Choosing the best algorithm for event detection based on the intended application: A conceptual framework for syndromic surveillance

Missingness and algorithmic bias: an example from the United States National Outbreak Reporting System, 2009-2019

Growing debates about algorithmic bias in public health surveillance lack specific examples. We tested a common assumption that exposure and illness periods coincide and demonstrated how algorithmic bias can arise due to missingness of critical information related to illness and exposure durations. We examined 9407 outbreaks recorded by the United States National Outbreak Reporting System (NORS) from January 1, 2009 through December 31, 2019 and detected algorithmic bias, a systematic over- or under-estimation of foodborne disease outbreak (FBDO) durations due to missing start and end dates. For 7037 (75%) FBDOs with complete date-time information, ~ 60% reported that the exposure period ended before the illness period started. For 2079 (87.7%) FBDOs with missing exposure dates, average illness durations were ~ 5.3 times longer (p < 0.001) than those with complete information, prompting the potential for algorithmic bias. Modern surveillance systems must be equipped with investigative capacities to examine and assess structural data missingness that can lead to bias.

Outbreak detection algorithms based on generalized linear model: a review with new practical examples

Public health surveillance serves a crucial function within health systems, enabling the monitoring, early detection, and warning of infectious diseases. Recently, outbreak detection algorithms have gained significant importance across various surveillance systems, particularly in light of the COVID-19 pandemic. These algorithms are approached from both theoretical and practical perspectives. The theoretical aspect entails the development and introduction of novel statistical methods that capture the interest of statisticians. In contrast, the practical aspect involves designing outbreak detection systems and employing diverse methodologies for monitoring syndromes, thus drawing the attention of epidemiologists and health managers. Over the past three decades, considerable efforts have been made in the field of surveillance, resulting in valuable publications that introduce new statistical methods and compare their performance. The generalized linear model (GLM) family has undergone various advancements in comparison to other statistical methods and models. This study aims to present and describe GLM-based methods, providing a coherent comparison between them. Initially, a historical overview of outbreak detection algorithms based on the GLM family is provided, highlighting commonly used methods. Furthermore, real data from Measles and COVID-19 are utilized to demonstrate examples of these methods. This study will be useful for researchers in both theoretical and practical aspects of outbreak detection methods, enabling them to familiarize themselves with the key techniques within the GLM family and facilitate comparisons, particularly for those with limited mathematical expertise.

Efficient algorithms for real-time syndromic surveillance

OBJECTIVE

Outbreaks of influenza-like diseases often cause spikes in the demand for hospital beds. Early detection of these outbreaks can enable improved management of hospital resources. The objective of this study was to test whether surveillance algorithms designed to be responsive to a wide range of anomalous decreases in the time between emergency department (ED) presentations with influenza-like illnesses provide efficient early detection of these outbreaks.

METHODS

Our study used data on ED presentations to major public hospitals in Queensland, Australia across 2017-2020. We developed surveillance algorithms for each hospital that flag potential outbreaks when the average time between successive ED presentations with influenza-like illnesses becomes anomalously small. We designed one set of algorithms to be responsive to a wide range of anomalous decreases in the time between presentations. These algorithms concurrently monitor three exponentially weighted moving averages (EWMAs) of the time between presentations and flag an outbreak when at least one EWMA falls below its control limit. We designed another set of algorithms to be highly responsive to narrower ranges of anomalous decreases in the time between presentations. These algorithms monitor one EWMA of the time between presentations and flag an outbreak when the EWMA falls below its control limit. Our algorithms use dynamic control limits to reflect that the average time between presentations depends on the time of year, time of day, and day of the week.

RESULTS

We compared the performance of the algorithms in detecting the start of two epidemic events at the hospital-level: the 2019 seasonal influenza outbreak and the early-2020 COVID-19 outbreak. The algorithm that concurrently monitors three EWMAs provided significantly earlier detection of these outbreaks than the algorithms that monitor one EWMA.

CONCLUSION

Surveillance algorithms designed to be responsive to a wide range of anomalous decreases in the time between ED presentations are highly efficient at detecting outbreaks of influenza-like diseases at the hospital level.

