Event Abstract

Back to Event

Development of spatio-temporal risk analytics for enhancing event-based surveillance systems

Julie Simon1, Victoria Ng2, 3, Jean Philippe Gilbert3, Kamran Khan4, 5, 6 and Erin E. Rees1, 3*
  • 1 Public Health Risk Sciences Division, National Microbiology Laboratory, Public Health Agency of Canada, Canada
  • 2 Public Health Risk Sciences Division, National Microbiology Laboratory, Public Health Agency of Canada, Canada
  • 3 Department of Geography, Faculty of Forestry, Geography and Geomatics, Laval University, Canada
  • 4 Other, Canada
  • 5 St. Michael's Hospital, Canada
  • 6 Department of Medicine, University of Toronto, Canada

A new generation of health intelligence surveillance systems are being developed to find evidence of emerging health threats. The approach focuses on integrating data analytics into event-based surveillance (EBS) systems [1]. Event-based surveillance systems monitor open-source information on the internet, such as media reports and social media, for evidence of emerging health threats. Recent technological advances using machine learning (ML) techniques are increasingly exploiting internet data to detect potential health threats [2]. EBS systems use human moderation, automation, or a combination of both techniques to review, validate, and assess threat levels of detected events. Human moderated events have a higher signal-to-noise ratio, however moderators can bias results and do not necessarily have the resources to provide a detailed risk assessment [3]. Fully moderated systems process internet data more quickly and at a lower cost but result in a lower signal-to-noise ratio [4]. Despite the advances in ML to increase automation and improve signal detection, efforts to integrate downstream analytics to assess whether detected threats warrant attention have not kept pace. We are developing a new EBS system strategy that includes data analytics for risk assessment of detected events. The Internet-based Surveillance Informing Global Health Threats (InSIGHT) module is designed as a system of connected components (Figure 1). In brief, unstructured open-source internet data flows in and gets ingested. ML algorithms provide structure to unstructured data by extracting information about what happened, where and when and storing the data in the article database. An additional database, the contextual information database, contains other relevant information for feeding into risk assessment models. Data can be extracted from both the article and contextual databases to produce information and knowledge for situational awareness and risk assessment, as displayed in a dashboard. A key question when assessing the risk of emerging infectious disease threats detected from EBS systems are whether the threat will spread to other susceptible populations. With globalization, spatio-temporal spread patterns of many infectious diseases can be explained using commercial air traffic data, for example, for Zika virus (ZIKV) [5], influenza A (H1N1) [6] and severe acute respiratory illness (SARS) [7]. The probability of pathogens spreading to naïve populations are commonly estimated using susceptible-exposed-infected-removed (SEIR) compartmental models [8][9]. This approach integrates passenger volume data to simulate the flow of passengers through the travel network in the SEIR health states. Though this metapopulation approach has proved valuable to retrospectively explain spatio-temporal patterns in infectious disease outbreaks, it would be difficult to harness in an EBS system. When EBS systems first detect the emergence of an infectious disease threat there may not be sufficient information to parameterize the SEIR disease dynamics, specifically being the transmission rate, delay from being exposed to infectious, and the recovery rate. Furthermore, air traffic data suppliers may not provide data in real-time but in a delayed release up to one month. We will showcase InSIGHT as a module that can be integrated into EBS systems to assess the risk that detected infectious disease events will impact populations and spread to other locations. Specifically, we will 1) describe the InSIGHT system architecture for the flow of information, and 2) present our process for developing a data analytic risk assessment tool that can enhance information gained from EBS systems when little is known about the disease dynamics. Our tool will use air traffic data to inform the risk of travel-related cases of infectious diseases acquired outside of the traveller’s home country. We demonstrate the tool using Zika virus (ZIKV) infection as the study system and show preliminary results. Methods We will present system architecture of the InSIGHT module, including the flow of information to and from the model that estimates infectious disease spread patterns using air traffic data. To develop the modeling tool we use monthly air traffic passenger-level flight itinerary data for 2015 from the International Air for Transport Association (IATA) and for 2016 from BlueDot’s Explorer (BE), a proprietary web platform allowing access to IATA data via matrices of monthly travel volume between paired airports. These data result in a list of 3 691 to 3 778 airports connected by 2 269 669 to 2 780 009 routes depending on the month, and capture approximately 90% of all commercial air travel between 163 countries. To estimate the potential of infected air travellers to spread ZIKV we calculated monthly expected force (ExF), developed by Lawyer (2015), from each airport of American countries between May 2015 and December 2016. The ExF of an airport is defined as the expectation of the force of infection generated by an epidemic process seeded from the airport into an otherwise susceptible air travel network, after two transmission events and no recovery [10]. We are also developing multiple versions of ExF that scale data up from the airport level to get a sense of the force of infection generated by a country as a whole. As a first step, we assess for temporal variations in air traffic passenger volume data and the multiple versions of ExF to determine if time-lagged ExF values are representative of current traffic volumes. As a second step we are developing regression models to determine the power of ExF to explain the number of ZIKV infected travellers returning to Canada, as compared to the predictive power of the basic reproduction number (R0) of ZIKV previously estimated for countries in Americas over the same time period [11]. To strengthen the predictive power of the ExF metrics we are also exploring the inclusion of other variables known to influence ZIKV exposure and transmission dynamics (i.e. climate data and socioeconomic data) in the regression modeling framework. Results ExF metrics were estimated for 46 countries of Americas from which Canadian travellers were tested for the ZIKV infection. Of the 6 548 travellers tested between May 2015 and December 2016, 324 (4.95%) tested positive for ZIKV infection. Analysis of temporal variations in air traffic passenger volume data and ExF showed significant change in the mean and variance for 18 countries (Figure 2). Preliminary analyses suggest that ExF metrics are a better predictor of the proportion of travel-related cases of ZIKV infection compared to passenger volumes. Further analysis is determining which form of the country-level ExF metric predicts best, and whether this metric can be used in place of R0 for predicting the number of travellers returning to Canada infected with ZIKV (Figure 3). Conclusions Event-based surveillance systems can notify of emerging infectious disease threats days to over a week in advance of official notification systems [12][13]. Given the increasing volume, variety and velocity of open-source internet data, a challenge of EBS systems is being able to efficiently assess the risk of detected threats. This study presents data analytical strategies for next generation of EBS systems as showcased through InSIGHT. We are finding that simple tools using data about human mobility can inform on health impacts and spread patterns in countries experiencing an outbreak of ZIKV. We continue to explore the potential for ExF metrics to inform about the risk variations between ZIKV-affected countries, and to predict rates of ZIKV spread at the international level. Our ongoing research will also explore the relevance of the using ExF to inform the risk for other infectious diseases. Ultimately, our InSIGHT module will include risk modeling tools, such as demonstrated in this study, that can provide accurate information sooner, and when not much is known about the infectious disease threat, and thus strengthen the ability of response systems to prepare mitigation strategies.

