A scoping review of malaria forecasting: past work and future directions

Fine-Scale Spatial Prediction on the Risk of Plasmodium vivax Infection in the Republic of Korea

BACKGROUND

Malaria elimination strategies in the Republic of Korea (ROK) have decreased malaria incidence but face challenges due to delayed case detection and response. To improve this, machine learning models for predicting malaria, focusing on high-risk areas, have been developed.

METHODS

The study targeted the northern region of ROK, near the demilitarized zone, using a 1-km grid to identify areas for prediction. Grid cells without residential buildings were excluded, leaving 8,425 cells. The prediction was based on whether at least one malaria case was reported in each grid cell per month, using spatial data of patient locations. Four algorithms were used: gradient boosted (GBM), generalized linear (GLM), extreme gradient boosted (XGB), and ensemble models, incorporating environmental, sociodemographic, and meteorological data as predictors. The models were trained with data from May to October (2019-2021) and tested with data from May to October 2022. Model performance was evaluated using the area under the receiver operating characteristic curve (AUROC).

RESULTS

The AUROC of the prediction models performed excellently (GBM = 0.9243, GLM = 0.9060, XGB = 0.9180, and ensemble model = 0.9301). Previous malaria risk, population size, and meteorological factors influenced the model most in GBM and XGB.

CONCLUSION

Machine-learning models with properly preprocessed malaria case data can provide reliable predictions. Additional predictors, such as mosquito density, should be included in future studies to improve the performance of models.

Time series analysis of malaria cases to assess the impact of various interventions over the last three decades and forecasting malaria in India towards the 2030 elimination goals

BACKGROUND

Despite the progress made in this decade towards malaria elimination, it remains a significant public health concern in India and many other countries in South Asia and Asia Pacific region. Understanding the historical trends of malaria incidence in relation to various commodity and policy interventions and identifying the factors associated with its occurrence can inform future intervention strategies for malaria elimination goals.

METHODS

This study analysed historical malaria cases in India from 1990 to 2022 to assess the annual trends and the impact of key anti-malarial interventions on malaria incidence. Factors associated with malaria incidence were identified using univariate and multivariate linear regression analyses. Generalized linear, smoothing, autoregressive integrated moving averages (ARIMA) and Holt's models were used to forecast malaria cases from 2023 to 2030.

RESULTS

The reported annual malaria cases in India during 1990-2000 were 2.38 million, which dropped to 0.73 million cases annually during 2011-2022. The overall reduction from 1990 (2,018,783) to 2022 (176,522) was 91%. The key interventions of the Enhanced Malaria Control Project (EMCP), Intensified Malaria Control Project (IMCP), use of bivalent rapid diagnostic tests (RDT-Pf/Pv), artemisinin-based combination therapy (ACT), and involvement of the Accredited Social Health Activists (ASHAs) as front-line workers were found to result in the decline of malaria significantly. The ARIMA and Holt's models projected a continued decline in cases with the potential for reaching zero indigenous cases by 2027-2028. Important factors influencing malaria incidence included tribal population density, literacy rate, health infrastructure, and forested and hard-to-reach areas.

CONCLUSIONS

Studies aimed at assessing the impact of major commodity and policy interventions on the incidence of disease and studies of disease forecasting will inform programmes and policymakers of steps needed during the last mile phase to achieve malaria elimination. It is proposed that these time series and disease forecasting studies should be performed periodically using granular (monthly) and meteorological data to validate predictions of prior studies and suggest any changes needed for elimination efforts at national and sub-national levels.

Exploring malaria prediction models in Togo: a time series forecasting by health district and target group

OBJECTIVES

Integrating malaria prediction models into malaria control strategies can help to anticipate the response to seasonal epidemics. This study aimed to explore the possibility of using routine malaria data and satellite-derived climate data to forecast malaria cases in Togo.

METHODS

Generalised additive (mixed) models were developed to forecast the monthly number of malaria cases in 40 health districts and three target groups. Routinely collected malaria data from 2013 to 2016 and meteorological and vegetation data with a time lag of 1 or 2 months were used for model training, while the year 2017 was used for model testing. Two methods for selecting lagged meteorological and environmental variables were compared: a first method based on statistical approach ('SA') and a second method based on biological reasoning ('BR'). Both methods were applied to obtain a model per target group and health district and a mixed model per target group and health region with the health district as a random effect. The predictive skills of the four models were compared for each health district and target group.

RESULTS

The most selected predictors in the models per district for the 'SA' method were the normalised difference vegetation index, minimum temperature and mean temperature. The 'SA' method provided the most accurate models for the training period, except for some health districts in children ≥5 years old and adults and in pregnant women. The most accurate models for the testing period varied by health district and target group, provided either by the 'SA' method or the 'BR' method. Despite the development of models with four different approaches, the number of malaria cases was inaccurately forecasted.

CONCLUSIONS

These models cannot be used as such in malaria control activities in Togo. The use of finer spatial and temporal scales and non-environmental data could improve malaria prediction.

Predicting the risk of malaria re-introduction in countries certified malaria-free: a systematic review

BACKGROUND

Predicting the risk of malaria in countries certified malaria-free is crucial for the prevention of re-introduction. This review aimed to identify and describe existing prediction models for malaria re-introduction risk in eliminated settings.

METHODS

A systematic literature search following the PRISMA guidelines was carried out. Studies that developed or validated a malaria risk prediction model in eliminated settings were included. At least two authors independently extracted data using a pre-defined checklist developed by experts in the field. The risk of bias was assessed using both the prediction model risk of bias assessment tool (PROBAST) and the adapted Newcastle-Ottawa Scale (aNOS).

RESULTS

A total 10,075 references were screened and 10 articles describing 11 malaria re-introduction risk prediction models in 6 countries certified malaria free. Three-fifths of the included prediction models were developed for the European region. Identified parameters predicting malaria re-introduction risk included environmental and meteorological, vectorial, population migration, and surveillance and response related factors. Substantial heterogeneity in predictors was observed among the models. All studies were rated at a high risk of bias by PROBAST, mostly because of a lack of internal and external validation of the models. Some studies were rated at a low risk of bias by the aNOS scale.

