Application of Artificial Neural Networks for Dengue Fever Outbreak Predictions in the Northwest Coast of Yucatan, Mexico and San Juan, Puerto Rico

Global infectious disease early warning models: An updated review and lessons from the COVID-19 pandemic

An early warning model for infectious diseases is a crucial tool for timely monitoring, prevention, and control of disease outbreaks. The integration of diverse multi-source data using big data and artificial intelligence techniques has emerged as a key approach in advancing these early warning models. This paper presents a comprehensive review of widely utilized early warning models for infectious diseases around the globe. Unlike previous review studies, this review encompasses newly developed approaches such as the combined model and Hawkes model after the COVID-19 pandemic, providing a thorough evaluation of their current application status and development prospects for the first time. These models not only rely on conventional surveillance data but also incorporate information from various sources. We aim to provide valuable insights for enhancing global infectious disease surveillance and early warning systems, as well as informing future research in this field, by summarizing the underlying modeling concepts, algorithms, and application scenarios of each model.

Machine learning in epidemiology: Neural networks forecasting of monkeypox cases

This study integrates advanced machine learning techniques, namely Artificial Neural Networks, Long Short-Term Memory, and Gated Recurrent Unit models, to forecast monkeypox outbreaks in Canada, Spain, the USA, and Portugal. The research focuses on the effectiveness of these models in predicting the spread and severity of cases using data from June 3 to December 31, 2022, and evaluates them against test data from January 1 to February 7, 2023. The study highlights the potential of neural networks in epidemiology, especially concerning recent monkeypox outbreaks. It provides a comparative analysis of the models, emphasizing their capabilities in public health strategies. The research identifies optimal model configurations and underscores the efficiency of the Levenberg-Marquardt algorithm in training. The findings suggest that ANN models, particularly those with optimized Root Mean Squared Error, Mean Absolute Percentage Error, and the Coefficient of Determination values, are effective in infectious disease forecasting and can significantly enhance public health responses.

Artificial intelligence in parasitic disease control: A paradigm shift in health care

Parasitic diseases, including malaria, leishmaniasis, and trypanosomiasis, continue to plague populations worldwide, particularly in resource-limited settings and disproportionately affecting vulnerable populations. It has limited the use of conventional health-care delivery and disease control approaches and necessitated exploring innovative strategies. In this direction, artificial intelligence (AI) has emerged as a transformative tool with immense promise in parasitic disease control, offering the potential for enhanced diagnostics, precision drug discovery, predictive modeling, and personalized treatment. Predictive AI algorithms have assisted in understanding parasite transmission patterns and outbreaks by analyzing vast amounts of epidemiological data, environmental factors, and population demographics. This has strengthened public health interventions, resource allocation, and outbreak preparedness strategies, enabling proactive measures to mitigate disease spread. In diagnostics, AI-enabled accurate and rapid identification of parasites by analyzing microscopic images. This capability is particularly valuable in remote regions with limited access to diagnostic facilities. AI-driven computational methods have also assisted in drug discovery for parasitic diseases by identifying novel drug targets and predicting the efficacy and safety of potential drug candidates. This approach has streamlined drug development, leading to more effective and targeted therapies. This article reviews these current developments and their transformative impacts on the health-care sector. It also assessed the hurdles that require attention before these transformations can be realized in real-life scenarios.

Weather integrated multiple machine learning models for prediction of dengue prevalence in India

Dengue is a rapidly spreading viral disease transmitted to humans by Aedes mosquitoes. Due to global urbanization and climate change, the number of dengue cases are gradually increasing in recent decades. Hence, an early prediction of dengue continues to be a major concern for public health in countries with high prevalence of dengue. Creating a robust forecast model for the accurate prediction of dengue is a complex task and can be done through various data modelling approaches. In the present study, we have applied vector auto regression, generalized boosted models, support vector regression, and long short-term memory (LSTM) to predict the dengue prevalence in Kerala state of the Indian subcontinent. We consider the number of dengue cases as the target variable and weather variables viz., relative humidity, soil moisture, mean temperature, precipitation, and NINO3.4 as independent variables. Various analytical models have been applied on both datasets and predicted the dengue cases. Among all the models, the LSTM model was outperformed with superior prediction capability (RMSE: 0.345 and R²:0.86) than the other models. However, other models are able to capture the trend of dengue cases but failed in predicting the outbreak periods when compared to LSTM. The findings of this study will be helpful for public health agencies and policymakers to draw appropriate control measures before the onset of dengue. The proposed LSTM model for dengue prediction can be followed by other states of India as well.

