Flusion: Integrating multiple data sources for accurate influenza predictions

Over the last ten years, the US Centers for Disease Control and Prevention (CDC) has organized an annual influenza forecasting challenge with the motivation that accurate probabilistic forecasts could improve situational awareness and yield more effective public health actions. Starting with the 2021/22 influenza season, the forecasting targets for this challenge have been based on hospital admissions reported in the CDC's National Healthcare Safety Network (NHSN) surveillance system. Reporting of influenza hospital admissions through NHSN began within the last few years, and as such only a limited amount of historical data are available for this target signal. To produce forecasts in the presence of limited data for the target surveillance system, we augmented these data with two signals that have a longer historical record: 1) ILI+, which estimates the proportion of outpatient doctor visits where the patient has influenza; and 2) rates of laboratory-confirmed influenza hospitalizations at a selected set of healthcare facilities. Our model, Flusion, is an ensemble model that combines two machine learning models using gradient boosting for quantile regression based on different feature sets with a Bayesian autoregressive model. The gradient boosting models were trained on all three data signals, while the autoregressive model was trained on only data for the target surveillance signal, NHSN admissions; all three models were trained jointly on data for multiple locations. In each week of the influenza season, these models produced quantiles of a predictive distribution of influenza hospital admissions in each state for the current week and the following three weeks; the ensemble prediction was computed by averaging these quantile predictions. Flusion emerged as the top-performing model in the CDC's influenza prediction challenge for the 2023/24 season. In this article we investigate the factors contributing to Flusion's success, and we find that its strong performance was primarily driven by the use of a gradient boosting model that was trained jointly on data from multiple surveillance signals and multiple locations. These results indicate the value of sharing information across multiple locations and surveillance signals, especially when doing so adds to the pool of available training data.

Artificial Neural Network Prediction of COVID-19 Daily Infection Count

This study addresses COVID-19 testing as a nonlinear sampling problem, aiming to uncover the dependence of the true infection count in the population on COVID-19 testing metrics such as testing volume and positivity rates. Employing an artificial neural network, we explore the relationship among daily confirmed case counts, testing data, population statistics, and the actual daily case count. The trained artificial neural network undergoes testing in in-sample, out-of-sample, and several hypothetical scenarios. A substantial focus of this paper lies in the estimation of the daily true case count, which serves as the output set of our training process. To achieve this, we implement a regularized backcasting technique that utilize death counts and the infection fatality ratio (IFR), as the death statistics and serological surveys (providing the IFR) as more reliable COVID-19 data sources. Addressing the impact of factors such as age distribution, vaccination, and emerging variants on the IFR time series is a pivotal aspect of our analysis. We expect our study to enhance our understanding of the genuine implications of the COVID-19 pandemic, subsequently benefiting mitigation strategies.

Reconstructing COVID-19 incidences from positive RT-PCR tests by deconvolution

BACKGROUND

The emergency of new COVID-19 variants over the past three years posed a serious challenge to the public health. Cities in China implemented mass daily RT-PCR tests by pooling strategies. However, a random delay exists between an infection and its first positive RT-PCR test. It is valuable for disease control to know the delay pattern and daily infection incidences reconstructed from RT-PCR test observations.

METHODS

We formulated the convolution model between daily incidences and positive RT-PCR test counts as a linear inverse problem with positivity restrictions. Consequently, the Richard-Lucy deconvolution algorithm was used to reconstruct COVID-19 incidences from daily PCR tests. A real-time deconvolution was further developed based on the same mathematical principle. The method was applied to an Omicron epidemic data set of a bar outbreak in Beijing and another in Wuxi in June 2022. We estimated the delay function by maximizing likelihood via an E-M algorithm.

RESULTS

The delay function of the bar-outbreak in 2022 differs from that reported in 2020. Its mode was shortened to 4 days by one day. A 95% confidence interval of the mean delay is [4.43,5.55] as evaluated by bootstrap. In addition, the deconvolved infection incidences successfully detected two associated infection events after the bar was closed. The application of the real-time deconvolution to the Wuxi data identified all explosive incidence increases. The results revealed the progression of the two COVID-19 outbreaks and provided new insights for prevention and control strategies, especially for the role of mass daily RT-PCR testing.

CONCLUSIONS

The proposed deconvolution method is generally applicable to other infectious diseases if the delay model can be assumed to be approximately valid. To ensure a fair reconstruction of daily infection incidences, the delay function should be estimated in a similar context in terms of virus variant and test protocol. Both the delay estimate from the E-M algorithm and the incidences resulted from deconvolution are valuable for epidemic prevention and control. The real-time feedback is particularly useful during the epidemic's acute phase because it can help the local disease control authorities modify the control measures more promptly and precisely.

Dynamics-informed deconvolutional neural networks for super-resolution identification of regime changes in epidemiological time series

The ability to infer the timing and amplitude of perturbations in epidemiological systems from their stochastically spread low-resolution outcomes is crucial for multiple applications. However, the general problem of connecting epidemiological curves with the underlying incidence lacks the highly effective methodology present in other inverse problems, such as super-resolution and dehazing from computer vision. Here, we develop an unsupervised physics-informed convolutional neural network approach in reverse to connect death records with incidence that allows the identification of regime changes at single-day resolution. Applied to COVID-19 data with proper regularization and model-selection criteria, the approach can identify the implementation and removal of lockdowns and other nonpharmaceutical interventions (NPIs) with 0.93-day accuracy over the time span of a year.