This project will combine large-scale electronic patient data, artificial intelligence algorithms, and mechanistic mathematical models, to develop systems that can improve the diagnosis, and hence treatment, of critically ill patients with acute respiratory distress syndrome (ARDS). The key idea is to use mechanistic virtual patient models as "filters" to extract relevant medical information on individual patients, significantly reducing biases introduced by machine learning on heterogeneous datasets, and allowing improved discovery of patient cohorts driven exclusively by medical conditions. I propose to establish a collaboration with Prof Andreas Schuppert at Aachen University and Dr Jörg Lippert at Bayer Healthcare in Germany that will give me access to large-scale patient data and internationally leading expertise in applying machine learning to real clinical problems. As noted recently by leading medical researchers in the journal Intensive Care Medicine, "Artificial Intelligence approaches such as machine learning may assist in identification of patients at risk of or fulfilling diagnostic criteria for ARDS, although this technology is not yet ready for clinical implementation". In ARDS, patient outcomes are poor, while hospital costs are huge - this collaboration will make breakthroughs in the clinical applicability of digital technologies for the earlier identification of ARDS, improving treatment of patients and reducing costs to healthcare providers.