Interpretable machine learning approach to analyze the effects of landscape and meteorological factors on mosquito occurrences in Seoul, South Korea

Mosquitoes are the underlying cause of various public health and economic problems. In this study, patterns of mosquito occurrence were analyzed based on landscape and meteorological factors in the metropolitan city of Seoul. We evaluated the influence of environmental factors on mosquito occurrence through the interpretation of prediction models with a machine learning algorithm. Through hierarchical cluster analysis, the study areas were classified into waterside and non-waterside areas, according to the landscape patterns. The mosquito occurrence was higher in the waterside area, and mosquito abundance was negatively affected by rainfall at the waterside. The mosquito occurrence was predicted in each cluster area based on the landscape and cumulative meteorological variables using a random forest algorithm. Both models exhibited good performance (both accuracy and AUROC > 0.8) in predicting the level of mosquito occurrence. The embedded relationship between the mosquito occurrence and the environmental factors in the models was explained using the Shapley additive explanation method. According to the variable importance and the partial dependence plots for each model, the waterside area was more influenced by the meteorological and land cover variables than the non-waterside area. Therefore, mosquito control strategies should consider the effects of landscape and meteorological conditions, including the temperature, rainfall, and the landscape heterogeneity. The present findings can contribute to the development of mosquito forecasting systems in metropolitan cities for the promotion of public health.

Artificial Neural Networks for the Prediction of Monkeypox Outbreak

While the world is still struggling to recover from the harm caused by the widespread COVID-19 pandemic, the monkeypox virus now poses a new threat of becoming a pandemic. Although it is not as dangerous or infectious as COVID-19, new cases of the disease are nevertheless being reported daily from many countries. In this study, we have used public datasets provided by the European Centre for Disease Prevention and Control for developing a prediction model for the spread of the monkeypox outbreak to and throughout the USA, Germany, the UK, France and Canada. We have used certain effective neural network models for this purpose. The novelty of this study is that a neural network model for a time series monkeypox dataset is developed and compared with LSTM and GRU models using an adaptive moment estimation (ADAM) optimizer. The Levenberg-Marquardt (LM) learning technique is used to develop and validate a single hidden layer artificial neural network (ANN) model. Different ANN model architectures with varying numbers of hidden layer neurons were trained, and the K-fold cross-validation early stopping validation approach was employed to identify the optimum structure with the best generalization potential. In the regression analysis, our ANN model gives a good R-value of almost 99%, the LSTM model gives almost 98% and the GRU model gives almost 98%. These three model fits demonstrated that there was a good agreement between the experimental data and the forecasted values. The results of our experiments show that the ANN model performs better than the other methods on the collected monkeypox dataset in all five countries. To the best of the authors' knowledge, this is the first report that has used ANN, LSTM and GRU to predict a monkeypox outbreak in all five countries.

Dengue Prediction in Latin America Using Machine Learning and the One Health Perspective: A Literature Review

Dengue fever is a serious and growing public health problem in Latin America and elsewhere, intensified by climate change and human mobility. This paper reviews the approaches to the epidemiological prediction of dengue fever using the One Health perspective, including an analysis of how Machine Learning techniques have been applied to it and focuses on the risk factors for dengue in Latin America to put the broader environmental considerations into a detailed understanding of the small-scale processes as they affect disease incidence. Determining that many factors can act as predictors for dengue outbreaks, a large-scale comparison of different predictors over larger geographic areas than those currently studied is lacking to determine which predictors are the most effective. In addition, it provides insight into techniques of Machine Learning used for future predictive models, as well as general workflow for Machine Learning projects of dengue fever.

