Model training periods impact estimation of COVID-19 incidence from wastewater viral loads

Predicting COVID-19 in Ohio: Insights from wastewater, demographic and socioeconomic data

More than four years into the COVID-19 pandemic, clear patterns have emerged showing that the virus does not affect all populations uniformly. Demographic and socioeconomic disparities play a significant role in the vulnerability to and spread of SARS-CoV-2. Analyzing these disparities can offer insights into the pandemic's dynamics, helping to identify critical factors that need to be addressed in efforts to mitigate the pandemic's impact globally. Wastewater-based surveillance (WBS), a crucial tool for tracking the virus, offers a unique perspective on how socioeconomic and demographic factors might influence infection rates across different communities. However, estimating and predicting the extent of the epidemic from WBS results is still challenging. In our study, we tried to address these challenges by analyzing data from 55 sites in Ohio, USA, with populations ranging from 3300 to 654,817, to better understand the pandemic's dynamics and WBS effectiveness in monitoring COVID-19 spread. Factors such as population size, poverty rate, racial demographics (specifically white and black populations), and median income showed the strongest correlations with both clinical cases and wastewater results, with population size being the most important factor. Moreover, among eight evaluated machine learning models, k-Nearest Neighbors (R2 = 0.873), Random Forest (R2 = 0.862), and XGBoost (R2 = 0.854) were the most effective in predicting clinical cases from WBS data across demographic and socioeconomic categories, while Linear (R2 = 0.578) and Ridge+Linear (R2 = 0.595) were least effective. Thus, these findings highlight the potential of machine learning to predict COVID-19 cases from WBS data across a wide range of demographic and socioeconomic categories.

Equity-centered adaptive sampling in sub-sewershed wastewater surveillance using census data

Sub-city, or sub-sewershed, wastewater monitoring for infectious diseases offers a data-driven strategy to inform local public health response and complements city-wide data from centralized wastewater treatment plants. Developing strategies for equitable representation of diverse populations in sub-city wastewater sampling frameworks is complicated by misalignment between demographic data and sampling zones. We address this challenge by: (1) developing a geospatial analysis tool that probabilistically assigns demographic data for subgroups aggregated by race and age from census blocks to sub-city sampling zones; (2) evaluating representativeness of subgroup populations for COVID-19 wastewater-based disease surveillance in Davis, California; and (3) demonstrating scenario planning that prioritizes vulnerable populations. We monitored SARS-CoV-2 in wastewater as a proxy for COVID-19 incidence in Davis (November 2021-September 2022). Daily city-wide sampling and thrice-weekly sub-city sampling from 16 maintenance holes covered nearly the entire city population. Sub-city wastewater data, aggregated as a population-weighted mean, correlated strongly with centralized treatment plant data (Spearman's correlation 0.909). Probabilistic assignment of demographic data can inform decisions when adapting sampling locations to prioritize vulnerable groups. We considered four scenarios that reduced the number of sampling zones from baseline by 25% and 50%, chosen randomly or to prioritize coverage of >65-year-old populations. Prioritizing representation increased coverage of >65-year-olds from 51.1% to 67.2% when removing half the zones, while increasing coverage of Black or African American populations from 67.5% to 76.7%. Downscaling had little effect on correlations between sub-city and centralized data (Spearman's correlations ranged from 0.875 to 0.917), with strongest correlations observed when prioritizing coverage of >65-year-old populations.

A modeling pipeline to relate municipal wastewater surveillance and regional public health data

As COVID-19 becomes endemic, public health departments benefit from improved passive indicators, which are independent of voluntary testing data, to estimate the prevalence of COVID-19 in local communities. Quantification of SARS-CoV-2 RNA from wastewater has the potential to be a powerful passive indicator. However, connecting measured SARS-CoV-2 RNA to community prevalence is challenging due to the high noise typical of environmental samples. We have developed a generalized pipeline using in- and out-of-sample model selection to test the ability of different correction models to reduce the variance in wastewater measurements and applied it to data collected from treatment plants in the Chicago area. We built and compared a set of multi-linear regression models, which incorporate pepper mild mottle virus (PMMoV) as a population biomarker, Bovine coronavirus (BCoV) as a recovery control, and wastewater system flow rate into a corrected estimate for SARS-CoV-2 RNA concentration. For our data, models with BCoV performed better than those with PMMoV, but the pipeline should be used to reevaluate any new data set as the sources of variance may change across locations, lab methods, and disease states. Using our best-fit model, we investigated the utility of RNA measurements in wastewater as a leading indicator of COVID-19 trends. We did this in a rolling manner for corrected wastewater data and for other prevalence indicators and statistically compared the temporal relationship between new increases in the wastewater data and those in other prevalence indicators. We found that wastewater trends often lead other COVID-19 indicators in predicting new surges.

