PROforma

INTRODUCTION

PROforma was developed at Cancer Research UK for the general purpose of building decision support and intelligent agents. The technology includes the PROforma language, a formal specification language (as that term is used in software engineering), a knowledge representation language (as understood in AI) and a set of prolog and Java tools for building applications in the language.

PROforma is essentially a first-order logic formalism extended to support decision making and plan execution, but it also incorporates a number of well known features of non-classical logics (e.g. modal logic, temporal logic) and two novel logics (LA, logic of argument and LOT, logic of obligation and time) to support decision making and action control.

The technology includes a suite of PROforma authoring and execution software that incorporate CASE and verification tools. It has been shown to meet specific requirements of medical applications though the language and tools are generic.

PROforma is a continuing area of active research at Cancer Research UK, particularly for safety-critical applications. PROforma is the platform for a number of clinical applications developed by the lab. (including REACT, RAGs, ERA). PROforma applications demonstrating guideline enactment over the WWW are currently under development.

The language, technology and illustrative applications are comprehensively described by J Fox and S Das in Safe and Sound: Artificial Intelligence in Hazardous Applications, AAAI and MIT Press, 2000.

The work was awarded the 20th Anniversary Gold Medal of the European Federation of Medical Informatics in Copenhagen in 1996.

BACKGROUND

Work leading to the design and implementation of the clinical guideline representation and interchange format, PROforma was carried out in a series of projects largely funded by European agencies, starting in the late 1980s. The work undertaken (over a period of around thirteen years) is described in detail in a paper published in 1998, Decision support and disease management: a Logic Engineering approach and in a new book, Safe and Sound: Artificial Intelligence in Hazardous Applications by John Fox and Subrata Das (AAAI and MIT Press, 2000). PROforma itself was a major result of the EC 4th Framework PROMPT project, which was completed in 1998.

The domino model

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. The model was developed largely developed under the RED (Rigorously Engineered Decisions) project, part of the Safety Critical Systems Programme funded by the UK Government's Department of Trade and Industry and the Engineering and Physical Sciences Research Council. RED explored the use of formal theories of reasoning and decision making, rigorous specification techniques, and systematic design and implementation of software in order to optimise soundness and safety in systems development.

The domino model, shown below, informally summarises the formal framework for describing clinical decision making and process management including task scheduling used in PROforma.

The Domino modelThe Domino model

Given a set of beliefs, an 'agent' may infer goals (e.g. to diagnose or treat a disease) and various solutions to these goals. If there are multiple options (such as alernative diagnoses or treatments) the system must consider the arguments for and against these alterantive and make decisions based on the validity of each of the arguments. A decision may commit to a new belief (e.g. what is wrong with the patient) or to a plan of action to achieve a goal (e.g. a clinical care plan) consisting of a set of actions carried out over time.

Each node of the domino represents information or data that are relevant to a particular clinical situation e.g. facts about a patient's history, decisions and other tasks in progress or actions which are planned.

Each arrow represents an inference procedure that can derive conclusions from available information. The inference procedure uses information of the type shown at the arrow's tail in conjunction with information from a patient record and/or general medical knowledge base, in order to generate information of the type shown at its head.

The decision logic is subsumed by the four nodes to the left of the domino model diagram; the processes required to schedule and enact tasks are summarised by the two nodes to the right.

PROforma software combines the logic-based representation developed in RED with an object-oriented representation of clinical tasks in order to achieve a balance between the clarity and precision of a formal specification language, and a more intuitive representation of clinical concepts and procedures. The domino model provides a good framework for defining the formal semantics of PROforma, but it is rather abstract. We have therefore developed a complementary view in which clinical procedures are conceived in a way that emphasises the clinical activities that the application is designed to support. These are captured naturally using the concepts of object-oriented programming to model "tasks" which are exploited in the Tallis process modelling tools (see this site) and the commercial platform Arezzo(R) developed by InferMed (www.infermed.com).

References
Fox J, Patkar V, Thomson R. Decision Support for Healthcare: the PROforma evidence base Informatics in Primary Care 2006; 14:49-54 [CRUCK]

Sutton DR, Fox J. The Syntax and Semantics of the PROforma guideline modelling language. J Am Med Inform Assoc. 2003 Sep-Oct;10(5):433-43. [PubMed]

Fox J., Thomson R. Decision Support and Disease Management: A Logic Engineering Approach. IEEE Transactions on Information Technology in Biomedicine, vol 2 no. 4, December 1998, pp217-228. [PubMed]

Fox J. and Das S. Safe and Sound: Artificial Intelligence in Hazardous Applications. Jointly published by the AAAI, Menlo Park, CA, and MIT Press, Cambridge Mass., 2000. ISBN 0-262-06211-9. [MIT Press]

Fox J., Johns N. & Rahmanzadeh A. Disseminating Medical Knowledge-The PROforma Approach. Artificial Intelligence in Medicine, 14, 1998, 157-181. [PubMed]

-Bury J., Fox J., & Sutton D. The PROforma guideline specification language: progress and prospects. Proceedings of the First European Workshop, Computer-based Support for Clinical Guidelines and Protocols (EWGLP 2000), Leipzig 13-14 Nov. 2000.
[CRUCK]

-Sutton DR, Taylor P, Earle K. Evaluation of PROforma as a language for implementing medical guidelines in a practical context. BMC Med Inform Decis Mak. 2006 Apr 5;6(1):20 [PubMed]

Contact
John Fox
Email: john.fox@eng.ox.ac.uk