CONCLUSIONS

Malaria re-introduction risk remains substantial in many countries that have eliminated malaria. Multiple factors were identified which could predict malaria risk in eliminated settings. Although the population movement is well acknowledged as a risk factor associated with the malaria re-introduction risk in eliminated settings, it is not frequently incorporated in the risk prediction models. This review indicated that the proposed models were generally poorly validated. Therefore, future emphasis should be first placed on the validation of existing models.

A systematic review of dengue outbreak prediction models: Current scenario and future directions

Dengue is among the fastest-spreading vector-borne infectious disease, with outbreaks often overwhelm the health system and result in huge morbidity and mortality in its endemic populations in the absence of an efficient warning system. A large number of prediction models are currently in use globally. As such, this study aimed to systematically review the published literature that used quantitative models to predict dengue outbreaks and provide insights about the current practices. A systematic search was undertaken, using the Ovid MEDLINE, EMBASE, Scopus and Web of Science databases for published citations, without time or geographical restrictions. Study selection, data extraction and management process were devised in accordance with the 'Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies' ('CHARMS') framework. A total of 99 models were included in the review from 64 studies. Most models sourced climate (94.7%) and climate change (77.8%) data from agency reports and only 59.6% of the models adjusted for reporting time lag. All included models used climate predictors; 70.7% of them were built with only climate factors. Climate factors were used in combination with climate change factors (13.4%), both climate change and demographic factors (3.1%), vector factors (6.3%), and demographic factors (5.2%). Machine learning techniques were used for 39.4% of the models. Of these, random forest (15.4%), neural networks (23.1%) and ensemble models (10.3%) were notable. Among the statistical (60.6%) models, linear regression (18.3%), Poisson regression (18.3%), generalized additive models (16.7%) and time series/autoregressive models (26.7%) were notable. Around 20.2% of the models reported no validation at all and only 5.2% reported external validation. The reporting of methodology and model performance measures were inadequate in many of the existing prediction models. This review collates plausible predictors and methodological approaches, which will contribute to robust modelling in diverse settings and populations.

An investigation of the efficacy of different statistical models in malaria forecasting in the semi-arid regions of Gujarat, India

BACKGROUND & OBJECTIVES

Robust forecasting of malaria cases is desirable as we are approaching towards malaria elimination in India. Methods enabling robust forecasting and timely case detection in unstable transmission areas are the need of the hour.

METHODS

Forecasting efficacy of the eight most prominent statistical models that are based on three statistical methods: Generalized linear model (Model A and Model B), Smoothing method (Model C), and SARIMA (Model D to model H) were compared using last twelve years (2008-19) monthly malaria data of two districts (Kheda and Anand) of Gujarat state of India.

RESULTS

The SARIMA Model F was found the most appropriate when forecasted for 2017 and 2018 using model-building data sets 1 and 2, respectively, for both the districts: Kheda and Anand. Model H followed by model C were the two models found appropriate in terms of point estimates for 2019. Still, we regretted these two because confidence intervals from these models are wider that they do not have any forecasting utility. Model F is the third one in terms of point prediction but gives a relatively better confidence interval. Therefore, model F was considered the most appropriate for the year 2019 for both districts.

INTERPRETATION & CONCLUSION

Model F was found relatively more appropriate than others and can be used to forecast malaria cases in both districts.

Machine-Learning-Based Forecasting of Dengue Fever in Brazilian Cities Using Epidemiologic and Meteorological Variables

Dengue is a serious public health concern in Brazil and globally. In the absence of a universal vaccine or specific treatments, prevention relies on vector control and disease surveillance. Accurate and early forecasts can help reduce the spread of the disease. In this study, we developed a model for predicting monthly dengue cases in Brazilian cities 1 month ahead, using data from 2007-2019. We compared different machine learning algorithms and feature selection methods using epidemiologic and meteorological variables. We found that different models worked best in different cities, and a random forests model trained on monthly dengue cases performed best overall. It produced lower errors than a seasonal naive baseline model, gradient boosting regression, a feed-forward neural network, or support vector regression. For each city, we computed the mean absolute error between predictions and true monthly numbers of dengue cases on the test data set. The median error across all cities was 12.2 cases. This error was reduced to 11.9 when selecting the optimal combination of algorithm and input features for each city individually. Machine learning and especially decision tree ensemble models may contribute to dengue surveillance in Brazil, as they produce low out-of-sample prediction errors for a geographically diverse set of cities.

Predicting malaria outbreaks from sea surface temperature variability up to 9 months ahead in Limpopo, South Africa, using machine learning

Malaria is the cause of nearly half a million deaths worldwide each year, posing a great socioeconomic burden. Despite recent progress in understanding the influence of climate on malaria infection rates, climatic sources of predictability remain poorly understood and underexploited. Local weather variability alone provides predictive power at short lead times of 1-2 months, too short to adequately plan intervention measures. Here, we show that tropical climatic variability and associated sea surface temperature over the Pacific and Indian Oceans are valuable for predicting malaria in Limpopo, South Africa, up to three seasons ahead. Climatic precursors of malaria outbreaks are first identified via lag-regression analysis of climate data obtained from reanalysis and observational datasets with respect to the monthly malaria case count data provided from 1998-2020 by the Malaria Institute in Tzaneen, South Africa. Out of 11 sea surface temperature sectors analyzed, two regions, the Indian Ocean and western Pacific Ocean regions, emerge as the most robust precursors. The predictive value of these precursors is demonstrated by training a suite of machine-learning classification models to predict whether malaria case counts are above or below the median historical levels and assessing their skills in providing early warning predictions of malaria incidence with lead times ranging from 1 month to a year. Through the development of this prediction system, we find that past information about SST over the western Pacific Ocean offers impressive prediction skills (~80% accuracy) for up to three seasons (9 months) ahead. SST variability over the tropical Indian Ocean is also found to provide good skills up to two seasons (6 months) ahead. This outcome represents an extension of the effective prediction lead time by about one to two seasons compared to previous prediction systems that were more computationally costly compared to the machine learning techniques used in the current study. It also demonstrates the value of climatic information and the prediction framework developed herein for the early planning of interventions against malaria outbreaks.

