Conditional predictive inference for online surveillance of spatial disease incidence

A Novel Bayesian Spatio-Temporal Surveillance Metric to Predict Emerging Infectious Disease Areas of High Disease Risk

Identification of areas of high disease risk has been one of the top goals for infectious disease public health surveillance. Accurate prediction of these regions leads to effective resource allocation and faster intervention. This paper proposes a novel prediction surveillance metric based on a Bayesian spatio-temporal model for infectious disease outbreaks. Exceedance probability, which has been commonly used for cluster detection in statistical epidemiology, was extended to predict areas of high risk. The proposed metric consists of three components: the area's risk profile, temporal risk trend, and spatial neighborhood influence. We also introduce a weighting scheme to balance these three components, which accommodates the characteristics of the infectious disease outbreak, spatial properties, and disease trends. Thorough simulation studies were conducted to identify the optimal weighting scheme and evaluate the performance of the proposed prediction surveillance metric. Results indicate that the area's own risk and the neighborhood influence play an important role in making a highly sensitive metric, and the risk trend term is important for the specificity and accuracy of prediction. The proposed prediction metric was applied to the COVID-19 case data of South Carolina from March 12, 2020, and the subsequent 30 weeks of data.

Evaluation of Bayesian spatiotemporal infectious disease models for prospective surveillance analysis

BACKGROUND

COVID-19 brought enormous challenges to public health surveillance and underscored the importance of developing and maintaining robust systems for accurate surveillance. As public health data collection efforts expand, there is a critical need for infectious disease modeling researchers to continue to develop prospective surveillance metrics and statistical models to accommodate the modeling of large disease counts and variability. This paper evaluated different likelihoods for the disease count model and various spatiotemporal mean models for prospective surveillance.

METHODS

We evaluated Bayesian spatiotemporal models, which are the foundation for model-based infectious disease surveillance metrics. Bayesian spatiotemporal mean models based on the Poisson and the negative binomial likelihoods were evaluated with the different lengths of past data usage. We compared their goodness of fit and short-term prediction performance with both simulated epidemic data and real data from the COVID-19 pandemic.

RESULTS

The simulation results show that the negative binomial likelihood-based models show better goodness of fit results than Poisson likelihood-based models as deemed by smaller deviance information criteria (DIC) values. However, Poisson models yield smaller mean square error (MSE) and mean absolute one-step prediction error (MAOSPE) results when we use a shorter length of the past data such as 7 and 3 time periods. Real COVID-19 data analysis of New Jersey and South Carolina shows similar results for the goodness of fit and short-term prediction results. Negative binomial-based mean models showed better performance when we used the past data of 52 time periods. Poisson-based mean models showed comparable goodness of fit performance and smaller MSE and MAOSPE results when we used the past data of 7 and 3 time periods.

CONCLUSION

We evaluate these models and provide future infectious disease outbreak modeling guidelines for Bayesian spatiotemporal analysis. Our choice of the likelihood and spatiotemporal mean models was influenced by both historical data length and variability. With a longer length of past data usage and more over-dispersed data, the negative binomial likelihood shows a better model fit than the Poisson likelihood. However, as we use a shorter length of the past data for our surveillance analysis, the difference between the Poisson and the negative binomial models becomes smaller. In this case, the Poisson likelihood shows robust posterior mean estimate and short-term prediction results.

Issues in Bayesian prospective surveillance of spatial health data

In this paper I review some of the major issues that arise when geo-referenced health data are to be the subject of prospective surveillance. The review focusses on modelbased approaches to this activity, and proposes the Bayesian paradigm as a convenient vehicle for modeling. Various posterior functional measures are discussed including the SCPO and SKL and a number of extensions to these are considered. Overall the value of Bayesian Hierarchical Modeling (BHM) with surveillance functionals is stressed in its relevance to early warning of adverse risk scenarios.

Bayesian spatiotemporal modeling with sliding windows to correct reporting delays for real-time dengue surveillance in Thailand

Background

The ability to produce timely and accurate estimation of dengue cases can significantly impact disease control programs. A key challenge for dengue control in Thailand is the systematic delay in reporting at different levels in the surveillance system. Efficient and reliable surveillance and notification systems are vital to monitor health outcome trends and early detection of disease outbreaks which vary in space and time.

Methods

Predicting the trend in dengue cases in real-time is a challenging task in Thailand due to a combination of factors including reporting delays. We present decision support using a spatiotemporal nowcasting model which accounts for reporting delays in a Bayesian framework with sliding windows. A case study is presented to demonstrate the proposed nowcasting method using weekly dengue surveillance data in Bangkok at district level in 2010.

Results

The overall real-time estimation accuracy was 70.69% with 59.05% and 79.59% accuracy during low and high seasons averaged across all weeks and districts. The results suggest the model was able to give a reasonable estimate of the true numbers of cases in the presence of delayed reports in the surveillance system. With sliding windows, models could also produce similar accuracy to estimation with the whole data.

Conclusions

A persistent challenge for the statistical and epidemiological communities is to transform data into evidence-based knowledge that facilitates policy making about health improvements and disease control at the individual and population levels. Improving real-time estimation of infectious disease incidence is an important technical development. The effort in this work provides a template for nowcasting in practice to inform decision making for dengue control.

