Sasikiran Kandula

In 2023 the European Centre for Disease Prevention and Control (ECDC) launched RespiCast, the first European Respiratory Diseases Forecasting Hub, to provide probabilistic forecasts for influenza-like illness (ILI) and acute respiratory infection (ARI) incidence across 26 European countries. During the 2023/24 and 2024/25 winter seasons, RespiCast collected one- to four-week-ahead forecasts from multiple models contributed by different international teams and combined them into an ensemble. Our analysis shows that, when evaluated using the weighted interval score (WIS) and the absolute error (AE), the ensemble consistently outperformed the baseline model (defined as a persistence model that projects the last observed value forward) as well as individual models across most countries and forecasting rounds for both ILI and ARI incidence in the two seasons. Analysis of ensemble coverage (defined as the proportion of times observed values fall within the specified prediction intervals) indicated that forecast prediction intervals were reliable, although a general overconfidence trend (i.e., prediction intervals that are too narrow) was observed, particularly in specific countries. The relative performance of the ensemble declined in certain weeks, likely due to reduced participation from modelling teams, epidemic dynamics, higher data noise, and reporting delays. Forecast scores varied across countries, with some exhibiting consistently higher relative errors than others. Overall, the findings highlight the strengths of ensemble approaches in improving the accuracy and reliability of epidemiological forecasts while identifying areas for improvement, such as managing overconfidence and addressing variability in performance across countries and over time.

BACKGROUND: Monitoring of mortality to identify trends and detect deviations from normal levels is an essential part of routine surveillance. In many European countries, disruptions in mortality patterns from the COVID-19 pandemic have required revisions to expected mortality estimates (and models) in the current endemic phase of SARS-CoV-2. AIM: To identify essential characteristics for future mortality surveillance and describe two Bayesian methods that satisfy these criteria while being robust to past periods of high COVID-19 mortality. We demonstrate their application in 19 European countries and subnational estimates in the United States, and report measures of model calibration. METHODS: We used a generalised additive model (GAM) with smoothed spline terms for annual trend and within-year seasonality and a generalised linear model (GLM) with a Serfling component for within-year seasonality and breakpoints to detect trend changes in trend. Both approaches modelled change in population size and group-specific (age and sex) mortality patterns. RESULTS: Models were well-calibrated and able to estimate national and group-specific mortality before and during the acute COVID-19 pandemic phase. The effect of inclusion of mortality from the acute pandemic period was primarily an increase in uncertainty in expected mortality over the projection period. The GAM approach had better calibration and less variability in bias among countries. CONCLUSION: Models that can adapt to mortality anomalies seen during the acute COVID-19 pandemic period without a need for adjustments to observational data, or tailoring of model specifications, are feasible. The proposed methods can complement operational national and inter-agency surveillance systems currently used in Europe.

In this study, we assessed the overall impact of the COVID-19 pandemic in the United States between 2020 and 2023 through estimates of excess all-cause mortality. Monthly mortality rates over a 19-year period, stratified by age, sex, and state of residence, were used to predict expected mortality for the pandemic years. A combination of models—2 timeseries, a spatial random-effects and a generalized additive—was used to better capture uncertainty. Results indicate that the national excess mortality for the United States decreased in 2023 to 157 000 (95% prediction interval: 35 000-282 000) from 502 000 (436 000-567 000), 574 000 (484 000-666 000) and 377 000 (264 000-484 000) during the years 2020-2022, respectively. Unlike in previous years, deaths with COVID-19 as the underlying cause of death possibly accounted for all excess deaths during 2023. While for the older age groups (>75 years), the year 2020—before vaccines were available—had the highest excess mortality rate, and the 2 younger age groups had the highest excess mortality in 2021. In each age group, women were estimated to have consistently lower excess mortality than men. West Virginia had the highest age-standardized excess mortality among all states in 2021 and 2022. Our findings demonstrate the value of a multimodel approach in capturing heterogeneity in excess mortality.

The spread of suicidal behavior among individuals is often described as a contagion; however, rigorous modeling of suicide as a dynamic, contagious process is minimal. Here, we develop and validate a model-inference system depicting suicide ideation and death and use it to quantify the contagion processes in the US associated with two prominent celebrity suicide events: Robin Williams during 2014 and Kate Spade and Anthony Bourdain, which occurred 3 days apart during 2018. We show that both events produced large transient increases of suicide contagion contact rates, i.e., the spread of suicidal thought and behavior, and a period of elevated suicidal ideation in the general population. Our modeling approach provides a framework for quantifying suicidal contagion and better understanding, preventing, and containing its spread.