Predicting the impact of climate change on the re-emergence of malaria cases in China using LSTMSeq2Seq deep learning model: a modelling and prediction analysis study

OBJECTIVES

Malaria is a vector-borne disease that remains a serious public health problem due to its climatic sensitivity. Accurate prediction of malaria re-emergence is very important in taking corresponding effective measures. This study aims to investigate the impact of climatic factors on the re-emergence of malaria in mainland China.

DESIGN

A modelling study.

SETTING AND PARTICIPANTS

Monthly malaria cases for four Plasmodium species (P. falciparum, P. malariae, P. vivax and other Plasmodium) and monthly climate data were collected for 31 provinces; malaria cases from 2004 to 2016 were obtained from the Chinese centre for disease control and prevention and climate parameters from China meteorological data service centre. We conducted analyses at the aggregate level, and there was no involvement of confidential information.

PRIMARY AND SECONDARY OUTCOME MEASURES

The long short-term memory sequence-to-sequence (LSTMSeq2Seq) deep neural network model was used to predict the re-emergence of malaria cases from 2004 to 2016, based on the influence of climatic factors. We trained and tested the extreme gradient boosting (XGBoost), gated recurrent unit, LSTM, LSTMSeq2Seq models using monthly malaria cases and corresponding meteorological data in 31 provinces of China. Then we compared the predictive performance of models using root mean squared error (RMSE) and mean absolute error evaluation measures.

RESULTS

The proposed LSTMSeq2Seq model reduced the mean RMSE of the predictions by 19.05% to 33.93%, 18.4% to 33.59%, 17.6% to 26.67% and 13.28% to 21.34%, for P. falciparum, P. vivax, P. malariae, and other plasmodia, respectively, as compared with other candidate models. The LSTMSeq2Seq model achieved an average prediction accuracy of 87.3%.

CONCLUSIONS

The LSTMSeq2Seq model significantly improved the prediction of malaria re-emergence based on the influence of climatic factors. Therefore, the LSTMSeq2Seq model can be effectively applied in the malaria re-emergence prediction.

Bayesian spatio-temporal distributed lag modeling for delayed climatic effects on sparse malaria incidence data

Background

In many areas of the Greater Mekong Subregion (GMS), malaria endemic regions have shrunk to patches of predominantly low-transmission. With a regional goal of elimination by 2030, it is important to use appropriate methods to analyze and predict trends in incidence in these remaining transmission foci to inform planning efforts. Climatic variables have been associated with malaria incidence to varying degrees across the globe but the relationship is less clear in the GMS and standard methodologies may not be appropriate to account for the lag between climate and incidence and for locations with low numbers of cases.

Methods

In this study, a methodology was developed to estimate the spatio-temporal lag effect of climatic factors on malaria incidence in Thailand within a Bayesian framework. A simulation was conducted based on ground truth of lagged effect curves representing the delayed relation with sparse malaria cases as seen in our study population. A case study to estimate the delayed effect of environmental variables was used with malaria incidence at a fine geographic scale of sub-districts in a western province of Thailand.

Results

From the simulation study, the model assumptions which accommodated both delayed effects and excessive zeros appeared to have the best overall performance across evaluation metrics and scenarios. The case study demonstrated lagged climatic effect estimation of the proposed modeling with real data. The models appeared to be useful to estimate the shape of association with malaria incidence.

Conclusions

A new method to estimate the spatiotemporal effect of climate on malaria trends in low transmission settings is presented. The developed methodology has potential to improve understanding and estimation of past and future trends in malaria incidence. With further development, this could assist policy makers with decisions on how to more effectively distribute resources and plan strategies for malaria elimination.

Spatial and spatio-temporal methods for mapping malaria risk: a systematic review

Background

Approaches in malaria risk mapping continue to advance in scope with the advent of geostatistical techniques spanning both the spatial and temporal domains. A substantive review of the merits of the methods and covariates used to map malaria risk has not been undertaken. Therefore, this review aimed to systematically retrieve, summarise methods and examine covariates that have been used for mapping malaria risk in sub-Saharan Africa (SSA).

Methods

A systematic search of malaria risk mapping studies was conducted using PubMed, EBSCOhost, Web of Science and Scopus databases. The search was restricted to refereed studies published in English from January 1968 to April 2020. To ensure completeness, a manual search through the reference lists of selected studies was also undertaken. Two independent reviewers completed each of the review phases namely: identification of relevant studies based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, data extraction and methodological quality assessment using a validated scoring criterion.

Results

One hundred and seven studies met the inclusion criteria. The median quality score across studies was 12/16 (range: 7–16). Approximately half (44%) of the studies employed variable selection techniques prior to mapping with rainfall and temperature selected in over 50% of the studies. Malaria incidence (47%) and prevalence (35%) were the most commonly mapped outcomes, with Bayesian geostatistical models often (31%) the preferred approach to risk mapping. Additionally, 29% of the studies employed various spatial clustering methods to explore the geographical variation of malaria patterns, with Kulldorf scan statistic being the most common. Model validation was specified in 53 (50%) studies, with partitioning data into training and validation sets being the common approach.

Conclusions

Our review highlights the methodological diversity prominent in malaria risk mapping across SSA. To ensure reproducibility and quality science, best practices and transparent approaches should be adopted when selecting the statistical framework and covariates for malaria risk mapping. Findings underscore the need to periodically assess methods and covariates used in malaria risk mapping; to accommodate changes in data availability, data quality and innovation in statistical methodology.

Malaria rapid diagnostic test (HRP2/pLDH) positivity, incidence, care accessibility and impact of community WASH Action programme in DR Congo: mixed method study involving 625 households

Background

Malaria is one of the most prevalent and deadliest illnesses in sub-Saharan Africa. Despite recent gains made towards its control, many African countries still have endemic malaria transmission. This study aimed to assess malaria burden at household level in Kongo central province, Democratic Republic of Congo (DRC), and the impact of community participatory Water, Sanitation and Hygiene (WASH) Action programme.