Evaluating and implementing temporal, spatial, and spatio-temporal methods for outbreak detection in a local syndromic surveillance system

The New York City Department of Health and Mental Hygiene has operated an emergency department syndromic surveillance system since 2001, using temporal and spatial scan statistics run on a daily basis for cluster detection. Since the system was originally implemented, a number of new methods have been proposed for use in cluster detection. We evaluated six temporal and four spatial/spatio-temporal detection methods using syndromic surveillance data spiked with simulated injections. The algorithms were compared on several metrics, including sensitivity, specificity, positive predictive value, coherence, and timeliness. We also evaluated each method's implementation, programming time, run time, and the ease of use. Among the temporal methods, at a set specificity of 95%, a Holt-Winters exponential smoother performed the best, detecting 19% of the simulated injects across all shapes and sizes, followed by an autoregressive moving average model (16%), a generalized linear model (15%), a modified version of the Early Aberration Reporting System's C2 algorithm (13%), a temporal scan statistic (11%), and a cumulative sum control chart (<2%). Of the spatial/spatio-temporal methods we tested, a spatial scan statistic detected 3% of all injects, a Bayes regression found 2%, and a generalized linear mixed model and a space-time permutation scan statistic detected none at a specificity of 95%. Positive predictive value was low (<7%) for all methods. Overall, the detection methods we tested did not perform well in identifying the temporal and spatial clusters of cases in the inject dataset. The spatial scan statistic, our current method for spatial cluster detection, performed slightly better than the other tested methods across different inject magnitudes and types. Furthermore, we found the scan statistics, as applied in the SaTScan software package, to be the easiest to program and implement for daily data analysis.

Bayesian prospective detection of small area health anomalies using Kullback-Leibler divergence

Early detection of unusual health events depends on the ability to rapidly detect any substantial changes in disease, thus facilitating timely public health interventions. To assist public health practitioners to make decisions, statistical methods are adopted to assess unusual events in real time. We introduce a surveillance Kullback-Leibler measure for timely detection of disease outbreaks for small area health data. The detection methods are compared with the surveillance conditional predictive ordinate within the framework of Bayesian hierarchical Poisson modeling and applied to a case study of a group of respiratory system diseases observed weekly in South Carolina counties. Properties of the proposed surveillance techniques including timeliness and detection precision are investigated using a simulation study.

The application of a novel 'rising activity, multi-level mixed effects, indicator emphasis' (RAMMIE) method for syndromic surveillance in England

MOTIVATION

Syndromic surveillance is the real-time collection and interpretation of data to allow the early identification of public health threats and their impact, enabling public health action. The 'rising activity, multi-level mixed effects, indicator emphasis' method was developed to provide a single robust method enabling detection of unusual activity across a wide range of syndromes, nationally and locally.

RESULTS

The method is shown here to have a high sensitivity (92%) and specificity (99%) compared to previous methods, whilst halving the time taken to detect increased activity to 1.3 days.

AVAILABILITY AND IMPLEMENTATION

The method has been applied successfully to syndromic surveillance systems in England providing realistic models for baseline activity and utilizing prioritization rules to ensure a manageable number of 'alarms' each day.

CONTACT

roger.morbey@phe.gov.uk.

Prospective analysis of infectious disease surveillance data using syndromic information

In this paper, we describe a Bayesian hierarchical Poisson model for the prospective analysis of data for infectious diseases. The proposed model consists of two components. The first component describes the behavior of disease during nonepidemic periods and the second component represents the increase in disease counts due to the presence of an epidemic. A novelty of our model formulation is that the parameters describing the spread of epidemics are allowed to vary in both space and time. We also show how syndromic information can be incorporated into the model to provide a better description of the data and more accurate one-step-ahead forecasts. These real-time forecasts can be used to identify high-risk areas for outbreaks and, consequently, to develop efficient targeted surveillance. We apply the methodology to weekly emergency room discharges for acute bronchitis in South Carolina.

A case-association cluster detection and visualisation tool with an application to Legionnaires' disease

Statistical methods used in spatio-temporal surveillance of disease are able to identify abnormal clusters of cases but typically do not provide a measure of the degree of association between one case and another. Such a measure would facilitate the assignment of cases to common groups and be useful in outbreak investigations of diseases that potentially share the same source. This paper presents a model-based approach, which on the basis of available location data, provides a measure of the strength of association between cases in space and time and which is used to designate and visualise the most likely groupings of cases. The method was developed as a prospective surveillance tool to signal potential outbreaks, but it may also be used to explore groupings of cases in outbreak investigations. We demonstrate the method by using a historical case series of Legionnaires' disease amongst residents of England and Wales.

Prospective surveillance of multivariate spatial disease data

Surveillance systems are often focused on more than one disease within a predefined area. On those occasions when outbreaks of disease are likely to be correlated, the use of multivariate surveillance techniques integrating information from multiple diseases allows us to improve the sensitivity and timeliness of outbreak detection. In this article, we present an extension of the surveillance conditional predictive ordinate to monitor multivariate spatial disease data. The proposed surveillance technique, which is defined for each small area and time period as the conditional predictive distribution of those counts of disease higher than expected given the data observed up to the previous time period, alerts us to both small areas of increased disease incidence and the diseases causing the alarm within each area. We investigate its performance within the framework of Bayesian hierarchical Poisson models using a simulation study. An application to diseases of the respiratory system in South Carolina is finally presented.