Evaluation of the application of sequence data to the identification of outbreaks of disease using anomaly detection methods

Anomaly detection methods have a great potential to assist the detection of diseases in animal production systems. We used sequence data of Porcine Reproductive and Respiratory Syndrome (PRRS) to define the emergence of new strains at the farm level. We evaluated the performance of 24 anomaly detection methods based on machine learning, regression, time series techniques and control charts to identify outbreaks in time series of new strains and compared the best methods using different time series: PCR positives, PCR requests and laboratory requests. We introduced synthetic outbreaks of different size and calculated the probability of detection of outbreaks (POD), sensitivity (Se), probability of detection of outbreaks in the first week of appearance (POD1w) and background alarm rate (BAR). The use of time series of new strains from sequence data outperformed the other types of data but POD, Se, POD1w were only high when outbreaks were large. The methods based on Long Short-Term Memory (LSTM) and Bayesian approaches presented the best performance. Using anomaly detection methods with sequence data may help to identify the emergency of cases in multiple farms, but more work is required to improve the detection with time series of high variability. Our results suggest a promising application of sequence data for early detection of diseases at a production system level. This may provide a simple way to extract additional value from routine laboratory analysis. Next steps should include validation of this approach in different settings and with different diseases.

Resurgence of Respiratory Syncytial Virus in Children: An Out-of-Season Epidemic in Portugal

INTRODUCTION

An out-of-season increase in respiratory syncytial virus (RSV) incidence was observed in Portugal from June 2021 onwards, revealing a continuing surge in cases throughout 2021/2022 autumn/winter. We aimed to describe this out-of-season epidemic and define its epidemic period, by analysing RSV incidence from week 40 of 2020 (2020-W40) to week 18 of 2022 (2022-W18).

MATERIAL AND METHODS

Surveillance data on weekly RSV laboratory confirmed cases, in Portugal, was used to monitor RSV incidence using CUSUM test methodology for count data.

RESULTS

In 2021-W23, the CUSUM score identified a significant increase in the risk of RSV. By that time, the percentage of RSV positive tests rose from 1% in 2021-W22 (3/265) to 6% in 2021-W23 (18/298). Despite a sharp decrease in RSV incidence on 2021-W33 and on 2022-W02, the CUSUM score stayed over the limit up to 2022-W07, indicating that the RSV activity remained at an epidemic level. Distinct peaks of RSV cases were observed between 2021-W30 and 2021-W32 (average of 77 RSV cases per week) and between 2021-W39 and 2021-W41 (average of 79 RSV cases per week) with positivity rates around 60%.

CONCLUSION

An out-of-season RSV epidemic was identified, with a longer epidemic period compared with previous seasons. Possible reasons include relaxation of COVID-19 physical distancing measures and a greater proportion of population susceptible to disease. As several factors may change the pattern of RSV activity, countries should implement year-round surveillance RSV surveillance systems. These findings might have an impact on public health planning regarding future RSV surges, namely, on the palivizumab prophylaxis period for high-risk infants.

Can North American animal poison control center call data provide early warning of outbreaks associated with contaminated pet food? Using the 2007 melamine pet food contamination incident as a case study

The 2007 melamine pet food contamination incident highlighted the need for enhanced reporting of toxicological exposures and development of a national quantitative disease surveillance system for companion animals. Data from poison control centers, such as the Animal Poison Control Center (APCC), may be useful for conducting real-time surveillance in this population. In this study, we explored the suitability of APCC call data for early warning of toxicological incidents in companion animal populations by using a-priori knowledge of the melamine-related nephrotoxicosis outbreak. Patient and household-level information regarding possible toxicological exposures in dogs and cats reported to the APCC from 2005 to 2007, inclusive, were extracted from the APCC's AnTox database. These data were used to examine the impact of surveillance outcome, statistical methodology, analysis level, and call source on the ability to detect the outbreak prior to the voluntary recall issued by the pet food manufacturer. Retrospective Poisson temporal scan tests were applied for each combination of outcome, method, level, and call source. The results showed that month-adjusted scans using syndromic data may have been able to help detect the outbreak up to two months prior to the voluntary recall although the success of these methods varied across call sources. We also demonstrated covariate month-adjustment can lead to vastly different results based on the surveillance outcome and call source to which it is applied. This illustrates care should be taken prior to arbitrarily selecting a surveillance outcome and statistical model for surveillance efforts and warns against ignoring the impacts of call source or key covariates when applying quantitative surveillance methods to APCC call data since these factors can lead to very different results. This study provides further evidence that APCC call data may be useful for conducting surveillance in the US companion animal population and further exploratory analyses and validation studies are warranted.

