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.

Background: The SARS-CoV-2 pandemic is currently leading to increasing numbers of COVID-19 patients all over the world. Clinical presentations range from asymptomatic, mild respiratory tract infection, to severe cases with acute respiratory distress syndrome, respiratory failure, and death. Reports on a dysregulated immune system in the severe cases call for a better characterization and understanding of the changes in the immune system.Methods: In order to dissect COVID-19-driven immune host responses, we performed RNA-seq of whole blood cell transcriptomes and granulocyte preparations from mild and severe COVID-19 patients and analyzed the data using a combination of conventional and data-driven co-expression analysis. Additionally, publicly available data was used to show the distinction from COVID-19 to other diseases. Reverse drug target prediction was used to identify known or novel drug candidates based on finding from data-driven findings.Results: Here, we profiled whole blood transcriptomes of 39 COVID-19 patients and 10 control donors enabling a data-driven stratification based on molecular phenotype. Neutrophil activation-associated signatures were prominently enriched in severe patient groups, which was corroborated in whole blood transcriptomes from an independent second cohort of 30 as well as in granulocyte samples from a third cohort of 16 COVID-19 patients (44 samples). Comparison of COVID-19 blood transcriptomes with those of a collection of over 3100 samples derived from 12 different viral infections, inflammatory diseases, and independent control samples revealed highly specific transcriptome signatures for COVID-19. Further, stratified transcriptomes predicted patient subgroup-specific drug candidates targeting the dysregulated systemic immune response of the host.Conclusions: Our study provides novel insights in the distinct molecular subgroups or phenotypes that are not simply explained by clinical parameters. We show that whole blood transcriptomes are extremely informative for COVID-19 since they capture granulocytes which are major drivers of disease severity.

After birth, the immune system matures via interactions with microbes in the gut. The S100 calcium binding proteins S100A8 and S100A9, and their extracellular complex form, S100A8-A9, are found in high amounts in human breast milk. We studied levels of S100A8-A9 in fecal samples (also called fecal calprotectin) from newborns and during infancy, and their effects on development of the intestinal microbiota and mucosal immune system. We collected stool samples (n = 517) from full-term (n = 72) and preterm infants (n = 49) at different timepoints over the first year of life (days 1, 3, 10, 30, 90, 180, and 360). We measured levels of S100A8-A9 by enzyme-linked immunosorbent assay and analyzed fecal microbiomes by 16S sRNA gene sequencing. We also obtained small and large intestine biopsies from 8 adults and 10 newborn infants without inflammatory bowel diseases (controls) and 8 infants with necrotizing enterocolitis and measured levels of S100A8 by immunofluorescence microscopy. Children were followed for 2.5 years and anthropometric data and medical information on infections were collected. We performed studies with newborn C57BL/6J wild-type and S100a9^-/- mice (which also lack S100A8). Some mice were fed or given intraperitoneal injections of S100A8 or subcutaneous injections of Staphylococcus aureus. Blood and intestine, mesenterial and celiac lymph nodes were collected; cells and cytokines were measured by flow cytometry and studied in cell culture assays. Colon contents from mice were analyzed by culture-based microbiology assays. Loss of S100A8 and S100A9 in mice altered the phenotypes of colonic lamina propria macrophages, compared with wild-type mice. Intestinal tissues from neonatal S100-knockout mice had reduced levels of CX3CR1 protein, and Il10 and Tgfb1 mRNAs, compared with wild-type mice, and fewer T-regulatory cells. S100-knockout mice weighed 21% more than wild-type mice at age 8 weeks and a higher proportion developed fatal sepsis during the neonatal period. S100-knockout mice had alterations in their fecal microbiomes, with higher abundance of Enterobacteriaceae. Feeding mice S100 at birth prevented the expansion of Enterobacteriaceae, increased numbers of T-regulatory cells and levels of CX3CR1 protein and Il10 mRNA in intestine tissues, and reduced body weight and death from neonatal sepsis. Fecal samples from term infants, but not preterm infants, had significantly higher levels of S100A8-A9 during the first 3 months of life than fecal samples from adults; levels decreased to adult levels after weaning. Fecal samples from infants born by cesarean delivery had lower levels of S100A8-A9 than from infants born by vaginal delivery. S100 proteins were expressed by lamina propria macrophages in intestinal tissues from infants, at higher levels than in intestinal tissues from adults. High fecal levels of S100 proteins, from 30 days to 1 year of age, were associated with higher abundance of Actinobacteria and Bifidobacteriaceae, and lower abundance of Gammaproteobacteria-particularly opportunistic Enterobacteriaceae. A low level of S100 proteins in infants' fecal samples associated with development of sepsis and obesity by age 2 years. S100A8 and S100A9 regulate development of the intestinal microbiota and immune system in neonates. Nutritional supplementation with these proteins might aide in development of preterm infants and prevent microbiota-associated disorders in later years.

Macrophages are the principal immune cells of the epididymis and testis, but their origins, heterogeneity, development, and maintenance are not well understood. Here, we describe distinct populations of epididymal and testicular macrophages that display an organ-specific cellular identity. Combining in vivo fate-mapping, chimeric and parabiotic mouse models with in-depth cellular analyses, we found that CD64^hiMHCII^lo and CD64^loMHCII^hi macrophage populations of epididymis and testis arise sequentially from yolk sac erythro-myeloid progenitors, embryonic hematopoiesis, and nascent neonatal monocytes. While monocytes were the major developmental source of both epididymal and testicular macrophages, both populations self-maintain in the steady-state independent of bone marrow hematopoietic precursors. However, after radiation-induced macrophage ablation or during infection, bone marrow-derived circulating monocytes are recruited to the epididymis and testis, giving rise to inflammatory macrophages that promote tissue damage. These results define the layered ontogeny, maintenance and inflammatory response of macrophage populations in the male reproductive organs. Significance Infection/inflammation is a major contributor to male infertility, and macrophages are likely to be key players in both pathological progression and resolution of the inflammation. We report that macrophage populations in the epididymis and testis are derived from fetal and neonatal monocytes, which are self-maintaining during adulthood. Furthermore, during inflammation, circulating monocytes recruited to the epididymis and testis give rise to inflammatory macrophages that promote tissue damage. These data are significant for our understanding of the origins and maintenance of macrophage subpopulations in both organs; our results point to a fundamental mechanism underpinning male infertility by infection and inflammation, and pave the way for the development of innovative therapeutics to treat this important and common condition.