Methods

Mixed method research was conducted in two semi-rural towns, Mbanza-Ngungu (a WASH action site) and Kasangulu (a WASH control site) in DRC between 1 January 2017 through March 2018, involving 625 households (3,712 household members). Baseline and post-intervention malaria surveys were conducted with the use of World Bank/WHO Malaria Indicator Questionnaire. An action research consisting of a six-month study was carried out which comprised two interventions: a community participatory WASH action programme aiming at eliminating mosquito breeding areas in the residential environment and a community anti-malaria education campaign. The latter was implemented at both study sites. In addition, baseline and post-intervention malaria rapid diagnostic test (RDT) was performed among the respondents. Furthermore, a six-month hospital-based epidemiological study was conducted at selected referral hospitals at each site from 1 January through June 2017 to determine malaria trend.

Results

Long-lasting insecticide-treated net (LLIN) was the most commonly used preventive measure (55%); 24% of households did not use any measures. Baseline malaria survey showed that 96% of respondents (heads of households) reported at least one episode occurring in the previous six months; of them only 66.5% received malaria care at a health setting. In the Action Research, mean incident household malaria cases decreased significantly at WASH action site (2.3 ± 2.2 cases vs. 1.2 ± 0.7 cases, respectively; p < 0.05), whereas it remained unchanged at the Control site. Similar findings were observed with RDT results. Data collected from referral hospitals showed high malaria incidence rate, 67.4%. Low household income (ORa = 2.37; 95%CI: 1.05–3.12; p < 0.05), proximity to high risk area for malaria (ORa = 5.13; 95%CI: 2–29-8.07; p < 0.001), poor WASH (ORa = 4.10; 95%CI: 2.11–7.08; p < 0.001) were predictors of household malaria.

Conclusion

This research showed high prevalence of positive malaria RDT among the responders and high household malaria incidence, which were reduced by a 6-month WASH intervention. DRC government should scale up malaria control strategy by integrating efficient indoor and outdoor preventive measures and improve malaria care accessibility.

Addressing challenges in routine health data reporting in Burkina Faso through Bayesian spatiotemporal prediction of weekly clinical malaria incidence

Sub-Saharan African (SSA) countries’ health systems are often vulnerable to unplanned situations that can hinder their effectiveness in terms of data completeness and disease control. For instance, in Burkina Faso following a workers' strike, comprehensive data on several diseases were unavailable for a long period in 2019. Weather, seasonal-malaria-chemoprevention (SMC), free healthcare, and other contextual data, which are purported to influence malarial disease, provide opportunities to fit models to describe the clinical malaria data and predict the disease spread. Bayesian spatiotemporal modeling was applied to weekly malaria surveillance data from Burkina Faso (2011–2018) while considering the effects of weather, health programs and contextual factors. Then, a prediction was used to deal with weekly missing data for the entire year of 2019, and SMC and free healthcare effects were quantified. Our proposed model accurately predicted weekly clinical malaria incidence (correlation coefficient, r = 0.90). The distribution of clinical malaria incidence was heterogeneous across the country. Overall, national predicted clinical malaria incidence in 2019 (605 per 1000 [95% CrI: 360–990]) increased by 24.7% compared with the year 2015. SMC and the interaction between free healthcare and health facility attendance were associated with a reduction in clinical malaria incidence. Our modeling approach could be a useful tool for strengthening health systems’ resilience by addressing data completeness and could support SSA countries in developing appropriate targets and indicators to facilitate the subnational control effort.

Data-driven malaria prevalence prediction in large densely populated urban holoendemic sub-Saharan West Africa

Over 200 million malaria cases globally lead to half-million deaths annually. The development of malaria prevalence prediction systems to support malaria care pathways has been hindered by lack of data, a tendency towards universal "monolithic" models (one-size-fits-all-regions) and a focus on long lead time predictions. Current systems do not provide short-term local predictions at an accuracy suitable for deployment in clinical practice. Here we show a data-driven approach that reliably produces one-month-ahead prevalence prediction within a densely populated all-year-round malaria metropolis of over 3.5 million inhabitants situated in Nigeria which has one of the largest global burdens of P. falciparum malaria. We estimate one-month-ahead prevalence in a unique 22-years prospective regional dataset of > 9 × 10⁴ participants attending our healthcare services. Our system agrees with both magnitude and direction of the prediction on validation data achieving MAE ≤ 6 × 10^-2, MSE ≤ 7 × 10^-3, PCC (median 0.63, IQR 0.3) and with more than 80% of estimates within a (+ 0.1 to - 0.05) error-tolerance range which is clinically relevant for decision-support in our holoendemic setting. Our data-driven approach could facilitate healthcare systems to harness their own data to support local malaria care pathways.

Malaria predictions based on seasonal climate forecasts in South Africa: A time series distributed lag nonlinear model

Although there have been enormous demands and efforts to develop an early warning system for malaria, no sustainable system has remained. Well-organized malaria surveillance and high-quality climate forecasts are required to sustain a malaria early warning system in conjunction with an effective malaria prediction model. We aimed to develop a weather-based malaria prediction model using a weekly time-series data including temperature, precipitation, and malaria cases from 1998 to 2015 in Vhembe, Limpopo, South Africa and apply it to seasonal climate forecasts. The malaria prediction model performed well for short-term predictions (correlation coefficient, r > 0.8 for 1- and 2-week ahead forecasts). The prediction accuracy decreased as the lead time increased but retained fairly good performance (r > 0.7) up to the 16-week ahead prediction. The demonstration of the malaria prediction process based on the seasonal climate forecasts showed the short-term predictions coincided closely with the observed malaria cases. The weather-based malaria prediction model we developed could be applicable in practice together with skillful seasonal climate forecasts and existing malaria surveillance data. Establishing an automated operating system based on real-time data inputs will be beneficial for the malaria early warning system, and can be an instructive example for other malaria-endemic areas.