Beyond linear regression: Modeling COVID-19 clinical cases with wastewater surveillance of SARS-CoV-2 for the city of Athens and Ohio University campus

Wastewater-based surveillance has emerged as a detection tool for population-wide infectious diseases, including coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Infected individuals shed the virus, which can be detected in wastewater using molecular techniques such as reverse transcription-digital polymerase chain reaction (RT-dPCR). This study examined the association between the number of clinical cases and the concentration of SARS-CoV-2 in wastewater beyond linear regression and for various normalizations of viral loads. Viral loads were measured in a total of 446 wastewater samples during the period from August 2021 to April 2022. These samples were collected from nine different locations, with 220 samples taken from four specific sites within the city of Athens and 226 samples from five sites within Ohio University. The correlation between COVID-19 cases and wastewater viral concentrations, which was estimated using the Pearson correlation coefficient, was statistically significant and ranged from 0.6 to 0.9. In addition, time-lagged cross correlation was applied to identify the lag time between clinical and wastewater data, estimated 4 to 7 days. While we also explored the effect on the correlation coefficients of various normalizations of viral loads accounting for procedural loss or amount of fecal material and of estimated lag times, these alternative specifications did not change our substantive conclusions. Additionally, several linear and non-linear regression models were applied to predict the COVID-19 cases given wastewater data as input. The non-linear approach was found to yield the highest R-squared and Pearson correlation and lowest Mean Absolute Error values between the predicted and actual number of COVID-19 cases for both aggregated OHIO Campus and city data. Our results provide support for previous studies on correlation and time lag and new evidence that non-linear models, approximated with artificial neural networks, should be implemented for WBS of contagious diseases.

Bayesian sequential approach to monitor COVID-19 variants through test positivity rate from wastewater

Deployment of clinical testing on a massive scale was an essential control measure for curtailing the burden of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections and the magnitude of the COVID-19 (coronavirus disease 2019) pandemic during its waves. As the pandemic progressed, new preventive and surveillance mechanisms emerged. Implementation of vaccine programs, wastewater (WW) surveillance, and at-home COVID-19 antigen tests reduced the demand for mass SARS-CoV-2 testing. Unfortunately, reductions in testing and test reporting rates also reduced the availability of public health data to support decision-making. This paper proposes a sequential Bayesian approach to estimate the COVID-19 test positivity rate (TPR) using SARS-CoV-2 RNA concentrations measured in WW through an adaptive scheme incorporating changes in virus dynamics. The proposed modeling framework was applied to WW surveillance data from two WW treatment plants in California; the City of Davis and the University of California, Davis campus. TPR estimates are used to compute thresholds for WW data using the Centers for Disease Control and Prevention thresholds for low (<5% TPR), moderate (5%-8% TPR), substantial (8%-10% TPR), and high (>10% TPR) transmission. The effective reproductive number estimates are calculated using TPR estimates from the WW data. This approach provides insights into the dynamics of the virus evolution and an analytical framework that combines different data sources to continue monitoring COVID-19 trends. These results can provide public health guidance to reduce the burden of future outbreaks as new variants continue to emerge. IMPORTANCE We propose a statistical model to correlate WW with TPR to monitor COVID-19 trends and to help overcome the limitations of relying only on clinical case detection. We pose an adaptive scheme to model the nonautonomous nature of the prolonged COVID-19 pandemic. The TPR is modeled through a Bayesian sequential approach with a beta regression model using SARS-CoV-2 RNA concentrations measured in WW as a covariable. The resulting model allows us to compute TPR based on WW measurements and incorporates changes in viral transmission dynamics through an adaptive scheme.