Euclidean Distance Modeling of Musi River in Controlling the Dengue Epidemic Transmission in Palembang City

BACKGROUND: Various attempts have been made to control the population of Aedes aegypti with the help of chemicals or by engineering Wolbachia pipentis, an obligate intracellular bacterium that is passed down through DENV and arbovirus infections to manipulate the monthly average reproductive yield. This study reviews the phenomenon of the river border area which is one of the habitats for the Aedes aegypti mosquito in the Musi River, Palembang City. AIM: The application of the euclidean distance method in this study was carried out to determine the environmental exposure of settlements along the river basin area. METHODS: The research methodology was carried out objectively related to data on dengue incidence in 2019. It was carried out by taking location coordinates through the application of geographic information systems and the use of satellite imagery for data acquisition of existing buildings. This stage is followed by bivariate statistical calculations using the application of WoE where the probability value of the measurement is described using the Area Under Curve. Processing and accumulation carried out with existing buildings will result in a calculation of the estimated size of the exposure area. RESULTS: The results obtained provide information, where the natural breaks jeanks value of 0.007-0.016 range results in 1465ha of heavily exposed building area. The value of the temporary bivariate statistical calculation will produce an AUC probability number of 0.44 which describes the relationship between the Musi river and the findings of dengue symptoms in the sub-districts around the Musi river border area, Palembang City. Swamp soil conditions are vulnerable to being a habitat where Aedes aegypti larvae are found. CONCLUSIONS: Based on the analysis that we obtained from the population of dengue incidence and the condition of the river basin area showed a significant structure with the distribution of dengue incidence, it is known that the presence of buildings on the river Musi banks has a greater risk of infectious diseases transmissions and natural disasters ranging from sanitation, hygiene, flooding to river erosion.

Cluster Analysis of Dengue Morbidity and Mortality in Mexico from 2007 to 2020: Implications for the Probable Case Definition

Dengue cases and deaths occur frequently in Mexico, although the trend is not uniform across the country. We performed a Spatio-temporal analysis of dengue cases and deaths in Mexico from 2007 to 2020, and clustered states according to whether there was a low, moderate, or high risk of dengue. A total of 501,600 confirmed dengue cases were registered from 2007 to 2020, with 378,122 cases classified as dengue fever (DF) and 123,478 cases classified as dengue hemorrhagic fever (DHF). For each confirmed case, there were 4.68 probable cases. There were 1,230 dengue deaths, with highest numbers reported in 2009, 2012, 2013, and 2019. The number of deaths had a significant correlation (P ≤ 0.01) with DF (r = 0.82), DHF (r = 0.94), and probable dengue cases (r = 0.84). States were clustered using Machine Learning technique according to select indices associated with dengue. Cluster 1 (low risk) primarily contained states in the northwest, northcentral, and east. Cluster 2 (moderate risk) includes states in the northeast. Cluster 3 (high risk) mostly contained coastal states in the southeast, southwest, and west. The generation of the clusters was supported by the Kruskal-Wallis test. A significant difference was found in the incidence, mortality rates, and case-fatality rates of dengue among the clusters (P ≤ 0.01). Notably, cluster 3 contributed 71.4% of the confirmed cases and 89.2% of the deaths. Public health and vector control strategies designed to mitigate the burden of dengue in Mexico should consider the states in cluster 3 as high priority areas.

Artificial Intelligence: Review of Current and Future Applications in Medicine

Background

The role of artificial intelligence (AI) in health care is expanding rapidly. Currently, there are at least 29 US Food and Drug Administration-approved AI health care devices that apply to numerous medical specialties and many more are in development.

Observations

With increasing expectations for all health care sectors to deliver timely, fiscally-responsible, high-quality health care, AI has potential utility in numerous areas, such as image analysis, improved workflow and efficiency, public health, and epidemiology, to aid in processing large volumes of patient and medical data. In this review, we describe basic terminology, principles, and general AI applications relating to health care. We then discuss current and future applications for a variety of medical specialties. Finally, we discuss the future potential of AI along with the potential risks and limitations of current AI technology.