Understanding and using time series analyses in addiction research

Time series analyses are statistical methods used to assess trends in repeated measurements taken at regular intervals and their associations with other trends or events, taking account of the temporal structure of such data. Addiction research often involves assessing associations between trends in target variables (e.g. population cigarette smoking prevalence) and predictor variables (e.g. average price of a cigarette), known as a multiple time series design, or interventions or events (e.g. introduction of an indoor smoking ban), known as an interrupted time series design. There are many analytical tools available, each with its own strengths and limitations. This paper provides addiction researchers with an overview of many of the methods available (GLM, GLMM, GLS, GAMM, ARIMA, ARIMAX, VAR, SVAR, VECM) and guidance on when and how they should be used, sample size det ermination, reporting and interpretation. The aim is to provide increased clarity for researchers proposing to undertake these analyses concerning what is likely to be acceptable for publication in journals such as Addiction. Given the large number of choices that need to be made when setting up time series models, the guidance emphasizes the importance of pre-registering hypotheses and analysis plans before the analyses are undertaken.

Satellite Earth Observation Data in Epidemiological Modeling of Malaria, Dengue and West Nile Virus: A Scoping Review

Earth Observation (EO) data can be leveraged to estimate environmental variables that influence the transmission cycle of the pathogens that lead to mosquito-borne diseases (MBDs). The aim of this scoping review is to examine the state-of-the-art and identify knowledge gaps on the latest methods that used satellite EO data in their epidemiological models focusing on malaria, dengue and West Nile Virus (WNV). In total, 43 scientific papers met the inclusion criteria and were considered in this review. Researchers have examined a wide variety of methodologies ranging from statistical to machine learning algorithms. A number of studies used models and EO data that seemed promising and claimed to be easily replicated in different geographic contexts, enabling the realization of systems on regional and national scales. The need has emerged to leverage furthermore new powerful modeling approaches, like artificial intelligence and ensemble modeling and explore new and enhanced EO sensors towards the analysis of big satellite data, in order to develop accurate epidemiological models and contribute to the reduction of the burden of MBDs.

Weather-based forecasting of mosquito-borne disease outbreaks in Canada

Early warning systems to predict infectious disease outbreaks have been identified as a key adaptive response to climate change. Warming, climate variability and extreme weather events associated with climate change are expected to drive an increase in frequency and intensity of mosquito-borne disease (MBD) outbreaks globally. In Canada, this will mean an increased risk of endemic and emerging MBD outbreaks such as West Nile virus and other MBDs. The availability of timely information on the risk of impending MBD outbreaks has important public health implications, by allowing implementation of mosquito control measures and targeted communications regarding the need for increased personal protective measures-before an outbreak occurs. In Canada, both mechanistic and statistical weather-based models have been developed to predict West Nile virus outbreaks. These include models for different species of mosquitoes that transmit West Nile virus in different geographical areas of Canada. Although initial results have been promising, further validation and assessment of forecasting skill are needed before wide scale implementation. Weather-based forecasting for other emerging MBDs in Canada, such as Eastern equine encephalitis, may also be feasible.

Complexity-Based Spatial Hierarchical Clustering for Malaria Prediction

Targeted intervention and resource allocation are essential in effective control of infectious diseases, particularly those like malaria that tend to occur in remote areas. Disease prediction models can help support targeted intervention, particularly if they have fine spatial resolution. But, choosing an appropriate resolution is a difficult problem since choice of spatial scale can have a significant impact on accuracy of predictive models. In this paper, we introduce a new approach to spatial clustering for disease prediction we call complexity-based spatial hierarchical clustering. The technique seeks to find spatially compact clusters that have time series that can be well characterized by models of low complexity. We evaluate our approach with 2 years of malaria case data from Tak Province in northern Thailand. We show that the technique's use of reduction in Akaike information criterion (AIC) and Bayesian information criterion (BIC) as clustering criteria leads to rapid improvement in predictability and significantly better predictability than clustering based only on minimizing spatial intra-cluster distance for the entire range of cluster sizes over a variety of predictive models and prediction horizons.

Big Data's Role in Precision Public Health

Precision public health is an emerging practice to more granularly predict and understand public health risks and customize treatments for more specific and homogeneous subpopulations, often using new data, technologies, and methods. Big data is one element that has consistently helped to achieve these goals, through its ability to deliver to practitioners a volume and variety of structured or unstructured data not previously possible. Big data has enabled more widespread and specific research and trials of stratifying and segmenting populations at risk for a variety of health problems. Examples of success using big data are surveyed in surveillance and signal detection, predicting future risk, targeted interventions, and understanding disease. Using novel big data or big data approaches has risks that remain to be resolved. The continued growth in volume and variety of available data, decreased costs of data capture, and emerging computational methods mean big data success will likely be a required pillar of precision public health into the future. This review article aims to identify the precision public health use cases where big data has added value, identify classes of value that big data may bring, and outline the risks inherent in using big data in precision public health efforts.

The associations between malaria, interventions, and the environment: a systematic review and meta-analysis

BACKGROUND

Malaria transmission is driven by multiple factors, including complex and multifaceted connections between malaria transmission, socioeconomic conditions, climate and interventions. Forecasting models should account for all significant drivers of malaria incidence although it is first necessary to understand the relationship between malaria burden and the various determinants of risk to inform the development of forecasting models. In this study, the associations between malaria risk, environmental factors, and interventions were evaluated through a systematic review.

METHODS

Five electronic databases (CAB Abstracts, EMBASE, Global Health, MEDLINE and ProQuest Dissertations & Theses) were searched for studies that included both the effects of the environment and interventions on malaria within the same statistical model. Studies were restricted to quantitative analyses and health outcomes of malaria mortality or morbidity, outbreaks, or transmission suitability. Meta-analyses were conducted on a subset of results using random-effects models.