Patterns of SARS-CoV-2 circulation revealed by a nationwide sewage surveillance programme, the Netherlands, August 2020 to February 2022

BackgroundSurveillance of SARS-CoV-2 in wastewater offers a near real-time tool to track circulation of SARS-CoV-2 at a local scale. However, individual measurements of SARS-CoV-2 in sewage are noisy, inherently variable and can be left-censored.AimWe aimed to infer latent virus loads in a comprehensive sewage surveillance programme that includes all sewage treatment plants (STPs) in the Netherlands and covers 99.6% of the Dutch population.MethodsWe applied a multilevel Bayesian penalised spline model to estimate time- and STP-specific virus loads based on water flow-adjusted SARS-CoV-2 qRT-PCR data for one to four sewage samples per week for each of the more than 300 STPs.ResultsThe model captured the epidemic upsurges and downturns in the Netherlands, despite substantial day-to-day variation in the measurements. Estimated STP virus loads varied by more than two orders of magnitude, from ca 1012 virus particles per 100,000 persons per day in the epidemic trough in August 2020 to almost 1015 per 100,000 in many STPs in January 2022. The timing of epidemics at the local level was slightly shifted between STPs and municipalities, which resulted in less pronounced peaks and troughs at the national level.ConclusionAlthough substantial day-to-day variation is observed in virus load measurements, wastewater-based surveillance of SARS-CoV-2 that is performed at high sampling frequency can track long-term progression of an epidemic at a local scale in near real time.

Wastewater-based modeling, reconstruction, and prediction for COVID-19 outbreaks in Hungary caused by highly immune evasive variants

(MOTIVATION)

Wastewater-based epidemiology (WBE) has emerged as a promising approach for monitoring the COVID-19 pandemic, since the measurement process is cost-effective and is exposed to fewer potential errors compared to other indicators like hospitalization data or the number of detected cases. Consequently, WBE was gradually becoming a key tool for epidemic surveillance and often the most reliable data source, as the intensity of clinical testing for COVID-19 drastically decreased by the third year of the pandemic. Recent results suggests that the model-based fusion of wastewater measurements with clinical data and other indicators is essential in future epidemic surveillance.

(METHOD)

In this work, we developed a wastewater-based compartmental epidemic model with a two-phase vaccination dynamics and immune evasion. We proposed a multi-step optimization-based data assimilation method for epidemic state reconstruction, parameter estimation, and prediction. The computations make use of the measured viral load in wastewater, the available clinical data (hospital occupancy, delivered vaccine doses, and deaths), the stringency index of the official social distancing rules, and other measures. The current state assessment and the estimation of the current transmission rate and immunity loss allow a plausible prediction of the future progression of the pandemic.

(RESULTS)

Qualitative and quantitative evaluations revealed that the contribution of wastewater data in our computational epidemiological framework makes predictions more reliable. Predictions suggest that at least half of the Hungarian population has lost immunity during the epidemic outbreak caused by the BA.1 and BA.2 subvariants of Omicron in the first half of 2022. We obtained a similar result for the outbreaks caused by the subvariant BA.5 in the second half of 2022.

(APPLICABILITY)

The proposed approach has been used to support COVID management in Hungary and could be customized for other countries as well.

The dynamic relationship between COVID-19 cases and SARS-CoV-2 wastewater concentrations across time and space: Considerations for model training data sets

During the COVID-19 pandemic, wastewater-based surveillance has been used alongside diagnostic testing to monitor infection rates. With the decline in cases reported to public health departments due to at-home testing, wastewater data may serve as the primary input for epidemiological models, but training these models is not straightforward. We explored factors affecting noise and bias in the ratio between wastewater and case data collected in 26 sewersheds in California from October 2020 to March 2022. The strength of the relationship between wastewater and case data appeared dependent on sampling frequency and population size, but was not increased by wastewater normalization to flow rate or case count normalization to testing rates. Additionally, the lead and lag times between wastewater and case data varied over time and space, and the ratio of log-transformed individual cases to wastewater concentrations changed over time. This ratio decreased between the Epsilon/Alpha and Delta variant surges of COVID-19 and increased during the Omicron BA.1 variant surge, and was also related to the diagnostic testing rate. Based on this analysis, we present a framework of scenarios describing the dynamics of the case to wastewater ratio to aid in data handling decisions for ongoing modeling efforts.