A novel machine learning-based approach for the detection and analysis of spontaneous synaptic currents

Spontaneous synaptic activity is a hallmark of biological neural networks. A thorough description of these synaptic signals is essential for understanding neurotransmitter release and the generation of a postsynaptic response. However, the complexity of synaptic current trajectories has either precluded an in-depth analysis or it has forced human observers to resort to manual or semi-automated approaches based on subjective amplitude and area threshold settings. Both procedures are time-consuming, error-prone and likely affected by human bias. Here, we present three complimentary methods for a fully automated analysis of spontaneous excitatory postsynaptic currents measured in major cell types of the mouse retina and in a primary culture of mouse auditory cortex. Two approaches rely on classical threshold methods, while the third represents a novel machine learning-based algorithm. Comparison with frequently used existing methods demonstrates the suitability of our algorithms for an unbiased and efficient analysis of synaptic signals in the central nervous system.

NextGen Public Health Surveillance and the Internet of Things (IoT)

Recent advances in technology have led to the rise of new-age data sources (e.g., Internet of Things (IoT), wearables, social media, and mobile health). IoT is becoming ubiquitous, and data generation is accelerating globally. Other health research domains have used IoT as a data source, but its potential has not been thoroughly explored and utilized systematically in public health surveillance. This article summarizes the existing literature on the use of IoT as a data source for surveillance. It presents the shortcomings of current data sources and how NextGen data sources, including the large-scale applications of IoT, can meet the needs of surveillance. The opportunities and challenges of using these modern data sources in public health surveillance are also explored. These IoT data ecosystems are being generated with minimal effort by the device users and benefit from high granularity, objectivity, and validity. Advances in computing are now bringing IoT-based surveillance into the realm of possibility. The potential advantages of IoT data include high-frequency, high volume, zero effort data collection methods, with a potential to have syndromic surveillance. In contrast, the critical challenges to mainstream this data source within surveillance systems are the huge volume and variety of data, fusing data from multiple devices to produce a unified result, and the lack of multidisciplinary professionals to understand the domain and analyze the domain data accordingly.

Development and validation of an automated emergency department-based syndromic surveillance system to enhance public health surveillance in Yukon: a lower-resourced and remote setting

BACKGROUND

Automated Emergency Department syndromic surveillance systems (ED-SyS) are useful tools in routine surveillance activities and during mass gathering events to rapidly detect public health threats. To improve the existing surveillance infrastructure in a lower-resourced rural/remote setting and enhance monitoring during an upcoming mass gathering event, an automated low-cost and low-resources ED-SyS was developed and validated in Yukon, Canada.

METHODS

Syndromes of interest were identified in consultation with the local public health authorities. For each syndrome, case definitions were developed using published resources and expert elicitation. Natural language processing algorithms were then written using Stata LP 15.1 (Texas, USA) to detect syndromic cases from three different fields (e.g., triage notes; chief complaint; discharge diagnosis), comprising of free-text and standardized codes. Validation was conducted using data from 19,082 visits between October 1, 2018 to April 30, 2019. The National Ambulatory Care Reporting System (NACRS) records were used as a reference for the inclusion of International Classification of Disease, 10th edition (ICD-10) diagnosis codes. The automatic identification of cases was then manually validated by two raters and results were used to calculate positive predicted values for each syndrome and identify improvements to the detection algorithms.

RESULTS

A daily secure file transfer of Yukon's Meditech ED-Tracker system data and an aberration detection plan was set up. A total of six syndromes were originally identified for the syndromic surveillance system (e.g., Gastrointestinal, Influenza-like-Illness, Mumps, Neurological Infections, Rash, Respiratory), with an additional syndrome added to assist in detecting potential cases of COVID-19. The positive predictive value for the automated detection of each syndrome ranged from 48.8-89.5% to 62.5-94.1% after implementing improvements identified during validation. As expected, no records were flagged for COVID-19 from our validation dataset.

CONCLUSIONS

The development and validation of automated ED-SyS in lower-resourced settings can be achieved without sophisticated platforms, intensive resources, time or costs. Validation is an important step for measuring the accuracy of syndromic surveillance, and ensuring it performs adequately in a local context. The use of three different fields and integration of both free-text and structured fields improved case detection.

Emergency Department Visits for Nonfatal Opioid Overdose During the COVID-19 Pandemic Across Six US Health Care Systems

Study objective

People with opioid use disorder are vulnerable to disruptions in access to addiction treatment and social support during the COVID-19 pandemic. Our study objective was to understand changes in emergency department (ED) utilization following a nonfatal opioid overdose during COVID-19 compared to historical controls in 6 healthcare systems across the United States.