RESULTS

Eleven studies of 2248 potentially relevant articles that met inclusion criteria were identified for the systematic review and two meta-analyses based upon five results each were performed. Normalized Difference Vegetation Index was not found to be statistically significant associated with malaria with a pooled OR of 1.10 (95% CI 0.07, 1.71). Bed net ownership was statistically associated with decreasing risk of malaria, when controlling for the effects of environment with a pooled OR of 0.75 (95% CI 0.60, 0.95). In general, environmental effects on malaria, while controlling for the effect of interventions, were variable and showed no particular pattern. Bed nets ownership, use and distribution, have a significant protective effect while controlling for environmental variables.

CONCLUSIONS

There are a limited number of studies which have simultaneously evaluated both environmental and interventional effects on malaria risk. Poor statistical reporting and a lack of common metrics were important challenges for this review, which must be addressed to ensure reproducibility and quality research. A comprehensive or inclusive approach to identifying malaria determinants using standardized indicators would allow for a better understanding of its epidemiology, which is crucial to improve future malaria risk estimations.

Using remote sensing environmental data to forecast malaria incidence at a rural district hospital in Western Kenya

Malaria surveillance data provide opportunity to develop forecasting models. Seasonal variability in environmental factors correlate with malaria transmission, thus the identification of transmission patterns is useful in developing prediction models. However, with changing seasonal transmission patterns, either due to interventions or shifting weather seasons, traditional modelling approaches may not yield adequate predictive skill. Two statistical models,a general additive model (GAM) and GAMBOOST model with boosted regression were contrasted by assessing their predictive accuracy in forecasting malaria admissions at lead times of one to three months. Monthly admission data for children under five years with confirmed malaria at the Siaya district hospital in Western Kenya for the period 2003 to 2013 were used together with satellite derived data on rainfall, average temperature and evapotranspiration(ET). There was a total of 8,476 confirmed malaria admissions. The peak of malaria season changed and malaria admissions reduced overtime. The GAMBOOST model at 1-month lead time had the highest predictive skill during both the training and test periods and thus can be utilized in a malaria early warning system.

Integrating Environmental Monitoring and Mosquito Surveillance to Predict Vector-borne Disease: Prospective Forecasts of a West Nile Virus Outbreak

INTRODUCTION

Predicting the timing and locations of future mosquito-borne disease outbreaks has the potential to improve the targeting of mosquito control and disease prevention efforts. Here, we present and evaluate prospective forecasts made prior to and during the 2016 West Nile virus (WNV) season in South Dakota, a hotspot for human WNV transmission in the United States.

METHODS

We used a county-level logistic regression model to predict the weekly probability of human WNV case occurrence as a function of temperature, precipitation, and an index of mosquito infection status. The model was specified and fitted using historical data from 2004-2015 and was applied in 2016 to make short-term forecasts of human WNV cases in the upcoming week as well as whole-year forecasts of WNV cases throughout the entire transmission season. These predictions were evaluated at the end of the 2016 WNV season by comparing them with spatial and temporal patterns of the human cases that occurred.

RESULTS

There was an outbreak of WNV in 2016, with a total of 167 human cases compared to only 40 in 2015. Model results were generally accurate, with an AUC of 0.856 for short-term predictions. Early-season temperature data were sufficient to predict an earlier-than-normal start to the WNV season and an above-average number of cases, but underestimated the overall case burden. Model predictions improved throughout the season as more mosquito infection data were obtained, and by the end of July the model provided a close estimate of the overall magnitude of the outbreak.

CONCLUSIONS

An integrated model that included meteorological variables as well as a mosquito infection index as predictor variables accurately predicted the resurgence of WNV in South Dakota in 2016. Key areas for future research include refining the model to improve predictive skill and developing strategies to link forecasts with specific mosquito control and disease prevention activities.

Integrating malaria surveillance with climate data for outbreak detection and forecasting: the EPIDEMIA system

Background

Early indication of an emerging malaria epidemic can provide an opportunity for proactive interventions. Challenges to the identification of nascent malaria epidemics include obtaining recent epidemiological surveillance data, spatially and temporally harmonizing this information with timely data on environmental precursors, applying models for early detection and early warning, and communicating results to public health officials. Automated web-based informatics systems can provide a solution to these problems, but their implementation in real-world settings has been limited.

Methods

The Epidemic Prognosis Incorporating Disease and Environmental Monitoring for Integrated Assessment (EPIDEMIA) computer system was designed and implemented to integrate disease surveillance with environmental monitoring in support of operational malaria forecasting in the Amhara region of Ethiopia. A co-design workshop was held with computer scientists, epidemiological modelers, and public health partners to develop an initial list of system requirements. Subsequent updates to the system were based on feedback obtained from system evaluation workshops and assessments conducted by a steering committee of users in the public health sector.

Results

The system integrated epidemiological data uploaded weekly by the Amhara Regional Health Bureau with remotely-sensed environmental data freely available from online archives. Environmental data were acquired and processed automatically by the EASTWeb software program. Additional software was developed to implement a public health interface for data upload and download, harmonize the epidemiological and environmental data into a unified database, automatically update time series forecasting models, and generate formatted reports. Reporting features included district-level control charts and maps summarizing epidemiological indicators of emerging malaria outbreaks, environmental risk factors, and forecasts of future malaria risk.

Conclusions

Successful implementation and use of EPIDEMIA is an important step forward in the use of epidemiological and environmental informatics systems for malaria surveillance. Developing software to automate the workflow steps while remaining robust to continual changes in the input data streams was a key technical challenge. Continual stakeholder involvement throughout design, implementation, and operation has created a strong enabling environment that will facilitate the ongoing development, application, and testing of the system.

Referral pathways for patients with TIA avoiding hospital admission: a scoping review

OBJECTIVE

To identify the features and effects of a pathway for emergency assessment and referral of patients with suspected transient ischaemic attack (TIA) in order to avoid admission to hospital.

DESIGN

Scoping review.

DATA SOURCES

PubMed, CINAHL Web of Science, Scopus.

STUDY SELECTION

Reports of primary research on referral of patients with suspected TIA directly to specialist outpatient services.

DATA EXTRACTION

We screened studies for eligibility and extracted data from relevant studies. Data were analysed to describe setting, assessment and referral processes, treatment, implementation and outcomes.