Conclusions

AI can improve diagnostic accuracy, increase patient safety, assist with patient triage, monitor disease progression, and assist with treatment decisions.

Comparison of different predictive models on HFMD based on weather factors in Zibo city, Shandong Province, China

The early identification and prediction of hand-foot-and-mouth disease (HFMD) play an important role in the disease prevention and control. However, suitable models are different in regions due to the differences in geography, social economy factors. We collected data associated with daily reported HFMD cases and weather factors of Zibo city in 2010~2019 and used the generalised additive model (GAM) to evaluate the effects of weather factors on HFMD cases. Then, GAM, support vectors regression (SVR) and random forest regression (RFR) models are used to compare predictive results. The annual average incidence was 129.72/100 000 from 2010 to 2019. Its distribution showed a unimodal trend, with incidence increasing from March, peaking from May to September. Our study revealed the nonlinear relationship between temperature, rainfall and relative humidity and HFMD cases and based on the predictive result, the performances of three models constructed ranked in descending order are: SVR > GAM> RFR, and SVR has the smallest prediction errors. These findings provide quantitative evidence for the prediction of HFMD for special high-risk regions and can help public health agencies implement prevention and control measures in advance.

A novel approach to modeling epidemic vulnerability, applied to Aedes aegypti-vectored diseases in Perú

Background

A proactive approach to preventing and responding to emerging infectious diseases is critical to global health security. We present a three-stage approach to modeling the spatial distribution of outbreak vulnerability to Aedes aegypti-vectored diseases in Perú.

Methods

Extending a framework developed for modeling hemorrhagic fever vulnerability in Africa, we modeled outbreak vulnerability in three stages: index case potential (stage 1), outbreak receptivity (stage 2), and epidemic potential (stage 3), stratifying scores on season and El Niño events. Subsequently, we evaluated the validity of these scores using dengue surveillance data and spatial models.

Results

We found high validity for stage 1 and 2 scores, but not stage 3 scores. Vulnerability was highest in Selva Baja and Costa, and in summer and during El Niño events, with index case potential (stage 1) being high in both regions but outbreak receptivity (stage 2) being generally high in Selva Baja only.

Conclusions

Stage 1 and 2 scores are well-suited to predicting outbreaks of Ae. aegypti-vectored diseases in this setting, however stage 3 scores appear better suited to diseases with direct human-to-human transmission. To prevent outbreaks, measures to detect index cases should be targeted to both Selva Baja and Costa, while Selva Baja should be prioritized for healthcare system strengthening. Successful extension of this framework from hemorrhagic fevers in Africa to an arbovirus in Latin America indicates its broad utility for outbreak and pandemic preparedness and response activities.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-021-06530-9.

Covid-19 Dynamic Monitoring and Real-Time Spatio-Temporal Forecasting

Background: Periodically, humanity is often faced with new and emerging viruses that can be a significant global threat. It has already been over a century post-the Spanish Flu pandemic, and we are witnessing a new type of coronavirus, the SARS-CoV-2, which is responsible for Covid-19. It emerged from the city of Wuhan (China) in December 2019, and within a few months, the virus propagated itself globally now resulting more than 50 million cases with over 1 million deaths. The high infection rates coupled with dynamic population movement demands for tools, especially within a Brazilian context, that will support health managers to develop policies for controlling and combating the new virus. Methods: In this work, we propose a tool for real-time spatio-temporal analysis using a machine learning approach. The COVID-SGIS system brings together routinely collected health data on Covid-19 distributed across public health systems in Brazil, as well as taking to under consideration the geographic and time-dependent features of Covid-19 so as to make spatio-temporal predictions. The data are sub-divided by federative unit and municipality. In our case study, we made spatio-temporal predictions of the distribution of cases and deaths in Brazil and in each federative unit. Four regression methods were investigated: linear regression, support vector machines (polynomial kernels and RBF), multilayer perceptrons, and random forests. We use the percentage RMSE and the correlation coefficient as quality metrics. Results: For qualitative evaluation, we made spatio-temporal predictions for the period from 25 to 27 May 2020. Considering qualitatively and quantitatively the case of the State of Pernambuco and Brazil as a whole, linear regression presented the best prediction results (thematic maps with good data distribution, correlation coefficient >0.99 and RMSE (%) <4% for Pernambuco and around 5% for Brazil) with low training time: [0.00; 0.04 ms], CI 95%. Conclusion: Spatio-temporal analysis provided a broader assessment of those in the regions where the accumulated confirmed cases of Covid-19 were concentrated. It was possible to differentiate in the thematic maps the regions with the highest concentration of cases from the regions with low concentration and regions in the transition range. This approach is fundamental to support health managers and epidemiologists to elaborate policies and plans to control the Covid-19 pandemics.