Methods

Opioid overdoses were retrospectively identified among adult visits to 25 EDs in Alabama, Colorado, Connecticut, North Carolina, Massachusetts, and Rhode Island from January 2018 to December 2020. Overdose visit counts and rates per 100 all-cause ED visits during the COVID-19 pandemic were compared with the levels predicted based on 2018 and 2019 visits using graphical analysis and an epidemiologic outbreak detection cumulative sum algorithm.

Results

Overdose visit counts increased by 10.5% (n=3486; 95% confidence interval [CI] 4.18% to 17.0%) in 2020 compared with the counts in 2018 and 2019 (n=3020 and n=3285, respectively), despite a 14% decline in all-cause ED visits. Opioid overdose rates increased by 28.5% (95% CI 23.3% to 34.0%) from 0.25 per 100 ED visits in 2018 to 2019 to 0.32 per 100 ED visits in 2020. Although all 6 studied health care systems experienced overdose ED visit rates more than the 95th percentile prediction in 6 or more weeks of 2020 (compared with 2.6 weeks as expected by chance), 2 health care systems experienced sustained outbreaks during the COVID-19 pandemic.

Conclusion

Despite decreases in ED visits for other medical emergencies, the numbers and rates of opioid overdose-related ED visits in 6 health care systems increased during 2020, suggesting a widespread increase in opioid-related complications during the COVID-19 pandemic. Expanded community- and hospital-based interventions are needed to support people with opioid use disorder and save lives during the COVID-19 pandemic.

National Ambulance Surveillance System: A novel method using coded Australian ambulance clinical records to monitor self-harm and mental health-related morbidity

Self-harm and mental health are inter-related issues that substantially contribute to the global burden of disease. However, measurement of these issues at the population level is problematic. Statistics on suicide can be captured in national cause of death data collected as part of the coroner's review process, however, there is a significant time-lag in the availability of such data, and by definition, these sources do not include non-fatal incidents. Although survey, emergency department, and hospitalisation data present alternative information sources to measure self-harm, such data do not include the richness of information available at the point of incident. This paper describes the mental health and self-harm modules within the National Ambulance Surveillance System (NASS), a unique Australian system for monitoring and mapping mental health and self-harm. Data are sourced from paramedic electronic patient care records provided by Australian state and territory-based ambulance services. A team of specialised research assistants use a purpose-built system to manually scrutinise and code these records. Specific details of each incident are coded, including mental health symptoms and relevant risk indicators, as well as the type, intent, and method of self-harm. NASS provides almost 90 output variables related to self-harm (i.e., type of behaviour, self-injurious intent, and method) and mental health (e.g., mental health symptoms) in the 24 hours preceding each attendance, as well as demographics, temporal and geospatial characteristics, clinical outcomes, co-occurring substance use, and self-reported medical and psychiatric history. NASS provides internationally unique data on self-harm and mental health, with direct implications for translational research, public policy, and clinical practice. This methodology could be replicated in other countries with universal ambulance service provision to inform health policy and service planning.

Evaluating 5.5 Years of Equinella: A Veterinary-Based Voluntary Infectious Disease Surveillance System of Equines in Switzerland

Equine health is important in regard to trade, economy, society, and the veterinary, as well as public health. To reduce the burden of equine infectious diseases internationally, it is important to collect, review, and distribute equine health surveillance data as accurate and timely as possible. Within this study, we aimed at providing a comprehensive descriptive analysis of data submitted to Equinella, a voluntary veterinary-based surveillance system of non-notifiable equine infectious diseases and clinical signs, in Switzerland. This was achieved by reviewing the reports submitted since its relaunch in November 2013 and until April 2019, as well as assessing the data validity, activeness of participating veterinarians, coverage of the equine population, geographical representativeness, and timeliness of the system. In total, 630 reports have been submitted. Data validity ranged between 88.2 and 100%. The coverage of Equinella was assessed to be 50.8% of the Swiss equine population. Over the 5.5 years, of all 102 registered veterinarians, 67 (65.7%) submitted at least one report. On average, these veterinarians submitted 1.7 reports per year (median = 4 reports). More recently, in 2018, approximately only one-third [29 (28.4%)] of all registered veterinarians submitted at least one report. However, 59 (57.8%) have responded to the monthly reminder emails to confirm that they have not observed any relevant clinical case to be reported at least once (median number of confirmation per veterinarian = 9 of 12 reminder emails). The incidence of reports varied between cantons (member states of the Swiss confederation). The median timeliness of report submission was found to be 7 days. Overall, Equinella has been receiving reports since its initiation and contributed continuously to the surveillance of infectious diseases in the Swiss equine population and provided an output for the international equine community. Challenges encountered in achieving a higher number of submitted reports and increasing the coverage of the equine population, as well as the overall activeness of veterinarians, require further work. With our study, we provide a comprehensive overview of a veterinary-based voluntary surveillance system for equine health, assessed challenges of such, and suggest concrete improvements with transdisciplinary approaches for similar veterinary-based surveillance systems.

