIMA Forecasts of Dengue Incidence in Brazil, Mexico, Singapore, Sri Lanka, and Thailand: Model Performance and the Significance of Reporting Delays . medRxiv 2020 Jun :2020.06.26.20141093. eprint: 2020.06.26.20141093 87. Anwar SM , Majid M , Qayyum A , Awais M , Alnowami M , and Khan MK . Medical Image Analysis using Convolutional Neural Networks: A Review . J. Med. Syst . 2018 Nov ; 42 : 1 – 13 . doi: 10.1007/s10916-018-1088-1 OpenUrl CrossRef 88. Papik K , Molnar B , Schaefer R , Dombovari Z , Tulassay Z , and Feher J. Application of neural networks in medicine - a review . Diagnostics and Medical Technology 1998 May ; 4 : MT538 – MT546 OpenUrl 89. Abiodun OI , Jantan A , Omolara AE , Dada KV , Mohamed NA , and Arshad H. State-of-the-art in artificial neural network applications: A survey . Heliyon 2018 Nov ; 4 . doi: 10.1016/j.heliyon.2018.e00938 OpenUrl CrossRef 90. Mienye ID , Swart TG , and Obaido G. Recurrent Neural Networks: A Comprehensive Review of Architectures, Variants, and Applications . Information 2024 Aug ; 15 : 517 . doi: 10.3390/info15090517 OpenUrl CrossRef 91. Cong S and Zhou Y. A review of convolutional neural network architectures and their optimizations . Artif. Intell. Rev . 2023 Mar ; 56 : 1905 – 69 . doi: 10.1007/s10462-022-10213-5 OpenUrl CrossRef 92. Rosenblatt F. The perceptron: a probabilistic model for information storage and organization in the brain . Psychol. Rev . 1958 Nov ; 65 : 386 – 408 . doi: 10.1037/h0042519 . eprint: 13602029 OpenUrl CrossRef PubMed Web of Science 93. Chowdhury AH . Comparison of Deep Learning and Gradient Boosting: ANN Versus XGBoost for Climate-Based Dengue Prediction in Bangladesh . Health Science Reports 2025 Apr ; 8 : e70714 . doi: 10.1002/hsr2.70714 OpenUrl CrossRef 94. Zahiruddin H , Zukarnain ZA , and Wijaya A. A scoping review and bibliometric analysis (ScoRBA) on dengue infection and machine learning research . JOIV : International Journal on Informatics Visualization 2024 Dec ; 8 : 2203 – 10 . doi: 10.62527/joiv.8.4.2249 OpenUrl CrossRef View the discussion thread. Back to top Previous Next Posted October 15, 2025. Download PDF Supplementary Material Data/Code Email Thank you for your interest in spreading the word about medRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Neural networks for dengue forecasting: a systematic review Message Subject (Your Name) has forwarded a page to you from medRxiv Message Body (Your Name) thought you would like to see this page from the medRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Neural networks for dengue forecasting: a systematic review L. Lober , F. A. Rodrigues , K. O. Roster medRxiv 2025.10.14.25338016; doi: https://doi.org/10.1101/2025.10.14.25338016 Share This Article: Copy Citation Tools Neural networks for dengue forecasting: a systematic review L. Lober , F. A. Rodrigues , K. O. Roster medRxiv 2025.10.14.25338016; doi: https://doi.org/10.1101/2025.10.14.25338016 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Infectious Diseases (except HIV/AIDS) Subject Areas All Articles Addiction Medicine (568) Allergy and Immunology (863) Anesthesia (299) Cardiovascular Medicine (4426) Dentistry and Oral Medicine (443) Dermatology (382) Emergency Medicine (607) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1507) Epidemiology (15222) Forensic Medicine (30) Gastroenterology (1123) Genetic and Genomic Medicine (6589) Geriatric Medicine (667) Health Economics (997) Health Informatics (4525) Health Policy (1368) Health Systems and Quality Improvement (1612) Hematology (540) HIV/AIDS (1264) Infectious Diseases (except HIV/AIDS) (15910) Intensive Care and Critical Care Medicine (1103) Medical Education (623) Medical Ethics (145) Nephrology (667) Neurology (6588) Nursing (346) Nutrition (998) Obstetrics and Gynecology (1143) Occupational and Environmental Health (956) Oncology (3331) Ophthalmology (971) Orthopedics (369) Otolaryngology (420) Pain Medicine (435) Palliative Medicine (129) Pathology (663) Pediatrics (1690) Pharmacology and Therapeutics (691) Primary Care Research (710) Psychiatry and Clinical Psychology (5440) Public and Global Health (9221) Radiology and Imaging (2195) Rehabilitation Medicine and Physical Therapy (1369) Respiratory Medicine (1196) Rheumatology (593) Sexual and Reproductive Health (710) Sports Medicine (529) Surgery (711) Toxicology (99) Transplantation (289) Urology (265) (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9ffebe664cae0704',t:'MTc3OTQ4MzU1Ng=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();