03 Sep 2020
Knowledge engineering for patient care: lessons from COVID-19
A new approach to rapid translation of clinical guidelines into point of care support services has produced a novel pattern for designing and deploying AI systems across healthcare
The COVID-19 pandemic has been a massive challenge to human expertise and organisation, including in the delivery of optimal care to COVID-19 patients. Artificial Intelligence (AI) is being pressed into service to help maintain and disseminate knowledge of best medical practice in the detection, diagnosis and management of COVID-19 infections.
Professor John Fox (pictured left), who retired from the Department of Engineering Science in 2017, has been involved in a pathfinder study of AI to rapidly formalise COVID-19 guidelines into an executable model of decision making and care pathways. This will be an invaluable resource for clinicians, researchers and healthcare providers to develop point of care products and services which embody best clinical practice across the COVID-19 patient journey. It is believed to be the first use of knowledge engineering methods for disseminating best practice in COVID-19 care.
This work follows on from the OpenClinical.net project, a service for disseminating clinical guidelines in executable form to improve quality of care, which was developed during Professor Fox’s time at Oxford. Traditionally, medical research is published in high quality journals and the knowledge produced is disseminated to healthcare professionals via clinical practice guidelines (CPGs). Although a vital way of disseminating up-to-date recommendations for safe clinical practice, CPGs take time to read and absorb, are difficult to keep up to date, and only provide general guidance rather than patient-specific recommendations.
OpenClinical.net is a web-based knowledge-sharing platform which uses an AI language called PROforma and knowledge engineering techniques to capture human expertise in decision-making and care planning, give patient-specific recommendations and empower healthcare professionals to author, share, critique, trial and revise models of best practice. The PROforma models which give these patient-specific recommendations have been validated in a wide range of clinical settings and specialities, with many successful trials published in high impact peer-reviewed journals.
Now the PROforma guideline modelling language and OpenClinical.net platform have been used to create a data model for care of COVID-19 patients, together with executable models of rules, decisions and plans that interpret patient data and give personalised care advice. They have been proved to be an effective combination for rapidly creating the COVID-19 model.
Professor Fox explains, “The Pathfinder 1 project was a successful demonstration of the power of established AI methods to help respond quickly to the COVID-19 pandemic and future similar emergencies. It also demonstrated a novel AI design pattern for rapidly developing sophisticated AI products and services in healthcare and potentially many other fields“.
A report of the work, ‘Rapid translation of clinical guidelines into executable knowledge: A case study of COVID-19 and online demonstration’, was published in the Learning Health Systems journal, with an on-line demonstration of www.OpenClinical.net in June 2020. The Pathfinder 2 project is consolidating the work to date and developing an advanced application which can support patient assessment, clinical decision making and management of care throughout the patient journey “from home to hospital to home”.