A tribute to ‘inspiring, witty and pioneering’ John Fox

John Fox was a pioneer, leader, educator, serial entrepreneur.

He was also an advocate in the area of high-quality knowledge-based clinical decision support systems (CDSS).

He built his career at Cancer Research UK and later Oxford University. He was also the founder and “driving force” of OpenClinical.

In late August this year John died from cancer, leaving behind an incredible legacy that will continue to have an impact in for years to come.

John’s greatest fame comes from the PROforma formal computer language for modelling clinical decisions and processes, which is rooted in cognitive theory and follows rigorous software engineering practices.

He implemented and trialled more than 15 CDSSs using PROforma, providing safe and sound decision-support to different healthcare organisations in different countries.

John also served our community as an editorial board member of the Journal of Biomedical Informatics for the past eight years.

From his roots in psychology and cognition, he developed a powerful formalism, PROforma, that reflects our understanding of human decision making, and has clear and consistent semantics.

John was an interdisciplinary researcher with theoretical interests in AI and computer science and an applied focus in medicine and medical software engineering.

After training in experimental psychology at Durham and Cambridge Universities in the UK, and post-doctoral fellowships at CMU (working with AI founders Herb Simon and Allen Newell) and Cornell in the USA, he joined the Medical Research Council in Sheffield UK to establish a programme of research in decision-making and clinical decision support.

In 1981 he set up the Advanced Computation Lab at the Imperial Cancer Research Fund (now Cancer Research UK) where research focused on applications of knowledge-based systems in oncology and primary care.

However, the group was explicitly multidisciplinary and made contributions in basic computer science, safety critical systems, AI and medical informatics, and developed a number of technologies which have been commercialised.

In 1996, he and his team were awarded the 20th Anniversary Gold Medal of the European Federation of Medical Informatics for the development of PROforma, one of the first formal computer languages for modelling clinical decisions and processes.

In 2007, John moved to the department of Engineering Science at Oxford University to set up an interdisciplinary research collaboration with Edinburgh University and UCL, and a clinical informatics incubator at the Royal Free Hospital in London.

The incubator ran a number of successful trials of AI technologies and spun out two companies to take the technologies into clinical use: OpenClinical CIC, and Deontics Ltd.

Standout contributions to research and health

John developed and left for our community many achievements, including symbolic reasoning and argumentation for decision support.

John’s early contribution to symbolic DSS came out of his time as a postdoctoral student with Simon and Newell, who established the paradigm as a way of modelling cognition.

John had early and famous clashes in the journals about whether cognition should be simulated symbolically or probabilistically.

This work developed into a full model of the cognitive processes surrounding the framing and making of decisions, and planning and execution of action in complex and uncertain domains such as medicine, which would come to underpin much of John’s work including the PROforma language.

John also had a strong interest in AI safety. Even now, AI safety remains an emerging issue. Twenty-one years ago, in 2000, John published the book ‘Safe and Sound’ based on research in the RED (rigorously engineered decisions) project carried out for the UK safety critical systems initiative.

In the book he presented an arguably more sophisticated treatment than anything available today. Toward this goal, he formulated a stack of principles (The CREDO stack) that can systematically be used to ensure that CDSS applications are safe and sound.

John was also founded OpenClinical a not-for-profit company supported by Oxford University Innovation whose mission is to promote the peer review and dissemination of AI models of best clinical practice based on a sound model of the life course of small fragments of executable knowledge John called “publets”.

It includes a repository of executable decision-support services and explicit (usually declarative) knowledge that can be shared for improving patient care through innovative AI and knowledge engineering techniques, such as PROforma.

A key objective is to empower and encourage healthcare professionals to author, share, critique, trial and revise these executable models of best practice in an ethical, transparent and trustworthy way.

OpenClinical is an operational publishing platform that uses the PROforma language to capture human expertise in decision-making, care planning and other cognitive skills.

OpenClinical’s open-access, open-source repository carries over fifty diverse examples of computer-executable guidelines and other artefacts.