RESULTS

8 international studies were identified, mostly cohort designs. 4 pathways were used by family doctors and 3 pathways by emergency department physicians. No pathways used by paramedics were found. Referrals were made to specialist clinic either directly or via a 24-hour helpline. Practitioners identified TIA symptoms and risk of further events using a checklist including the ABCD2 tool or clinical assessment. Antiplatelet medication was often given, usually aspirin unless contraindicated. Some patients underwent tests before referral and discharge. 5 studies reported reduced incident of stroke at 90 days, from 6-10% predicted rate to 1.3-2.1% actual rate. Between 44% and 83% of suspected TIA cases in these studies were referred through the pathways.

CONCLUSIONS

Research literature has focused on assessment and referral by family doctors and ED physicians to reduce hospitalisation of patients with TIA. No pathways for paramedical use were reported. We will use results of this scoping review to inform development of a paramedical referral pathway to be tested in a feasibility trial.

TRIAL REGISTRATION NUMBER

ISRCTN85516498. Stage: pre-results.

Analysing trends and forecasting malaria epidemics in Madagascar using a sentinel surveillance network: a web-based application

Background

The use of a malaria early warning system (MEWS) to trigger prompt public health interventions is a key step in adding value to the epidemiological data routinely collected by sentinel surveillance systems.

Methods

This study describes a system using various epidemic thresholds and a forecasting component with the support of new technologies to improve the performance of a sentinel MEWS. Malaria-related data from 21 sentinel sites collected by Short Message Service are automatically analysed to detect malaria trends and malaria outbreak alerts with automated feedback reports.

Results

Roll Back Malaria partners can, through a user-friendly web-based tool, visualize potential outbreaks and generate a forecasting model. The system already demonstrated its ability to detect malaria outbreaks in Madagascar in 2014.

Conclusion

This approach aims to maximize the usefulness of a sentinel surveillance system to predict and detect epidemics in limited-resource environments.

Assessing temporal associations between environmental factors and malaria morbidity at varying transmission settings in Uganda

Background

Environmental factors play a major role in transmission of malaria given their relationship to both the development and survival of the mosquito and parasite. The associations between environmental factors and malaria can be used to inform the development of early warning systems for increases in malaria burden. The objective of this study was to assess temporal relationships between rainfall, temperature and vegetation with malaria morbidity across three different transmission settings in Uganda.

Methods

Temporal relationships between environmental factors (weekly total rainfall, mean day time temperature and enhanced vegetation index series) and malaria morbidity (weekly malaria case count data and test positivity rate series) over the period January 2010–May 2013 in three sites located in varying malaria transmission settings in Uganda was explored using cross-correlation with pre-whitening. Sites included Kamwezi (low transmission), Kasambya (moderate transmission) and Nagongera (high transmission).

Results

Nagongera received the most rain (30.6 mm) and experienced, on average, the highest daytime temperatures (29.8 °C) per week. In the study period, weekly TPR and number of malaria cases were highest at Kasambya and lowest at Kamwezi. The largest cross-correlation coefficients between environmental factors and malaria morbidity for each site was 0.27 for Kamwezi (rainfall and cases), 0.21 for Kasambya (vegetation and TPR), and −0.27 for Nagongera (daytime temperature and TPR). Temporal associations between environmental factors (rainfall, temperature and vegetation) with malaria morbidity (number of malaria cases and TPR) varied by transmission setting. Longer time lags were observed at Kamwezi and Kasambya compared to Nagongera in the relationship between rainfall and number of malaria cases. Comparable time lags were observed at Kasambya and Nagongera in the relationship between temperature and malaria morbidity. Temporal analysis of vegetation with malaria morbidity revealed longer lags at Kasambya compared to those observed at the other two sites.

Conclusions

This study showed that temporal associations between environmental factors with malaria morbidity vary by transmission setting in Uganda. This suggests the need to incorporate local transmission differences when developing malaria early warning systems that have environmental predictors in Uganda. This will result in development of more accurate early warning systems, which are a prerequisite for effective malaria control in such a setting.

Remote Sensing-Driven Climatic/Environmental Variables for Modelling Malaria Transmission in Sub-Saharan Africa

Malaria is a serious public health threat in Sub-Saharan Africa (SSA), and its transmission risk varies geographically. Modelling its geographic characteristics is essential for identifying the spatial and temporal risk of malaria transmission. Remote sensing (RS) has been serving as an important tool in providing and assessing a variety of potential climatic/environmental malaria transmission variables in diverse areas. This review focuses on the utilization of RS-driven climatic/environmental variables in determining malaria transmission in SSA. A systematic search on Google Scholar and the Institute for Scientific Information (ISI) Web of Knowledge(SM) databases (PubMed, Web of Science and ScienceDirect) was carried out. We identified thirty-five peer-reviewed articles that studied the relationship between remotely-sensed climatic variable(s) and malaria epidemiological data in the SSA sub-regions. The relationship between malaria disease and different climatic/environmental proxies was examined using different statistical methods. Across the SSA sub-region, the normalized difference vegetation index (NDVI) derived from either the National Oceanic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR) or Moderate-resolution Imaging Spectrometer (MODIS) satellite sensors was most frequently returned as a statistically-significant variable to model both spatial and temporal malaria transmission. Furthermore, generalized linear models (linear regression, logistic regression and Poisson regression) were the most frequently-employed methods of statistical analysis in determining malaria transmission predictors in East, Southern and West Africa. By contrast, multivariate analysis was used in Central Africa. We stress that the utilization of RS in determining reliable malaria transmission predictors and climatic/environmental monitoring variables would require a tailored approach that will have cognizance of the geographical/climatic setting, the stage of malaria elimination continuum, the characteristics of the RS variables and the analytical approach, which in turn, would support the channeling of intervention resources sustainably.

Forecasting malaria in a highly endemic country using environmental and clinical predictors

Background

Malaria thrives in poor tropical and subtropical countries where local resources are limited. Accurate disease forecasts can provide public and clinical health services with the information needed to implement targeted approaches for malaria control that make effective use of limited resources. The objective of this study was to determine the relevance of environmental and clinical predictors of malaria across different settings in Uganda.