Model training periods impact estimation of COVID-19 incidence from wastewater viral loads

Wastewater-based epidemiology (WBE) has been deployed broadly as an early warning tool for emerging COVID-19 outbreaks. WBE can inform targeted interventions and identify communities with high transmission, enabling quick and effective responses. As the wastewater (WW) becomes an increasingly important indicator for COVID-19 transmission, more robust methods and metrics are needed to guide public health decision-making. This research aimed to develop and implement a mathematical framework to infer incident cases of COVID-19 from SARS-CoV-2 levels measured in WW. We propose a classification scheme to assess the adequacy of model training periods based on clinical testing rates and assess the sensitivity of model predictions to training periods. A testing period is classified as adequate when the rate of change in testing is greater than the rate of change in cases. We present a Bayesian deconvolution and linear regression model to estimate COVID-19 cases from WW data. The effective reproductive number is estimated from reconstructed cases using WW. The proposed modeling framework was applied to three Northern California communities served by distinct WW treatment plants. The results showed that training periods with adequate testing are essential to provide accurate projections of COVID-19 incidence.

An exploration of challenges associated with machine learning for time series forecasting of COVID-19 community spread using wastewater-based epidemiological data

Wastewater-based epidemiology (WBE) has gained increasing attention as a complementary tool to conventional surveillance methods with potential for significant resource and labour savings when used for public health monitoring. Using WBE datasets to train machine learning algorithms and develop predictive models may also facilitate early warnings for the spread of outbreaks. The challenges associated with using machine learning for the analysis of WBE datasets and timeseries forecasting of COVID-19 were explored by running Random Forest (RF) algorithms on WBE datasets across 108 sites in five regions: Scotland, Catalonia, Ohio, the Netherlands, and Switzerland. This method uses measurements of SARS-CoV-2 RNA fragment concentration in samples taken at the inlets of wastewater treatment plants, providing insight into the prevalence of infection in upstream wastewater catchment populations. RF's forecasting performance at each site was quantitatively evaluated by determining mean absolute percentage error (MAPE) values, which was used to highlight challenges affecting future implementations of RF for WBE forecasting efforts. Performance was generally poor using WBE datasets from Catalonia, Scotland, and Ohio with 'reasonable' or better forecasts constituting 0 %, 5 %, and 0 % of these regions' forecasts, respectively. RF's performance was much stronger with WBE data from the Netherlands and Switzerland, which provided 55 % and 45 % 'reasonable' or better forecasts respectively. Sampling frequency and training set size were identified as key factors contributing to accuracy, while inclusion of too many unnecessary variables (or e.g., flow data) was identified as a contributing factor to poor performance. The contribution of catchment population on forecast accuracy was more ambiguous. This study determined that the factors governing RF's forecast performance are complicated and interrelated, which presents challenges for further work in this space. A sufficiently accurate further iteration of the tool discussed within this study would provide significant but varying value for public health departments for monitoring future, or ongoing outbreaks, assisting the implementation of on-time health response measures.

Bayesian sequential approach to monitor COVID-19 variants through positivity rate from wastewater

Trends in COVID-19 infection have changed throughout the pandemic due to myriad factors, including changes in transmission driven by social behavior, vaccine development and uptake, mutations in the virus genome, and public health policies. Mass testing was an essential control measure for curtailing the burden of COVID-19 and monitoring the magnitude of the pandemic during its multiple phases. However, as the pandemic progressed, new preventive and surveillance mechanisms emerged. Implementing vaccine programs, wastewater (WW) surveillance, and at-home COVID-19 tests reduced the demand for mass severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) testing. This paper proposes a sequential Bayesian approach to estimate the COVID-19 positivity rate (PR) using SARS-CoV-2 RNA concentrations measured in WW through an adaptive scheme incorporating changes in virus dynamics. PR estimates are used to compute thresholds for WW data using the CDC thresholds for low, substantial, and high transmission. The effective reproductive number estimates are calculated using PR estimates from the WW data. This approach provides insights into the dynamics of the virus evolution and an analytical framework that combines different data sources to continue monitoring the COVID-19 trends. These results can provide public health guidance to reduce the burden of future outbreaks as new variants continue to emerge. The proposed modeling framework was applied to the City of Davis and the campus of the University of California Davis.