Figure 1
Image
Figure 2
Image
Figure 3
Image

Acknowledgements

This project is funded by Canadian Security and Safety Program of the Department of National Defence.

References

1. Șerban O, Thapen N, Maginnis B, Hankin C, Foot V (2018) Real-time processing of social media with SENTINEL: A syndromic surveillance system incorporating deep learning for health classification. Inf Process Manag. doi:10.1016/j.ipm.2018.04.011. 2. Rees E, Ng V, Gachon P, Mawudeku A, McKenney D, et al. (2019) Risk assessment strategies for early detection and prediction of infectious disease outbreaks associated with climate change. Canada Commun Dis Rep 45: 119–126. 3. Carrion M, Madoff LC (2017) ProMED-mail: 22 years of digital surveillance of emerging infectious diseases. Int Health 9: 177–183. doi:10.1093/inthealth/ihx014. 4. Belyaeva J, Linge JP (2476) How to maximise event-based surveillance web-systems: the example of ECDC/JRC collaboration to improve the performance of MedISys. Luxembourg. 1–22 p. Available: http://ipsc.jrc.ec.europa.eu/. 5. Zhang Q, Sun K, Chinazzi M, Pastore y Piontti A, Dean NE, et al. (2017) Spread of Zika virus in the Americas. Proc Natl Acad Sci 114: E4334–E4343. Available: http://www.pnas.org/lookup/doi/10.1073/pnas.1620161114. 6. Khan K, Arino J, Hu W, Raposo P, Sears J, et al. (2009) Spread of a Novel Influenza A (H1N1) Virus via Global Airline Transportation. N Engl J Med 361: 212–214. doi:10.1056/nejmc0904559. 7. Ruan S, Wang W, Levin SA (2005) The effect of global travel on the spread of SARS. Math Biosci Eng 3: 205–218. Available: http://www.aimsciences.org/journals/displayArticles.jsp?paperID=1714. 8. Balcan D, Colizza V, Goncalves B, Hu H, Ramasco JJ, et al. (2009) Multiscale mobility networks and the large scale spreading of infectious diseases. Available: http://arxiv.org/abs/0907.3304. 9. Colizza V, Barrat A, Barthelemy M, Vespignani A (2006) The role of the airline transportation network in the prediction and predictability of global epidemics. Proc Natl Acad Sci 103: 2015–2020. Available: http://www.pnas.org/cgi/doi/10.1073/pnas.0510525103. 10. Lawyer G (2015) Understanding the influence of all nodes in a network. Sci Rep 5: 1–9. doi:10.1038/srep08665. 11. Ogden NH, Fazil A, Safronetz D, Drebot M, Wallace J, et al. (2017) Risk of travel-related cases of Zika virus infection is predicted by transmission intensity. Parasit Vectors 10: 1–9. Available: http://dx.doi.org/10.1186/s13071-017-1977-z. 12. Barboza P, Vaillant L, Mawudeku A, Nelson NP, Hartley DM, et al. (2013) Evaluation of Epidemic Intelligence Systems Integrated in the Early Alerting and Reporting Project for the Detection of A/H5N1 Influenza Events. PLoS One 8: 1–9. doi:10.1371/journal.pone.0057252. 13. Keller M, Blench M, Tolentino H, Freifeld CC, Mandl KD, et al. (2009) Use of unstructured event-based reports for global infectious disease surveillance. Emerg Infect Dis 15: 689–695. doi:10.3201/eid1505.081114.

Keywords: Infectious disease spread, Human mobility, Air traffic networks, health intelligence surveillance systems, Event-based surveillance systems, Risk modeling., Zika virus

Conference: GeoVet 2019. Novel spatio-temporal approaches in the era of Big Data, Davis, United States, 8 Oct - 10 Oct, 2019.

Presentation Type: Senior oral presentation

Topic: Spatio-temporal surveillance and modeling approaches

Citation: Simon J, Ng V, Gilbert J, Khan K and Rees EE (2019). Development of spatio-temporal risk analytics for enhancing event-based surveillance systems. Front. Vet. Sci. Conference Abstract: GeoVet 2019. Novel spatio-temporal approaches in the era of Big Data. doi: 10.3389/conf.fvets.2019.05.00044

Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.

The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.

Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.

For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.

Received: 31 May 2019; Published Online: 27 Sep 2019.

* Correspondence: Dr. Erin E Rees, Public Health Risk Sciences Division, National Microbiology Laboratory, Public Health Agency of Canada, Saint-Hyacinthe, Canada, erin.rees@canada.ca