Comparison of Dengue Predictive Models Developed Using Artificial Neural Network and Discriminant Analysis with Small Dataset

In Indonesia, dengue has become one of the hyperendemic diseases. Dengue consists of three clinical phases—febrile phase, critical phase, and recovery phase. Many patients have died in the critical phase due to the lack of proper and timely treatment. Therefore, we developed models that can predict the severity level of dengue based on the laboratory test results of the corresponding patients using Artificial Neural Network (ANN) and Discriminant Analysis (DA). In developing the models, we used a very small dataset. It is shown that ANN models developed using logistic and hyperbolic tangent activation function with 70% training data yielded the highest accuracy (90.91%), sensitivity (91.11%), and specificity (95.51%). This is the proposed model in this research. The proposed model will be able to help physicians in predicting the severity level of dengue patients before entering the critical phase. Furthermore, it will ease physicians in treating dengue patients early, so fatal cases or deaths can be avoided.

Prediction of dengue outbreak in Selangor Malaysia using machine learning techniques

Dengue fever is a mosquito-borne disease that affects nearly 3.9 billion people globally. Dengue remains endemic in Malaysia since its outbreak in the 1980's, with its highest concentration of cases in the state of Selangor. Predictors of dengue fever outbreaks could provide timely information for health officials to implement preventative actions. In this study, five districts in Selangor, Malaysia, that demonstrated the highest incidence of dengue fever from 2013 to 2017 were evaluated for the best machine learning model to predict Dengue outbreaks. Climate variables such as temperature, wind speed, humidity and rainfall were used in each model. Based on results, the SVM (linear kernel) exhibited the best prediction performance (Accuracy = 70%, Sensitivity = 14%, Specificity = 95%, Precision = 56%). However, the sensitivity for SVM (linear) for the testing sample increased up to 63.54% compared to 14.4% for imbalanced data (original data). The week-of-the-year was the most important predictor in the SVM model. This study exemplifies that machine learning has respectable potential for the prediction of dengue outbreaks. Future research should consider boosting, or using, nature inspired algorithms to develop a dengue prediction model.

Machine learning and dengue forecasting: Comparing random forests and artificial neural networks for predicting dengue burden at national and sub-national scales in Colombia

The robust estimate and forecast capability of random forests (RF) has been widely recognized, however this ensemble machine learning method has not been widely used in mosquito-borne disease forecasting. In this study, two sets of RF models were developed at the national (pooled department-level data) and department level in Colombia to predict weekly dengue cases for 12-weeks ahead. A pooled national model based on artificial neural networks (ANN) was also developed and used as a comparator to the RF models. The various predictors included historic dengue cases, satellite-derived estimates for vegetation, precipitation, and air temperature, as well as population counts, income inequality, and education. Our RF model trained on the pooled national data was more accurate for department-specific weekly dengue cases estimation compared to a local model trained only on the department's data. Additionally, the forecast errors of the national RF model were smaller to those of the national pooled ANN model and were increased with the forecast horizon increasing from one-week-ahead (mean absolute error, MAE: 9.32) to 12-weeks ahead (MAE: 24.56). There was considerable variation in the relative importance of predictors dependent on forecast horizon. The environmental and meteorological predictors were relatively important for short-term dengue forecast horizons while socio-demographic predictors were relevant for longer-term forecast horizons. This study demonstrates the potential of RF in dengue forecasting with a feasible approach of using a national pooled model to forecast at finer spatial scales. Furthermore, including sociodemographic predictors is likely to be helpful in capturing longer-term dengue trends.

