Deontics Clinical Futures

Science & Bibliography

The team behind Deontics have many years of research and clinical experience aimed at solving the practical problems of implementing clinical guidelines in real clinical environments, effectively, and at scale.

Deontics products are built solidly on this scientific foundation. The company grew out of a research and development laboratory with a reputation for being at the forefront of academic research in Artificial Intelligence and Cognitive Psychology as applied to clinical decision support systems.

There are four main strands of scientific work that underpin the Deontics technology stack:

1 — The process of decision-making & decision framing in the real-world

Traditional theories of the most rational choice to make in particular well-specified circumstances give good mathematical analysis where the probabilities and utilities of all outcomes and options are known, but this is rarely the case in any complex real-world situation. Such theories say nothing about how to frame a decision – the process of working out what decision needs to be made, what the various options are and the implications of each.

The Deontics approach is based on a richer theory of the way people – especially clinicians – actually make decisions in the real world, where probabilities and utilities are often unknown and may have outcomes of radically different types that cannot easily be compared. Our cognitive framework, based on studies of clinical decision-making, encompasses the full range of processes surrounding framing, analysing, making and following up decisions. The PROforma logic model developed by the Deontics team for representing clinical pathways is directly based on this cognitive framework, in order to provide knowledge representation for processes surrounding clinical decisions and workflows which fits the way clinicians naturally think.

Selected References

Fox, J (2016). Cognitive systems at the point of care: the CREDO program

Retrospective review of the work of the team behind Deontics over more than 20 years in modelling and understanding clinical thinking and developing computer systems to assist clinicians.

Fox, J., Glasspool, D. W., Patkar, V., Austin, M., Black, L., South, M., et al. (2010). Delivering clinical decision support services: there is nothing as practical as a good theory. J. Biomed. Inform. 43, 831–843.

Shows how the Domino model and associated cognitive theory successfully underpins a wide range of real-world clinical applications.

Fox, J. and Glasspool, D. W. (2006). Knowledge, arguments, and intentions in clinical decision-making. In R. Paton & L. McNamara (Eds.) Multidisciplinary Approaches to Theory in Medicine. Amsterdam: Elsevier. pp. 103-129.

Discusses the full range of processes that must be taken into account in order to understand clinical decision making, and shows how the Domino model and Argumentation Logic capture them.

Fox, J., Beveridge, M., and Glasspool, D. W. (2003). Understanding intelligent agents: analysis and synthesis. AI Commun. 16, 139–152.

Shows how the PROforma language is based on the Domino model, and how this underlies its clinical usefulness.

Fox, J., and Das, S. (2000). Safe and Sound: Artificial Intelligence in Hazardous Applications. Cambridge: American Association for Artificial Intelligence and The MIT Press.

Original introduction of the Domino model and PROforma.

Back to top

2 — Multi-criteria and non-numeric decision theory

Understanding how to decide between options that are not strictly comparable, or where the evidence supporting them is uncertain, or is qualitative (e.g. “the patient has a poor prognosis” as opposed to a specific numeric probability of relapse) is well known to be very difficult and is not well covered by many formal approaches. However this is how many real world decisions – especially in medicine – look.

The Deontics team has been at the forefront of research in an approach known as Argumentation Logic, now widely recognised to be an attractive way to understand human deliberation. Argumentation provides a natural and expressive way of making decisions based on qualitative criteria when that is all that is possible, yet can incorporate quantitative probabilities and utilities when this is practical and helpful. Moreover it naturally allows knowledge from different sources (comparable to different “rule sets” in a more traditional approach) to be combined and added freely. Argumentation is a natural way to formalise clinical knowledge, and is particularly suited to presenting complex criteria in a way that is easy for clinicians to take in.

Selected References

Fox, J., Glasspool, D. W. and Bury, J. (2001). Quantitative and Qualitative Approaches to Reasoning under Uncertainty in Medical Decision Making In: S. Quaglini, P. Barahona, S Andreasson (Eds.): AIME 2001. Springer-Verlag, Berlin. pp. 272-282.

Shows how argumentation logic, which has become a strong area of research in recent years, is ideally suited to medical reasoning.

Fox, J., Krause, P., and Elvang-Gøransson, M. (1993). “Argumentation as a general framework for uncertain reasoning,” in Proceedings of the Ninth Conference Annual Conference on Uncertainty in Artificial Intelligence (UAI-93). San Francisco: Morgan Kaufmann.

Back to top

3 — Clinical workflows and processes

Although there are well-established representations for business processes and workflow (simple ones like flowcharts or more complex like BPMN), clinical processes are not like manufacturing or business processes, just as patients are not like cars on a production line; a different type of representation is needed to make it easy to express the inherent flexibility, and to embrace uncertainty, conflict and continual change.

In particular over many years studying the processes and workflow in clinical practice we have found that a more practical representation is one that focusses on the patient and clinician, and the processes and decisions they must undertake, rather than on the institution, frameworks and resources that they must work within.

Our clinical workflow language, PROforma, specifies tasks, decisions and data and the constraints between them. It is based on our standard framework of the extended process surrounding framing, making and following up decisions, and has developed into a highly capable but simple and concise model for describing processes that may be quite different for each individual traversing them.

Selected References

Fox, J., and Das, S. (2000). Safe and Sound: Artificial Intelligence in Hazardous Applications. Cambridge: American Association for Artificial Intelligence and The MIT Press.

Introduces the PROforma language with a detailed motivation for its structure and approach.

Back to top

4 — Human-computer interaction and interface design for clinical contexts

The Deontics team has carried out much research on practical techniques for communicating effectively between computer and clinician. The clinical environment is often pressured and complex, and it may be chaotic and distracting.

The amount of information that a clinician may need to filter and take in to make a good decision can be huge. We have focussed on the areas of clinical decision making that put the most strain on clinicians’ ability to take in and understand information (areas of highest “cognitive load”) and developed ways to allow the computer to take some of the strain without taking over the clinician’s job – clinicians are excellent at making decisions, the role of the computer is to give them the information they need to do so, at their fingertips.

Deontics user interfaces capitalise on the ability of the Argumentation Logic approach to give the reasons for and against each option in a clear and comprehensible way, and allow clinicians to explore the consequences of different choices quickly and intuitively.

Selected References

Glasspool, D., Oettinger, A, Braithwaite, D. and Fox, J. (2010). Interactive Decision Support for Risk Management: a Qualitative Evaluation in Cancer Genetic Counselling Sessions. In press, Journal of Cancer Education.

Analysis of a user interface design using many of the principles currently employed in Deontics technology. The principles are evaluated in actual use by a group of clinicians.

Glasspool, D. W., Oettinger, A., Smith-Spark, J. H., Castillo, F. C., Monaghan, V. E. L. and Fox, J. (2007). Supporting medical planning by mitigating cognitive load. Methods of Information in Medicine, 46 (6): 636-640.

Experimental analysis of some points of high cognitive load involved in making naturalistic clinical decisions, and the degree to which they can be mitigated by effective user interface design.

Back to top