Accurate forecasts can enable more effective public health responses during seasonal influenza epidemics. Forecasting teams were asked to provide national and jurisdiction-specific probabilistic predictions of weekly confirmed influenza hospital admissions for one through four weeks ahead for the 2021-22 and 2022-23 influenza seasons. Across both seasons, 26 teams submitted forecasts, with the submitting teams varying between seasons. Forecast skill was evaluated using the Weighted Interval Score (WIS), relative WIS, and coverage. Six out of 23 models outperformed the baseline model across forecast weeks and locations in 2021-22 and 12 out of 18 models in 2022-23. Averaging across all forecast targets, the FluSight ensemble was the 2 ^nd most accurate model measured by WIS in 2021-22 and the 5 ^th most accurate in the 2022-23 season. Forecast skill and 95% coverage for the FluSight ensemble and most component models degraded over longer forecast horizons and during periods of rapid change. Current influenza forecasting efforts help inform situational awareness, but research is needed to address limitations, including decreased performance during periods of changing epidemic dynamics.

Suicide rates in the United States have increased over the past 15 years, with substantial geographic variation in these increases; yet there have been few attempts to cluster counties by the magnitude of suicide rate changes according to intercept and slope or to identify the economic precursors of increases. We used vital statistics data and growth mixture models to identify clusters of counties by their magnitude of suicide growth from 2008 to 2020 and examined associations with county economic and labor indices. Our models identified 5 clusters, each differentiated by intercept and slope magnitude, with the highest-rate cluster (4% of counties) being observed mainly in sparsely populated areas in the West and Alaska, starting the time series at 25.4 suicides per 100,000 population, and exhibiting the steepest increase in slope (0.69/100,000/year). There was no cluster for which the suicide rate was stable or declining. Counties in the highest-rate cluster were more likely to have agricultural and service economies and less likely to have urban professional economies. Given the increased burden of suicide, with no clusters of counties improving over time, additional policy and prevention efforts are needed, particularly targeted at rural areas in the West.

Deaths by suicide, as well as suicidal ideations, plans and attempts, have been increasing in the US for the past two decades. Deployment of effective interventions would require timely, geographically well-resolved estimates of suicide activity. In this study, we evaluated the feasibility of a two-step process for predicting suicide mortality: a) generation of hindcasts, mortality estimates for past months for which observational data would not have been available if forecasts were generated in real-time; and b) generation of forecasts with observational data augmented with hindcasts. Calls to crisis hotline services and online queries to the Google search engine for suicide-related terms were used as proxy data sources to generate hindcasts. The primary hindcast model (auto) is an Autoregressive Integrated Moving average model (ARIMA), trained on suicide mortality rates alone. Three regression models augment hindcast estimates from auto with call rates (calls), GHT search rates (ght) and both datasets together (calls_ght). The 4 forecast models used are ARIMA models trained with corresponding hindcast estimates. All models were evaluated against a baseline random walk with drift model. Rolling monthly 6-month ahead forecasts for all 50 states between 2012 and 2020 were generated. Quantile score (QS) was used to assess the quality of the forecast distributions. Median QS for auto was better than baseline (0.114 vs. 0.21. Median QS of augmented models were lower than auto, but not significantly different from each other (Wilcoxon signed-rank test, p > .05). Forecasts from augmented models were also better calibrated. Together, these results provide evidence that proxy data can address delays in release of suicide mortality data and improve forecast quality. An operational forecast system of state-level suicide risk may be feasible with sustained engagement between modelers and public health departments to appraise data sources and methods as well as to continuously evaluate forecast accuracy.

Background: Suicide is one of the leading causes of death in the United States and population risk prediction models can inform the type, location, and timing of public health interventions. Here, we report the development of a prediction model of suicide risk using population characteristics. Methods: All suicide deaths reported to the Nation Vital Statistics System between 2005-2019 were identified, and age, sex, race, and county-of-residence of the decedents were extracted to calculate baseline risk. County-wise annual measures of socioeconomic predictors of suicide risk — unemployment, weekly wage, poverty prevalence, median household income, and population density — along with two state-wise measures of prevalence of major depressive disorder and firearm ownership were compiled from public sources. Conditional autoregressive (CAR) models, which account for spatiotemporal autocorrelation in response and predictors, were used to estimate county-level risk. Results: Estimates derived from CAR models were more accurate than from models not adjusted for spatiotemporal autocorrelation. Inclusion of suicide risk/protective covariates further reduced errors. Suicide risk was estimated to increase with each standard deviation increase in firearm ownership (2.8%), prevalence of major depressive episode (1%) and unemployment (2.8%). Conversely, risk was estimated to decrease by 4.3% for each standard deviation increase in both median household income and population density. Increased heterogeneity of risk across counties was also noted. Conclusions: Area-level characteristics and the CAR model structure can estimate population-level suicide risk and thus inform decisions on resource allocation and focused interventions during outbreaks.