Semi-Supervised Text Classification Framework: An Overview of Dengue Landscape Factors and Satellite Earth Observation

In recent years there has been an increasing use of satellite Earth observation (EO) data in dengue research, in particular the identification of landscape factors affecting dengue transmission. Summarizing landscape factors and satellite EO data sources, and making the information public are helpful for guiding future research and improving health decision-making. In this case, a review of the literature would appear to be an appropriate tool. However, this is not an easy-to-use tool. The review process mainly includes defining the topic, searching, screening at both title/abstract and full-text levels and data extraction that needs consistent knowledge from experts and is time-consuming and labor intensive. In this context, this study integrates the review process, text scoring, active learning (AL) mechanism, and bidirectional long short-term memory (BiLSTM) networks, and proposes a semi-supervised text classification framework that enables the efficient and accurate selection of the relevant articles. Specifically, text scoring and BiLSTM-based active learning were used to replace the title/abstract screening and full-text screening, respectively, which greatly reduces the human workload. In this study, 101 relevant articles were selected from 4 bibliographic databases, and a catalogue of essential dengue landscape factors was identified and divided into four categories: land use (LU), land cover (LC), topography and continuous land surface features. Moreover, various satellite EO sensors and products used for identifying landscape factors were tabulated. Finally, possible future directions of applying satellite EO data in dengue research in terms of landscape patterns, satellite sensors and deep learning were proposed. The proposed semi-supervised text classification framework was successfully applied in research evidence synthesis that could be easily applied to other topics, particularly in an interdisciplinary context.

Weathering the pandemic: How the Caribbean Basin can use viral and environmental patterns to predict, prepare, and respond to COVID-19

The 2020 coronavirus pandemic is developing at different paces throughout the world. Some areas, like the Caribbean Basin, have yet to see the virus strike at full force. When it does, there is reasonable evidence to suggest the consequent COVID‐19 outbreaks will overwhelm healthcare systems and economies. This is particularly concerning in the Caribbean as pandemics can have disproportionately higher mortality impacts on lower and middle‐income countries. Preliminary observations from our team and others suggest that temperature and climatological factors could influence the spread of this novel coronavirus, making spatiotemporal predictions of its infectiousness possible. This review studies geographic and time‐based distribution of known respiratory viruses in the Caribbean Basin in an attempt to foresee how the pandemic will develop in this region. This review is meant to aid in planning short‐ and long‐term interventions to manage outbreaks at the international, national, and subnational levels in the region.

Inter‐tropical regions have seen a slower spread of SARS‐CoV‐2 compared to temperate ones, suggesting viral spreading could likely be influenced by environmental factors.

Other coronaviruses also seem to respond to environmental factors, with peaks closely following the geotemporal patterns of influenza.

In the Caribbean, timing interventions to fight COVID‐19 around the projected peaks of influenza is a reasonable public health approach.

Prediction mapping of human leptospirosis using ANN, GWR, SVM and GLM approaches

Background

Recent reports of the National Ministry of Health and Treatment of Iran (NMHT) show that Gilan has a higher annual incidence rate of leptospirosis than other provinces across the country. Despite several efforts of the government and NMHT to eradicate leptospirosis, it remains a public health problem in this province. Modelling and Prediction of this disease may play an important role in reduction of the prevalence.

Methods

This study aims to model and predict the spatial distribution of leptospirosis utilizing Geographically Weighted Regression (GWR), Generalized Linear Model (GLM), Support Vector Machine (SVM) and Artificial Neural Network (ANN) as capable approaches. Five environmental parameters of precipitation, temperature, humidity, elevation and vegetation are used for modelling and predicting of the disease. Data of 2009 and 2010 are used for training, and 2011 for testing and evaluating the models.

