Simultaneous Identification of Changepoints and Model Parameters in Switching Dynamical Systems (preprint)

Many complex natural systems undergo shifts in dynamics at particular points in time. Examples include phase transitions in gene expression during the cell cycle, introduced species affecting predator-prey interactions, and disease outbreaks responding to intervention measures. Such changepoints partition timeseries into different dynamical regimes characterized by distinct parameter sets, and inference on both the changepoints and regime-specific dynamical parameters is of primary interest. Conventional approaches to analyzing switching dynamical systems first estimate changepoints, and then estimate dynamical parameters assuming the changepoints are fixed and known. Such two-stage approaches are ad-hoc, can introduce biases in the analysis, and do not fully account for uncertainty. Here, we introduce a rigorous, simulation-based inference framework that simultaneously estimates changepoints and model parameters from noisy data while admitting full uncertainty. We use simulation studies of oscillatory predator-prey dynamics and stochastic gene expression to demonstrate that our method yields accurate estimates of changepoints and model parameters together with appropriate uncertainty bounds. We then apply our approach to a real-world case study of COVID-19 intervention effects, and show that our inferred changepoints aligned closely with the actual dates of intervention implementation. Taken together, these results suggest that our framework will have broad utility in diverse scientific domains.

Autonomous and policy-induced behavior change during the COVID-19 pandemic: Towards understanding and modeling the interplay of behavioral adaptation (preprint)

Changes in human behaviors, such as reductions of physical contacts and the adoption of preventive measures, impact the transmission of infectious diseases considerably. Behavioral adaptations may be the result of individuals aiming to protect themselves or mere responses to public containment measures, or a combination of both. What drives autonomous and policy-induced adaptation, how they are related and change over time is insufficiently understood. Here, we develop a framework for more precise analysis of behavioral adaptation, focusing on confluence, interactions and time variance of autonomous and policy-induced adaptation. We carry out an empirical analysis of Germany during the fall of 2020 and beyond. Subsequently, we discuss how behavioral adaptation processes can be better represented in behavioral-epidemiological models. We find that our framework is useful to understand the interplay of autonomous and policy-induced adaptation as a “moving target”. Our empirical analysis suggests that mobility patterns in Germany changed significantly due to both autonomous and policy-induced adaption, with potentially weaker effects over time due to decreasing risk signals, diminishing risk perceptions and an erosion of trust in the government. We find that while a number of simulation and prediction models have made great efforts to represent behavioral adaptation, the interplay of autonomous and policy-induced adaption needs to be better understood to construct convincing counterfactual scenarios for policy analysis. The insights presented here are of interest to modelers and policy makers aiming to understand and account for behaviors during a pandemic response more accurately.

Household transmission of SARS-CoV-2 during the Omicron wave in Shanghai, China:a case-ascertained study (preprint)

Background Since late 2021, the highly transmissible SARS-CoV-2 Omicron variant has driven a new surge of infections across the world. We used a case-ascertained study to determine the features of household transmission of SARS-CoV-2 Omicron variant in Shanghai, China. Methods We collected detailed information on 323 pediatric cases and their 951 household members in April 2022 during the Omicron outbreak. All household members received consecutively intensive RT-PCR testing for SARS-CoV-2 and routine symptom monitoring within 14 days after exposure to a confirmed case. We described the characteristics of study participants and estimated the transmission parameters. Both secondary infection attack rates (SAR I ) and secondary clinical attack rates (SAR C ) among adult household contacts were computed, through which the transmission heterogeneities in infectivity and susceptibility were characterized and the vaccine effectiveness were estimated. Results We estimated the mean incubation period of SARS-CoV-2 Omicron variant to be 4.6 (median: 4.4, IQR: 3.1-6.0) days and the mean serial interval to be 3.9 (median:4.0, IQR: 1.4-6.5) days. The overall SAR I and SAR C among adult household contacts were 77.11% (95% confidence interval [CI]: 73.58%-80.63%) and 67.03% (63.09%-70.98%). We found higher household susceptibility in females, while infectivity was not significantly different in primary cases by age, sex, vaccination status and clinical severity. The estimated VEs of full vaccination was 14.8% (95% CI: 5.8%-22.9%) against Omicron infection and 21.5% (95% CI: 10.4%-31.2%) against symptomatic disease. The booster vaccination was 18.9% (95% CI: 9.0%-27.7%) and 24.3% (95% CI: 12.3%-34.7%) effective against infection and symptomatic disease, respectively. Conclusions We found high household transmission during the Omicron wave in Shanghai due to asymptomatic and pre-symptomatic transmission in the context of city-wide lockdown, indicating the importance of early detection and timely isolation of SARS-CoV-2 infections and quarantine of close contacts. Marginal effectiveness of inactivated vaccines against Omicron infection poses great challenge for prevention and control of the SARS-CoV-2 Omicron variant.