Project overview The primary healthcare issue addressed by this project is the underutilization of the vast amounts of data contained in electronic health records (EHRs). EHRs hold a treasure trove of patient information, including symptoms, diagnoses, treatment histories, and outcomes. However, the sheer volume and complexity of this data, coupled with concerns about privacy and interoperability, have posed significant challenges to its effective use in improving healthcare outcomes. The difficulty lies in extracting meaningful insights from heterogeneous and unstructured data sources, which are critical for advancing predictive healthcare, personalizing treatments, and making informed clinical decisions. The project aims to develop robust, scalable, and highly accurate predictive models leveraging the vast amounts of data available in electronic health records (EHRs). This project will focus on constructing foundation models that can understand and predict a wide range of health outcomes by integrating and analysing diverse data types, including clinical notes, laboratory results, medication records, and patient demographics. By developing transformer-based models, the project seeks to uncover hidden patterns and relationships within the data, facilitating personalised patient care, improving disease diagnosis and prognosis, and optimising treatment strategies. The main contribution will be a new foundational model, focussed on medical knowledge and electronic health records. It can be used for simulating clinical trajectories, predicting patient outcomes, or discovering novel medical patterns. As the EHRs contain unique types of data in many different formats, the project would also develop methods for integrating these data sources into the language modelling framework (for example, linearising numerical tables or representing structured metadata) The Research environment highly diverse, stimulating and multi-disciplinary. Imperial is consistently ranked in the top 10 of university world rankings; it has the highest proportion of world-leading research. The Department of Computing is the top-ranked Computer Science Department in the country by a substantial margin, and the Faculty of Engineering is the top-ranked engineering school in the 2021 REF. The project is embedded in UKRI AI Centre in Digital Healthcare: https://ai4health.io/training/ The studentship It covers the Home/UK PhD tuition fees, an annual bursary (UKRI rate £21,237 for 2024-25) per year, for at least 3 years, with option to extend to max 4 years, subject to satisfying progress requirements. Students will also have the opportunity of attending leading AI conferences. UK home rate students are candidates with unrestricted access on how long they can remain in the UK (e.g. have settled status or indefinite leave to remain etc.). The Tuition Fee status is determined by the university’s Registry at point of application. Entry requirements and start date Successful applicants are expected to have a First Class (4-year) Undergraduate degree and Distinction level Master’s degree (or equivalent degrees) in a relevant scientific or technical discipline, including computer science, engineering, mathematics, physics, statistics, as well as biological sciences, or medicine. Expected core skills include mathematics, programming, statistics, or data science. We would like to encourage students from groups that are currently underrepresented in postgraduate science research to apply, including black and minority ethnic students and those from a socio-economically disadvantaged background. Start date is up to 20 January 2025 Contact: Centre Manager: Britta Ross, b.rossimperial.ac.uk Home/UK PhD tuition fees, an annual bursary (UKRI rate £21,237 for 2024-25) per year, for at least 3 years, with option to extend to max 4 years, subject to satisfying progress requirements