Accessibility of covariance information creates vulnerability in Federated Learning frameworks.

MOTIVATION: Federated Learning (FL) is gaining traction in various fields as it enables integrative data analysis without sharing sensitive data, such as in healthcare. However, the risk of data leakage caused by malicious attacks must be considered. In this study, we introduce a novel attack algorithm that relies on being able to compute sample means, sample covariances, and construct known linearly independent vectors on the data owner side. RESULTS: We show that these basic functionalities, which are available in several established FL frameworks, are sufficient to reconstruct privacy-protected data. Additionally, the attack algorithm is robust to defense strategies that involve adding random noise. We demonstrate the limitations of existing frameworks and propose potential defense strategies analyzing the implications of using differential privacy. The novel insights presented in this study will aid in the improvement of FL frameworks. AVAILABILITY AND IMPLEMENTATION: The code examples are provided at GitHub (https://github.com/manuhuth/Data-Leakage-From-Covariances.git). The CNSIM1 dataset, which we used in the manuscript, is available within the DSData R package (https://github.com/datashield/DSData/tree/main/data).

Integrative modelling of reported case numbers and seroprevalence reveals time-dependent test efficiency and infectious contacts.

Mathematical models have been widely used during the ongoing SARS-CoV-2 pandemic for data interpretation, forecasting, and policy making. However, most models are based on officially reported case numbers, which depend on test availability and test strategies. The time dependence of these factors renders interpretation difficult and might even result in estimation biases. Here, we present a computational modelling framework that allows for the integration of reported case numbers with seroprevalence estimates obtained from representative population cohorts. To account for the time dependence of infection and testing rates, we embed flexible splines in an epidemiological model. The parameters of these splines are estimated, along with the other parameters, from the available data using a Bayesian approach. The application of this approach to the official case numbers reported for Munich (Germany) and the seroprevalence reported by the prospective COVID-19 Cohort Munich (KoCo19) provides first estimates for the time dependence of the under-reporting factor. Furthermore, we estimate how the effectiveness of non-pharmaceutical interventions and of the testing strategy evolves over time. Overall, our results show that the integration of temporally highly resolved and representative data is beneficial for accurate epidemiological analyses.

Seroepidemiology and model-based prediction of SARS-CoV-2 in Ethiopia: longitudinal cohort study among front-line hospital workers and communities.

BACKGROUND: Over 1 year since the first reported case, the true COVID-19 burden in Ethiopia remains unknown due to insufficient surveillance. We aimed to investigate the seroepidemiology of SARS-CoV-2 among front-line hospital workers and communities in Ethiopia. METHODS: We did a population-based, longitudinal cohort study at two tertiary teaching hospitals involving hospital workers, rural residents, and urban communities in Jimma and Addis Ababa. Hospital workers were recruited at both hospitals, and community participants were recruited by convenience sampling including urban metropolitan settings, urban and semi-urban settings, and rural communities. Participants were eligible if they were aged 18 years or older, had provided written informed consent, and were willing to provide blood samples by venepuncture. Only one participant per household was recruited. Serology was done with Elecsys anti-SARS-CoV-2 anti-nucleocapsid assay in three consecutive rounds, with a mean interval of 6 weeks between tests, to obtain seroprevalence and incidence estimates within the cohorts. FINDINGS: Between Aug 5, 2020, and April 10, 2021, we did three survey rounds with a total of 1104 hospital workers and 1229 community residents participating. SARS-CoV-2 seroprevalence among hospital workers increased strongly during the study period: in Addis Ababa, it increased from 10·9% (95% credible interval [CrI] 8·3-13·8) in August, 2020, to 53·7% (44·8-62·5) in February, 2021, with an incidence rate of 2223 per 100 000 person-weeks (95% CI 1785-2696); in Jimma Town, it increased from 30·8% (95% CrI 26·9-34·8) in November, 2020, to 56·1% (51·1-61·1) in February, 2021, with an incidence rate of 3810 per 100 000 person-weeks (95% CI 3149-4540). Among urban communities, an almost 40% increase in seroprevalence was observed in early 2021, with incidence rates of 1622 per 100 000 person-weeks (1004-2429) in Jimma Town and 4646 per 100 000 person-weeks (2797-7255) in Addis Ababa. Seroprevalence in rural communities increased from 18·0% (95% CrI 13·5-23·2) in November, 2020, to 31·0% (22·3-40·3) in March, 2021. INTERPRETATION: SARS-CoV-2 spread in Ethiopia has been highly dynamic among hospital worker and urban communities. We can speculate that the greatest wave of SARS-CoV-2 infections is currently evolving in rural Ethiopia, and thus requires focused attention regarding health-care burden and disease prevention. FUNDING: Bavarian State Ministry of Sciences, Research, and the Arts; Germany Ministry of Education and Research; EU Horizon 2020 programme; Deutsche Forschungsgemeinschaft; and Volkswagenstiftung.

Complex pattern formation driven by the interaction of stable fronts in a competition-diffusion system.

The ecological invasion problem in which a weaker exotic species invades an ecosystem inhabited by two strongly competing native species is modelled by a three-species competition-diffusion system. It is known that for a certain range of parameter values competitor-mediated coexistence occurs and complex spatio-temporal patterns are observed in two spatial dimensions. In this paper we uncover the mechanism which generates such patterns. Under some assumptions on the parameters the three-species competition-diffusion system admits two planarly stable travelling waves. Their interaction in one spatial dimension may result in either reflection or merging into a single homoclinic wave, depending on the strength of the invading species. This transition can be understood by studying the bifurcation structure of the homoclinic wave. In particular, a time-periodic homoclinic wave (breathing wave) is born from a Hopf bifurcation and its unstable branch acts as a separator between the reflection and merging regimes. The same transition occurs in two spatial dimensions: the stable regular spiral associated to the homoclinic wave destabilizes, giving rise first to an oscillating breathing spiral and then breaking up producing a dynamic pattern characterized by many spiral cores. We find that these complex patterns are generated by the interaction of two planarly stable travelling waves, in contrast with many other well known cases of pattern formation where planar instability plays a central role.

Ecological invasion in competition-diffusion systems when the exotic species is either very strong or very weak.

Reaction-diffusion systems with a Lotka-Volterra-type reaction term, also known as competition-diffusion systems, have been used to investigate the dynamics of the competition among m ecological species for a limited resource necessary to their survival and growth. Notwithstanding their rather simple mathematical structure, such systems may display quite interesting behaviours. In particular, while for [Formula: see text] no coexistence of the two species is usually possible, if [Formula: see text] we may observe coexistence of all or a subset of the species, sensitively depending on the parameter values. Such coexistence can take the form of very complex spatio-temporal patterns and oscillations. Unfortunately, at the moment there are no known tools for a complete analytical study of such systems for [Formula: see text]. This means that establishing general criteria for the occurrence of coexistence appears to be very hard. In this paper we will instead give some criteria for the non-coexistence of species, motivated by the ecological problem of the invasion of an ecosystem by an exotic species. We will show that when the environment is very favourable to the invading species the invasion will always be successful and the native species will be driven to extinction. On the other hand, if the environment is not favourable enough, the invasion will always fail.