Fast and reliable detection of patients with severe and heterogeneous illnesses is a major goal of precision medicine^1,2. Patients with leukaemia can be identified using machine learning on the basis of their blood transcriptomes³. However, there is an increasing divide between what is technically possible and what is allowed, because of privacy legislation^4,5. Here, to facilitate the integration of any medical data from any data owner worldwide without violating privacy laws, we introduce Swarm Learning—a decentralized machine-learning approach that unites edge computing, blockchain-based peer-to-peer networking and coordination while maintaining confidentiality without the need for a central coordinator, thereby going beyond federated learning. To illustrate the feasibility of using Swarm Learning to develop disease classifiers using distributed data, we chose four use cases of heterogeneous diseases (COVID-19, tuberculosis, leukaemia and lung pathologies). With more than 16,400 blood transcriptomes derived from 127 clinical studies with non-uniform distributions of cases and controls and substantial study biases, as well as more than 95,000 chest X-ray images, we show that Swarm Learning classifiers outperform those developed at individual sites. In addition, Swarm Learning completely fulfils local confidentiality regulations by design. We believe that this approach will notably accelerate the introduction of precision medicine.

While the leading symptoms during coronavirus disease 2019 (COVID-19) are acute and the majority of patients fully recover, a significant fraction of patients now increasingly experience long-term health consequences. However, most data available focus on health-related events after severe infection and hospitalisation. We present a longitudinal, prospective analysis of health consequences in patients who initially presented with no or minor symptoms of severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2) infection. Hence, we focus on mild COVID-19 in non-hospitalised patients. 958 Patients with confirmed SARS-CoV-2 infection were observed from April 6th to December 2nd 2020 for long-term symptoms and SARS-CoV-2 antibodies. We identified anosmia, ageusia, fatigue or shortness of breath as most common, persisting symptoms at month 4 and 7 and summarised presence of such long-term health consequences as post-COVID syndrome (PCS). Predictors of long-term symptoms were assessed using an uni- and multivariable logistic regression model. We observed 442 and 353 patients over four and seven months after symptom onset, respectively. Four months post SARS-CoV-2 infection, 8•6% (38/442) of patients presented with shortness of breath, 12•4% (55/442) with anosmia, 11•1% (49/442) with ageusia and 9•7% (43/442) with fatigue. At least one of these characteristic symptoms was present in 27•8% (123/442) and 34•8% (123/353) at month 4 and 7 post-infection, respectively. A lower baseline level of SARS-CoV-2 IgG, anosmia and diarrhoea during acute COVID-19 were associated with higher risk to develop long-term symptoms. The on-going presence of either shortness of breath, anosmia, ageusia or fatigue as long-lasting symptoms even in non-hospitalised patients was observed at four and seven months post-infection and summarised as post-COVID syndrome (PCS). The continued assessment of patients with PCS will become a major task to define and mitigate the socioeconomic and medical long-term effects of COVID-19. COVIM:“NaFoUniMedCovid19”(FKZ: 01KX2021)

Background: The investigation of the antibody response to SARS-CoV-2 represents a key aspect in facing the COVID-19 pandemic. In the present study, we compared the new Immundiagnostik IDK® anti-SARS-CoV-2 S1 IgG assay with four widely-used commercial serological assays for the detection of antibodies targeting S (spike) and NC (nucleocapsid) proteins. Methods: Serum samples were taken from an unbiased group of convalescent patients and from a negative control group. Sample were simultaneously analyzed by the new Immundiagnostik IDK^® anti-SARS-CoV-2 S1 IgG assay, by the DiaSorin LIAISON^® SARS-CoV-2 S1/S2 IgG assay, and by the Euroimmun anti-SARS-CoV-2 S1 IgG ELISA. Antibodies binding NC were detected by the Abbott SARS-CoV-2 IgG assay and by the pan-immunoglobulin immunoassay Roche Elecsys^® anti-SARS-CoV-2. Moreover, we investigated samples of a group of COVID-19 convalescent subjects that were primarily tested S1 IgG non-reactive. Samples were also tested by live virus and pseudovirus neutralization tests. Results: Overall, the IDK® anti-SARS-CoV-2 S1 IgG assay showed the highest sensitivity among the evaluated spike (S) protein-based assays. Additionally, the Immundiagnostik assay correlated well with serum-neutralizing activity. Conclusions: The novel IDK^® anti-SARS-CoV-2 S1 IgG assay showed high sensitivity and specificity, representing a valid option for use in the routine diagnostic.

Background: While the short-term symptoms of post-COVID syndromes (PCS) are well-known, the long-term clinical characteristics, risk factors and outcomes of PCS remain unclear. Moreover, there is ongoing discussion about the effectiveness of post-infection vaccination against severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2) to aid in PCS recovery.Methods: In this longitudinal and observational case-control study we aimed at identifying long-term PCS courses and evaluating the effects of post-infection vaccinations on PCS recovery. Individuals with initial mild COVID-19 were followed for a period of 15 months after primary infection. We assessed PCS outcomes, distinct symptom clusters (SC), and SARS-CoV-2 immunoglobulin G (IgG) levels in patients who received SARS-CoV-2 vaccination, as well as those who did not. To identify potential associating factors with PCS, we used binomial regression models and reported the results as odds ratios (OR) with 95% confidence intervals (95%CI).Results: Out of 958 patients, follow-up data at 15 month after infection was obtained for 222 (23.2%) outpatients. Of those individuals, 36.5% (81/222) and 31.1% (69/222) were identified to have PCS at month 10 and 15, respectively. Fatigue and dyspnea (SC2) rather than anosmia and ageusia (SC1) constituted PCS at month 15. SARS-CoV-2 IgG levels were equally distributed over time among age groups, sex, and absence/presence of PCS. Of the 222 patients, 77.0% (171/222) were vaccinated between 10- and 15-months post-infection, but vaccination did not affect PCS recovery at month 15. 26.3% of unvaccinated and 25.8% of vaccinated outpatients improved from PCS (p= .9646). Baseline headache (SC4) and diarrhoea (SC5) were risk factors for PCS at months 10 and 15 (SC4: OR 1.85 (95%CI 1.04-3.26), p=.0390; SC5: OR 3.27(95%CI 1.54-6.64), p=.0009).Conclusion: Based on the specific symptoms of PCS our findings show a shift in the pattern of recovery. We found no effect of SARS-CoV-2 vaccination on PCS recovery and recommend further studies to identify predicting biomarkers and targeted PCS therapeutics.