Results

Results indicate that utilized approaches in this study can model and predict leptospirosis with high significance level. To evaluate the efficiency of the approaches, MSE (GWR = 0.050, SVM = 0.137, GLM = 0.118 and ANN = 0.137), MAE (0.012, 0.063, 0.052 and 0.063), MRE (0.011, 0.018, 0.017 and 0.018) and R² (0.85, 0.80, 0.78 and 0.75) are used.

Conclusion

Results indicate the practical usefulness of approaches for spatial modelling and predicting leptospirosis. The efficiency of models is as follow: GWR > SVM > GLM > ANN. In addition, temperature and humidity are investigated as the most influential parameters. Moreover, the suitable habitat of leptospirosis is mostly within the central rural districts of the province.

A dynamic neural network model for predicting risk of Zika in real time

BACKGROUND

In 2015, the Zika virus spread from Brazil throughout the Americas, posing an unprecedented challenge to the public health community. During the epidemic, international public health officials lacked reliable predictions of the outbreak's expected geographic scale and prevalence of cases, and were therefore unable to plan and allocate surveillance resources in a timely and effective manner.

METHODS

In this work, we present a dynamic neural network model to predict the geographic spread of outbreaks in real time. The modeling framework is flexible in three main dimensions (i) selection of the chosen risk indicator, i.e., case counts or incidence rate; (ii) risk classification scheme, which defines the high-risk group based on a relative or absolute threshold; and (iii) prediction forecast window (1 up to 12 weeks). The proposed model can be applied dynamically throughout the course of an outbreak to identify the regions expected to be at greatest risk in the future.

RESULTS

The model is applied to the recent Zika epidemic in the Americas at a weekly temporal resolution and country spatial resolution, using epidemiological data, passenger air travel volumes, and vector habitat suitability, socioeconomic, and population data for all affected countries and territories in the Americas. The model performance is quantitatively evaluated based on the predictive accuracy of the model. We show that the model can accurately predict the geographic expansion of Zika in the Americas with the overall average accuracy remaining above 85% even for prediction windows of up to 12 weeks.

CONCLUSIONS

Sensitivity analysis illustrated the model performance to be robust across a range of features. Critically, the model performed consistently well at various stages throughout the course of the outbreak, indicating its potential value at any time during an epidemic. The predictive capability was superior for shorter forecast windows and geographically isolated locations that are predominantly connected via air travel. The highly flexible nature of the proposed modeling framework enables policy makers to develop and plan vector control programs and case surveillance strategies which can be tailored to a range of objectives and resource constraints.

A GIS-Based Artificial Neural Network Model for Spatial Distribution of Tuberculosis across the Continental United States

Despite the usefulness of artificial neural networks (ANNs) in the study of various complex problems, ANNs have not been applied for modeling the geographic distribution of tuberculosis (TB) in the US. Likewise, ecological level researches on TB incidence rate at the national level are inadequate for epidemiologic inferences. We collected 278 exploratory variables including environmental and a broad range of socio-economic features for modeling the disease across the continental US. The spatial pattern of the disease distribution was statistically evaluated using the global Moran’s I, Getis–Ord General G, and local Gi* statistics. Next, we investigated the applicability of multilayer perceptron (MLP) ANN for predicting the disease incidence. To avoid overfitting, L1 regularization was used before developing the models. Predictive performance of the MLP was compared with linear regression for test dataset using root mean square error, mean absolute error, and correlations between model output and ground truth. Results of clustering analysis showed that there is a significant spatial clustering of smoothed TB incidence rate (p < 0.05) and the hotspots were mainly located in the southern and southeastern parts of the country. Among the developed models, single hidden layer MLP had the best test accuracy. Sensitivity analysis of the MLP model showed that immigrant population (proportion), underserved segments of the population, and minimum temperature were among the factors with the strongest contributions. The findings of this study can provide useful insight to health authorities on prioritizing resource allocation to risk-prone areas.