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).

Big data- and artificial intelligence-based hot-spot analysis of COVID-19: Gauteng, South Africa, as a case study.

The coronavirus disease 2019 (COVID-19) has developed into a pandemic. Data-driven techniques can be used to inform and guide public health decision- and policy-makers. In generalizing the spread of a virus over a large area, such as a province, it must be assumed that the transmission occurs as a stochastic process. It is therefore very difficult for policy and decision makers to understand and visualize the location specific dynamics of the virus on a more granular level. A primary concern is exposing local virus hot-spots, in order to inform and implement non-pharmaceutical interventions. A hot-spot is defined as an area experiencing exponential growth relative to the generalised growth of the pandemic. This paper uses the first and second waves of the COVID-19 epidemic in Gauteng Province, South Africa, as a case study. The study aims provide a data-driven methodology and comprehensive case study to expose location specific virus dynamics within a given area. The methodology uses an unsupervised Gaussian Mixture model to cluster cases at a desired granularity. This is combined with an epidemiological analysis to quantify each cluster's severity, progression and whether it can be defined as a hot-spot.

Privacy-Preserving Workflow for the Cross-Border Federated Analysis of Clinical Data.

The motivation behind this research is to perform a privacy-preserving analysis of data located at remote sites and in different jurisdictions with no possibility of sharing individual-level information. Here, we present key findings from requirements analysis and a resulting federated data analysis workflow built using open-source research software, where patient-level information is securely stored and never exposed during the analysis process. We present additional improvements to further strengthen the security of the workflow. We emphasize and showcase the use of data harmonization in the analysis. The data analysis is done using the R language for statistical computing and DataSHIELD libraries for non-disclosive analysis of sensitive data. The workflow was validated against two data analysis scenarios, confirming the results obtained with a centralized analysis approach. The clinical datasets are part of the large Pan-European SARS-Cov-2 cohort, collected and managed by the ORCHESTRA project. We demonstrate the viability of establishing a cross-border federated data analysis framework and conducting an analysis without exposing patient-level information, achieving results equivalent to centralized non-secure analysis. However, it is vital to ensure requirements associated with data harmonization, anonymization and IT infrastructure to maintain availability, usability and data security.

Clinical phenotypes and quality of life to define post-COVID-19 syndrome: a cluster analysis of the multinational, prospective ORCHESTRA cohort.

Background: Lack of specific definitions of clinical characteristics, disease severity, and risk and preventive factors of post-COVID-19 syndrome (PCS) severely impacts research and discovery of new preventive and therapeutics drugs. Methods: This prospective multicenter cohort study was conducted from February 2020 to June 2022 in 5 countries, enrolling SARS-CoV-2 out- and in-patients followed at 3-, 6-, and 12-month from diagnosis, with assessment of clinical and biochemical features, antibody (Ab) response, Variant of Concern (VoC), and physical and mental quality of life (QoL). Outcome of interest was identification of risk and protective factors of PCS by clinical phenotype, setting, severity of disease, treatment, and vaccination status. We used SF-36 questionnaire to assess evolution in QoL index during follow-up and unsupervised machine learning algorithms (principal component analysis, PCA) to explore symptom clusters. Severity of PCS was defined by clinical phenotype and QoL. We also used generalized linear models to analyse the impact of PCS on QoL and associated risk and preventive factors. CT registration number: NCT05097677. Findings: Among 1796 patients enrolled, 1030 (57%) suffered from at least one symptom at 12-month. PCA identified 4 clinical phenotypes: chronic fatigue-like syndrome (CFs: fatigue, headache and memory loss, 757 patients, 42%), respiratory syndrome (REs: cough and dyspnoea, 502, 23%); chronic pain syndrome (CPs: arthralgia and myalgia, 399, 22%); and neurosensorial syndrome (NSs: alteration in taste and smell, 197, 11%). Determinants of clinical phenotypes were different (all comparisons p < 0.05): being female increased risk of CPs, NSs, and CFs; chronic pulmonary diseases of REs; neurological symptoms at SARS-CoV-2 diagnosis of REs, NSs, and CFs; oxygen therapy of CFs and REs; and gastrointestinal symptoms at SARS-CoV-2 diagnosis of CFs. Early treatment of SARS-CoV-2 infection with monoclonal Ab (all clinical phenotypes), corticosteroids therapy for mild/severe cases (NSs), and SARS-CoV-2 vaccination (CPs) were less likely to be associated to PCS (all comparisons p < 0.05). Highest reduction in QoL was detected in REs and CPs (43.57 and 43.86 vs 57.32 in PCS-negative controls, p < 0.001). Female sex (p < 0.001), gastrointestinal symptoms (p = 0.034) and renal complications (p = 0.002) during the acute infection were likely to increase risk of severe PCS (QoL <50). Vaccination and early treatment with monoclonal Ab reduced the risk of severe PCS (p = 0.01 and p = 0.03, respectively). Interpretation: Our study provides new evidence suggesting that PCS can be classified by clinical phenotypes with different impact on QoL, underlying possible different pathogenic mechanisms. We identified factors associated to each clinical phenotype and to severe PCS. These results might help in designing pathogenesis studies and in selecting high-risk patients for inclusion in therapeutic and management clinical trials. Funding: The study received funding from the Horizon 2020 ORCHESTRA project, grant 101016167; from the Netherlands Organisation for Health Research and Development (ZonMw), grant 10430012010023; from Inserm, REACTing (REsearch & ACtion emergING infectious diseases) consortium and the French Ministry of Health, grant PHRC 20-0424.

Clinical Impact of Monoclonal Antibodies in the Treatment of High-Risk Patients with SARS-CoV-2 Breakthrough Infections: The ORCHESTRA Prospective Cohort Study.

The clinical impact of anti-spike monoclonal antibodies (mAb) in Coronavirus Disease 2019 (COVID-19) breakthrough infections is unclear. We present the results of an observational prospective cohort study assessing and comparing COVID-19 progression in high-risk outpatients receiving mAb according to primary or breakthrough infection. Clinical, serological and virological predictors associated with 28-day COVID-19-related hospitalization were identified using multivariate logistic regression and summarized with odds ratio (aOR) and 95% confidence interval (CI). A total of 847 COVID-19 outpatients were included: 414 with primary and 433 with breakthrough infection. Hospitalization was observed in 42/414 (10.1%) patients with primary and 8/433 (1.8%) patients with breakthrough infection (p < 0.001). aOR for hospitalization was significantly lower for breakthrough infection (aOR 0.12, 95%CI: 0.05-0.27, p < 0.001) and higher for immunocompromised status (aOR:2.35, 95%CI:1.08-5.08, p = 0.003), advanced age (aOR:1.06, 95%CI: 1.03-1.08, p < 0.001), and male gender (aOR:1.97, 95%CI: 1.04-3.73, p = 0.037). Among the breakthrough infection group, the median SARS-CoV-2 anti-spike IgGs was lower (p < 0.001) in immunocompromised and elderly patients >75 years compared with that in the immunocompetent patients. Our findings suggest that, among mAb patients, those with breakthrough infection have significantly lower hospitalization risk compared with patients with primary infection. Prognostic algorithms combining clinical and immune-virological characteristics are needed to ensure appropriate and up-to-date clinical protocols targeting high-risk categories.