PROforma – a method and language for specifying clinical guidelines and protocols, rooted in robust cognitive theory and software engineering is an executable process modeling language used to build and deploy decision support systems, clinical protocols, guidelines, and other clinical applications in a machine-executable format.

The PROforma formalism and method for modelling clinical knowledge is based on the domino model, a generalised model of clinical decision making and protocol management.

PROforma is a first-order logic formalism extended to support decision making and plan execution. It incorporates non-classical logics (e.g., modal logic, temporal logic, deontic logic), LA, logic of argument and LOT (logic of obligation and time) to support decision making and action control.

Work leading to the introduction of PROforma started in the late 1980s with funding from European agencies. PROforma development was also supported by Cancer Research UK with the purpose of building decision support and intelligent agents to improve cancer care.

PROforma comprises a formal specification language, a knowledge representation language and a suite of authoring tools and execution software for building applications. It forms the knowledge representation layer in the CREDO knowledge stack. CREDO is a framework for understanding human expertise and for designing and deploying systems that support cognitive tasks such as situation and risk assessment, decision-making, therapy planning and workflow management.

The Tallis technology stack was developed by John’s group at Cancer Research UK to support the authoring and execution of PROforma applications.

PROforma has been validated in a wide range of applications in diverse clinical settings and specialties, with trials of these published in high impact peer-reviewed journals. Trials have studied patient workup and risk assessment; decision support (e.g., diagnosis, test and treatment selection, prescribing); adaptive care pathways and care planning.

More than 15 clinical applications have been built using PROforma technology, including:

• ORAMA (Optimal Renal Anaemia Management Assessment)
• Retrogram HIV treatment
• MACRO as a web-based clinical trial management system)
• ARNO for cancer pain control
• Evidence without Effort Clinical decision support
• RAGs for genetic risk management
• ERA for early referral for suspected cancer
• CAPSULE as a decision support for prescribing in primary care
• e-Asma as a decision support for asthma management
• HOMEY for the development of natural language and speech interfaces for patient home monitoring and clinical services
• LISA a decision support for collaborative care in childhood acute lymphoblastic leukemia
• CREDO, a clinical trial of decision-making and workflow management in the care of women at risk of or with a proven diagnosis of breast cancer
• And a computer-interpretable version of the AACE, AME, ETA medical guidelines for clinical practice for the diagnosis and management of thyroid nodules

In 2020, PROforma and OpenClinical were 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. This project demonstrated a novel and promising approach to the rapid translation of clinical guidelines into point-of-care services, and a foundation for rapid learning systems in many areas of healthcare.

John also co-founded two companies that develop tools supporting clinical decision-support system implementations based on PROforma.

The first PROforma-based technology for authoring and executing clinical guidelines was Arezzo. This software was designed and built by Cancer Research UK and InferMed to develop enactable clinical guidelines.

Deontics, of which John was a founder and a director, is a spinout company from Oxford University, University College London Business and Cancer Research UK.

It produces a mobile-friendly CDS application supporting clinical and operational pathway management and treatment decision making, along with analytics tools for running PROforma guidelines against longitudinal retrospective patient data, and a PROforma guideline authoring system.

A witty and humorous family man

Away from work, John was a much loved family man and friend.

Over dinners and family visits, and when chatting before Skype/Zoom calls, John shared his way of seeing the world through his wit and humour, which is what we all remember.

John cared about what is important – family, education, ethics, healthcare and other societal challenges.

He not only cared, but did something about those challenges within his grasp, especially where he felt decision making could be improved.

John was a deeply principled and good man, and was utterly passionate about his work and seeing his informatics research translate and touch the lives of real people through improved healthcare.

He was inspiring, focused, full of energy and had so much passion for research, and was a mentor to many students and young researchers. He was a brilliant speaker and generously shared his knowledge and experience with others, training many PhD students, some of whom are now leaders in our field.

He was an ethical man and a man of strong principles. We will miss John, his sense of humour and his thoughtful insights, questions and comments.

He was a wonderful mentor and an inspirational person in our lives.