TASH - Trustworthy Autonomous Systems for Health

点击翻译按钮获取中文摘要

As advances are made in Artificial Intelligence (AI), the technology is being applied to health-related problems. How do we evaluate the trustworthiness of these systems in the health context? This project will push the boundaries of our understanding of trustworthy autonomous systems in health through a mix of computer science and sociotechnical research, driven by the implementation of a real system. The project will develop and expand an existing autonomous system for early detection of deteriorating health in hospital patients, using safety, reliability and acceptability of a machine learning clinical deterioration score (CARDS) in comparison to the existing extensively used basic Early Warning Score (NEWS2). Using this system, and existing trustworthy autonomous frameworks, the project will first develop a further understanding of the healthcare ecosystem into which the system will be deployed along with the practical, professional and societal hurdles to its implementation. The project will then identify variations and extensions required by medical autonomous trustworthy systems leading to the development of a framework for trustworthy systems, focused on the specific needs of the health ecosystem, including organizational requirements, clinical interactions, user interfaces and regulatory bodies, etc.