Methods

Forecasting models were based on health facility data collected by the Uganda Malaria Surveillance Project and satellite-derived rainfall, temperature, and vegetation estimates from 2006 to 2013. Facility-specific forecasting models of confirmed malaria were developed using multivariate autoregressive integrated moving average models and produced weekly forecast horizons over a 52-week forecasting period.

Results

The model with the most accurate forecasts varied by site and by forecast horizon. Clinical predictors were retained in the models with the highest predictive power for all facility sites. The average error over the 52 forecasting horizons ranged from 26 to 128% whereas the cumulative burden forecast error ranged from 2 to 22%.

Conclusions

Clinical data, such as drug treatment, could be used to improve the accuracy of malaria predictions in endemic settings when coupled with environmental predictors. Further exploration of malaria forecasting is necessary to improve its accuracy and value in practice, including examining other environmental and intervention predictors, including insecticide-treated nets.

Electronic supplementary material

The online version of this article (doi:10.1186/s12936-015-0758-4) contains supplementary material, which is available to authorized users.

Fuzzy association rule mining and classification for the prediction of malaria in South Korea

Background

Malaria is the world’s most prevalent vector-borne disease. Accurate prediction of malaria outbreaks may lead to public health interventions that mitigate disease morbidity and mortality.

Methods

We describe an application of a method for creating prediction models utilizing Fuzzy Association Rule Mining to extract relationships between epidemiological, meteorological, climatic, and socio-economic data from Korea. These relationships are in the form of rules, from which the best set of rules is automatically chosen and forms a classifier. Two classifiers have been built and their results fused to become a malaria prediction model. Future malaria cases are predicted as LOW, MEDIUM or HIGH, where these classes are defined as a total of 0–2, 3–16, and above 17 cases, respectively, for a region in South Korea during a two-week period. Based on user recommendations, HIGH is considered an outbreak.

Results

Model accuracy is described by Positive Predictive Value (PPV), Sensitivity, and F-score for each class, computed on test data not previously used to develop the model. For predictions made 7–8 weeks in advance, model PPV and Sensitivity are 0.842 and 0.681, respectively, for the HIGH classes. The F0.5 and F3 scores (which combine PPV and Sensitivity) are 0.804 and 0.694, respectively, for the HIGH classes. The overall FARM results (as measured by F-scores) are significantly better than those obtained by Decision Tree, Random Forest, Support Vector Machine, and Holt-Winters methods for the HIGH class. For the MEDIUM class, Random Forest and FARM obtain comparable results, with FARM being better at F0.5, and Random Forest obtaining a higher F3.

Conclusions

A previously described method for creating disease prediction models has been modified and extended to build models for predicting malaria. In addition, some new input variables were used, including indicators of intervention measures. The South Korea malaria prediction models predict LOW, MEDIUM or HIGH cases 7–8 weeks in the future. This paper demonstrates that our data driven approach can be used for the prediction of different diseases.

Using age, triage score, and disposition data from emergency department electronic records to improve Influenza-like illness surveillance

OBJECTIVE

Markers of illness severity are increasingly captured in emergency department (ED) electronic systems, but their value for surveillance is not known. We assessed the value of age, triage score, and disposition data from ED electronic records for predicting influenza-related hospitalizations.

MATERIALS AND METHODS

From June 2006 to January 2011, weekly counts of pneumonia and influenza (P&I) hospitalizations from five Montreal hospitals were modeled using negative binomial regression. Over lead times of 0-5 weeks, we assessed the predictive ability of weekly counts of 1) total ED visits, 2) ED visits with influenza-like illness (ILI), and 3) ED visits with ILI stratified by age, triage score, or disposition. Models were adjusted for secular trends, seasonality, and autocorrelation. Model fit was assessed using Akaike information criterion, and predictive accuracy using the mean absolute scaled error (MASE).

RESULTS

Predictive accuracy for P&I hospitalizations during non-pandemic years was improved when models included visits from patients ≥65 years old and visits resulting in admission/transfer/death (MASE of 0.64, 95% confidence interval (95% CI) 0.54-0.80) compared to overall ILI visits (0.89, 95% CI 0.69-1.10). During the H1N1 pandemic year, including visits from patients <18 years old, visits with high priority triage scores, or visits resulting in admission/transfer/death resulted in the best model fit.

DISCUSSION

Age and disposition data improved model fit and moderately reduced the prediction error for P&I hospitalizations; triage score improved model fit only during the pandemic year.

CONCLUSION

Incorporation of age and severity measures available in ED records can improve ILI surveillance algorithms.

Generalized seasonal autoregressive integrated moving average models for count data with application to malaria time series with low case numbers

INTRODUCTION

With the renewed drive towards malaria elimination, there is a need for improved surveillance tools. While time series analysis is an important tool for surveillance, prediction and for measuring interventions' impact, approximations by commonly used Gaussian methods are prone to inaccuracies when case counts are low. Therefore, statistical methods appropriate for count data are required, especially during "consolidation" and "pre-elimination" phases.

METHODS

Generalized autoregressive moving average (GARMA) models were extended to generalized seasonal autoregressive integrated moving average (GSARIMA) models for parsimonious observation-driven modelling of non Gaussian, non stationary and/or seasonal time series of count data. The models were applied to monthly malaria case time series in a district in Sri Lanka, where malaria has decreased dramatically in recent years.

RESULTS

The malaria series showed long-term changes in the mean, unstable variance and seasonality. After fitting negative-binomial Bayesian models, both a GSARIMA and a GARIMA deterministic seasonality model were selected based on different criteria. Posterior predictive distributions indicated that negative-binomial models provided better predictions than Gaussian models, especially when counts were low. The G(S)ARIMA models were able to capture the autocorrelation in the series.

CONCLUSIONS

G(S)ARIMA models may be particularly useful in the drive towards malaria elimination, since episode count series are often seasonal and non-stationary, especially when control is increased. Although building and fitting GSARIMA models is laborious, they may provide more realistic prediction distributions than do Gaussian methods and may be more suitable when counts are low.