Acute respiratory distress syndrome (ARDS) is a major contributor to ICU mortality and ischaracterised by hypoxaemia and pulmonary oedema. Pathomechanisms include barrierbreakdown, immunopathology, haemostatic derailment and dysbiosis; however, the actualsequence of events and how they cumulatively lead to lung failure remains unclear.Although ARDS is frequently triggered by pneumonia, it can also occur as a result oftrauma, aspiration or non-pulmonary causes. Importantly, ARDS is highly heterogeneous;growing evidence points to aetiology-specific pathomechanisms - a circumstance thatexplains why attempts to develop specific drugs or timely diagnostic markers have so farfailed.A comprehensive analysis of key microenvironmental and haemostasis-related parameters ofthe lung, combined with multidimensional quantitative image features derived from chestCT scans (radiomics), will enable us to i) identify ARDS phenotypes with differentbiological characteristics and ii) generate new hypotheses regarding aetiology- orsubgroup-specific mechanisms, molecular markers and therapeutic options.Our approach is based on ICU management of our patients guided by bronchoalveolar lavagefluid (BALF). Together with previously sampled cases and new samples collected as part ofthis study, our cohort will consist of patients with i) COVID-19-associated ARDS, ii)ARDS associated with other viral pneumonia, iii) ARDS associated with bacterialpneumonia, and iv) ARDS of non-pulmonary origin. Bacterial and fungal co-infections andsuperinfections are recorded in all patients and taken into account in thestratification. Patients with pneumonia without ARDS, as well as ventilated patientswithout underlying lung disease, serve as controls. To characterise the microbial lungmicroenvironment, the investigators combine data from routine microbiological diagnosticswith microbiome sequencing and metabolomics. In addition, the investigators conductcomprehensive and longitudinal immune and haemostatic profiling by regularly analysingimmune cells, cytokines and parameters of immune thrombosis in BALF and blood.Multi-omics integration then identifies phenotypic subgroups by merging all multimodaldatasets - including radiomics. Selected samples from identified clusters are thenfurther characterised using single-cell sequencing to uncover specific features/markersand pathomechanisms of the respective ARDS subtypes.Although it is clear that the pathogenesis of ARDS is multifactorial, comprehensivestudies that integrate all relevant parameters are rare. Radiomics is increasinglyrecognised as a powerful tool for capturing the clinical status of ARDS in detail;however, to date, this imaging data has not been systematically linked to other omicsreadouts. The investigators aim to bridging this gap by conducting a thoroughinvestigation across various ARDS aetiologies in the present study, incorporating allidentifiable key factors.Our interdisciplinary team comprises basic immunologists, infectious disease andcomputational biologists, as well as clinicians with expertise in ARDS, infectiousdiseases, immunothrombosis and radiology.