A semi-parametric, state-space compartmental model with time-dependent parameters for forecasting COVID-19 cases, hospitalizations, and deaths

Short-term forecasts of the dynamics of COVID-19 in the period up to its decline following mass vaccination was a task that received much attention but proved difficult to do with high accuracy. However, the availability of standardized forecasts and versioned data sets from this period allows for continued work in this area. Here we introduce the Gaussian Infection State Space with Time-dependence (GISST) forecasting model. We evaluate its performance in 1-4 week ahead forecasts of COVID-19 cases, hospital admissions, and deaths in the state of California made with official reports of COVID-19, Googles mobility reports, and vaccination data available each week. Evaluation of these forecasts with a weighted interval score shows them to consistently outperform a naive baseline forecast and often score closer to or better than a high-performing ensemble forecaster. The GISST model also provides parameter estimates for a compartmental model of COVID-19 dynamics, includes a regression submodel for the transmission rate, and allows for parameters to vary over time according to a random walk. GISST provides a novel, balanced combination of computational efficiency, model interpretability, and applicability to large multivariate data sets that may prove useful in improving the accuracy of infectious disease forecasts.

Overlapping Time Scales Obscure Early Warning Signals for the Second COVID-19 Wave

Early warning indicators based on critical slowing down have been suggested as a model-independent and low-cost tool to anticipate the (re)emergence of infectious diseases. We studied whether such indicators could reliably have anticipated the second COVID-19 wave in European countries. Contrary to theoretical predictions, we found that characteristic early warning indicators generally decreased rather than increased prior to the second wave. A model explains this unexpected finding as a result of transient dynamics and the multiple time scales of relaxation during a non-stationary epidemic. Particularly, if an epidemic that seems initially contained after a first wave does not fully settle to its new quasi-equilibrium prior to changing circumstances or conditions that force a second wave, then indicators will show a decreasing rather than an increasing trend as a result of the persistent transient trajectory of the first wave. Our simulations show that this lack of time scale separation was to be expected during the second European epidemic wave of COVID-19. Overall, our results emphasize that the theory of critical slowing down applies only when the external forcing of the system across a critical point is slow relative to the internal system dynamics.

Evaluation of individual and ensemble probabilistic forecasts of COVID-19 mortality in the US

Short-term probabilistic forecasts of the trajectory of the COVID-19 pandemic in the United States have served as a visible and important communication channel between the scientific modeling community and both the general public and decision-makers. Forecasting models provide specific, quantitative, and evaluable predictions that inform short-term decisions such as healthcare staffing needs, school closures, and allocation of medical supplies. Starting in April 2020, the US COVID-19 Forecast Hub (https://covid19forecasthub.org/) collected, disseminated, and synthesized tens of millions of specific predictions from more than 90 different academic, industry, and independent research groups. A multi-model ensemble forecast that combined predictions from dozens of different research groups every week provided the most consistently accurate probabilistic forecasts of incident deaths due to COVID-19 at the state and national level from April 2020 through October 2021. The performance of 27 individual models that submitted complete forecasts of COVID-19 deaths consistently throughout this year showed high variability in forecast skill across time, geospatial units, and forecast horizons. Two-thirds of the models evaluated showed better accuracy than a naive baseline model. Forecast accuracy degraded as models made predictions further into the future, with probabilistic error at a 20-week horizon 3-5 times larger than when predicting at a 1-week horizon. This project underscores the role that collaboration and active coordination between governmental public health agencies, academic modeling teams, and industry partners can play in developing modern modeling capabilities to support local, state, and federal response to outbreaks. Significance StatementThis paper compares the probabilistic accuracy of short-term forecasts of reported deaths due to COVID-19 during the first year and a half of the pandemic in the US. Results show high variation in accuracy between and within stand-alone models, and more consistent accuracy from an ensemble model that combined forecasts from all eligible models. This demonstrates that an ensemble model provided a reliable and comparatively accurate means of forecasting deaths during the COVID-19 pandemic that exceeded the performance of all of the models that contributed to it. This work strengthens the evidence base for synthesizing multiple models to support public health action.