Glossary for public health surveillance in the age of data science

Public health surveillance is the ongoing systematic collection, analysis and interpretation of data, closely integrated with the timely dissemination of the resulting information to those responsible for preventing and controlling disease and injury. With the rapid development of data science, encompassing big data and artificial intelligence, and with the exponential growth of accessible and highly heterogeneous health-related data, from healthcare providers to user-generated online content, the field of surveillance and health monitoring is changing rapidly. It is, therefore, the right time for a short glossary of key terms in public health surveillance, with an emphasis on new data-science developments in the field.

Simulation Based Evaluation of Time Series for Syndromic Surveillance of Cattle in Switzerland

Choosing the syndrome time series to monitor in a syndromic surveillance system is not a straight forward process. Defining which syndromes to monitor in order to maximize detection performance has been recently identified as one of the research priorities in Syndromic surveillance. Estimating the minimum size of an epidemic that could potentially be detected in a specific syndrome could be used as a criteria for comparing the performance of different syndrome time series, and could provide some guidance for syndrome selection. The aim of our study was to estimate the potential value of different time series for building a national syndromic surveillance system for cattle in Switzerland. Simulations were used to produce outbreaks of different size and shape and to estimate the ability of each time series and aberration detection algorithm to detect them with high sensitivity, specificity and timeliness. Two temporal aberration detection algorithms were also compared: Holt-Winters generalized exponential smoothing (HW) and Exponential Weighted Moving Average (EWMA). Our results indicated that a specific aberration detection algorithm should be used for each time series. In addition, time series with high counts per unit of time had good overall detection performance, but poor detection performance for small epidemics making them of limited use for an early detection system. Estimating the minimum size of simulated epidemics that could potentially be detected in syndrome TS-event detection pairs can help surveillance system designers choosing the most appropriate syndrome TS to include in their early epidemic surveillance system.

Resumeq: A Novel Way of Monitoring Equine Diseases Through the Centralization of Necropsy Data

The French surveillance network for causes of equine mortality (Resumeq) was created in 2015 for the qualitative surveillance of equine mortality through the centralization in a national database of necropsy data and their subsequent epidemiological analysis. It was designed to identify the causes of equine mortality, monitor their evolution over time and space, and detect emerging diseases as early as possible. Resumeq is an event-based surveillance system involving various players and structures. It is organized around a steering body, a scientific and technical support committee and a coordination unit. Different tools have been developed specifically for Resumeq. These include standardized necropsy protocols, a thesaurus for the anatomopathological terms and the causes of equine death, and an interactive web application so that network contributors can display data analysis results. The four French veterinary schools, seventeen veterinary laboratories, and ten veterinary clinics already contribute to the production and centralization of standardized data. To date, the data from around 1,000 equine necropsies have been centralized. While most deaths were located in western France, the geographic coverage is gradually improving. Data analysis allows the main causes of death to be ranked and major threats identified on a local, regional or national level. Initial results demonstrate the feasibility and benefits of this national surveillance tool. Moreover, in the future, this surveillance could take an international dimension if several countries decided to jointly capitalize on their necropsy data.

A systematic review of aberration detection algorithms used in public health surveillance

The algorithms used for detecting anomalies have evolved substantially over the last decade to take advantage of advances in informatics and to accommodate changes in surveillance data. We identified 145 studies since 2007 that evaluated statistical methods used to detect aberrations in public health surveillance data. For each study, we classified the analytic methods and reviewed the evaluation metrics. We also summarized the practical usage of the detection algorithms in public health surveillance systems worldwide. Traditional methods (e.g., control charts, linear regressions) were the focus of most evaluation studies and continue to be used commonly in practice. There was, however, an increase in the number of studies using forecasting methods and studies applying machine learning methods, hidden Markov models, and Bayesian framework to multivariate datasets. Evaluation studies demonstrated improved accuracy with more sophisticated methods, but these methods do not appear to be used widely in public health practice.