Reconciling model predictions with low reported cases of COVID-19 in Sub-Saharan Africa: Insights from Madagascar

The COVID-19 pandemic has wreaked havoc globally, and there has been a particular concern for sub-Saharan Africa (SSA), where models suggest that the majority of the population will become infected. Conventional wisdom suggests that the continent will bear a higher burden of COVID-19 for the same reasons it suffers high burdens of other infectious diseases: ecology, socio-economic conditions, lack of water and sanitation infrastructure, and weak health systems. However, so far SSA has reported lower incidence and fatalities compared to the predictions of standard models and the experience of other regions of the world. There are three leading explanations, each with very different implications for the final epidemic burden: (1) low case detection, (2) differences in COVID-19 epidemiology (e.g. low R0), and (3) policy interventions. The low number of cases to date have led some SSA governments to relax these policy interventions. Will this result in a resurgence of cases? To understand how to interpret the lower-than-expected COVID-19 case data in Madagascar, we use a simple age-structured model to explore each of these explanations and predict the epidemic impact associated with them. We show that the current incidence of COVID-19 cases can be explained by any combination of the late introduction of first imported cases, early implementation of non-pharmaceutical interventions (NPIs), and low case detection rates. This analysis reinforces that Madagascar, along with other countries in SSA, remains at risk of an impending health crisis. If NPIs remain enforced, up to 50,000 lives may be saved. Even with NPIs, without vaccines and new therapies, COVID-19 could infect up to 30% of the population, making it the largest public health threat in Madagascar until early 2021, hence the importance of conducting clinical trials and continually improving access to healthcare. ResumeLa pandemie de COVID-19 a eu des consequences nefastes partout dans le monde, et il y a une preoccupation particuliere pour lAfrique subsaharienne (ASS), ou des modeles suggerent que la majorite de la population sera infectee. Il est craint que le continent supportera un fardeau plus elevee de COVID-19 pour les memes raisons quil souffre davantage dautres maladies infectieuses: ecologie, conditions socio-economiques, manque dinfrastructures deau et dassainissement, et faiblesse des systemes de sante. Cependant, jusqua present, lASS a rapporte une incidence et une mortalite bien inferieure a celle des previsions des modeles pour cette region, ainsi quau nombre observe dans dautres regions du monde. Il y a trois explications principales pour ce phenomene, chacune ayant des implications tres differentes pour le fardeau epidemique final: (1) detection faible des cas, (2) differences dans lepidemiologie COVID-19 (par exemple faible R0), et (3) interventions et politiques mises en place. Le faible nombre de cas a ce jour a conduit certains gouvernements dASS a assouplir ces interventions. Cela entrainera-t-il une resurgence de cas? Pour comprendre comment interpreter le fait que les cas COVID-19 rapportes sont plus faibles que prevu a Madagascar, nous utilisons un modele de transmission structure par groupe dage pour explorer chacune de ces explications et predire limpact epidemique qui leur est associe. Nous montrons que lincidence actuelle des cas de COVID-19 peut sexpliquer par leffet cumule de lintroduction tardive des premiers cas importes, la mise en uvre rapide dinterventions non pharmaceutiques (INP) et de faibles taux de detection des cas. Cette analyse renforce le fait que Madagascar, ainsi que dautres pays dAfrique subsaharienne, reste a risque dune crise sanitaire imminente. Si les INP restent appliques, jusqua 50 000 vies pourraient etre sauvees. Meme avec des INP, tant quil ny aura pas des vaccins ni des nouvelles therapies efficaces, COVID-19 pourrait infecter jusqua 30% de la population, ce qui constituerait la plus grande menace pour la sante publique a Madagascar jusquau debut de 2021, dou limportance de la realisation des essais cliniques et de lamelioration continuelle de lacces aux soins. Summary BoxO_LIThe lower-than-expected number of reported cases of COVID-19 in Madagascar can be explained by a combination of the relatively late introduction of the disease, low detection rates, and low transmission rates due to the early and effective implementation of non-pharmaceutical interventions that reduced contact rates. C_LIO_LICOVID-19 is predicted to be the largest public health problem in Madagascar in 2020, but non-pharmaceutical interventions at an effectiveness similar to those seen in the first few months could save up to 50,000 lives. C_LIO_LIHealth systems in SSA remain at risk of an impending health crisis due to COVID-19, stressing the importance of ongoing clinical trials and improving health care access. C_LI

Protective Population Behavior Change in Outbreaks of Emerging Infectious Disease

During outbreaks of emerging infections, the lack of effective drugs and vaccines increases reliance on non-pharmacologic public health interventions and behavior change to limit human-to-human transmission. Interventions that increase the speed with which infected individuals remove themselves from the susceptible population are paramount, particularly isolation and hospitalization. Ebola virus disease (EVD), Severe Acute Respiratory Syndrome (SARS), and Middle East Respiratory Syndrome (MERS) are zoonotic viruses that have caused significant recent outbreaks with sustained human-to-human transmission. This investigation quantified changing mean removal rates (MRR) and days from symptom onset to hospitalization (DSOH) of infected individuals from the population in seven different outbreaks of EVD, SARS, and MERS, to test for statistically significant differences in these metrics between outbreaks. We found that epidemic week and viral serial interval were correlated with the speed with which populations developed and maintained health behaviors in each outbreak.