The requirements analysis & design for a clinical information system : a formal approach
Following a number of recent far-reaching reforms to the UK NHS, St Thomas' Hospital (where this work was based) introduced a management structure based on the 'Clinical Directorate'. In order to lessen the increased workload commensurate with this measure, it was decided at St Thomas' that a new type of information system - the Directorate Information System or DIS - would introduced. This system was to 'support the business of the clinical directorate'. As part of the DIS project, a small study was set up toinvestigate the problems associated with the introduction of such an information system, and to suggest a design. This thesis reports on the study. The design of information systems in general, and clinical information systems in particular, seems to be an extremely difficult endeavour: many systems development projects end in failure. It is widely considered that the problems lie in inadequate requirements analysis and specification: consequently it was here that the project concentrated most of its efforts. It was recognised that when in use, the terms, quantities, and entities stored and displayed by an information system are interpreted by its users as terms, quantities, and entities in the organisation that is being supported (also called the domain in the thesis). This is perhaps the fundamental requirement of an information system: that it represents the organisation and processes it is to support. To assess the degree to which a design satisfies this requirement entails the development and use of three descriptions, or theories. The first is the theory of the domain; the second is a theory, or specification, of the proposed information system; the third is a theory of the way in which the information system is interpreted into the domain - this is called the interaction theory and is a composition of the first two theories. By inspecting the interaction theory inadequacies in the representation of the domain by the information system can be identified and, if necessary, rectified. There are four ways in which we are encouraged to modify information system designs so that they more accurately reflect the behaviour of the domain. These are called the four developmental motives. Through the use of a well constructed interaction theory, and guided by the desire for system simplicity on one hand and the four developmental motives on the other, an improved information system design can be engineered. For an interaction theory to be constructed and provide useful insight, both the domain theory and the information system specification must be semantically rich. Conventional analysis notations are inadequate for the task: mathematics (in this case set theory) is needed to represent and explore the domain, the information system, and the interpretation of the latter into the former. The construction of a good domain theory is the hardest part of the process. Representing the organisation as it is perceived by workers (in this case clinicians) as a set theoretic construction is fraught with difficulties. However, the judicious use of an adaptation of the scientific method means that we can have increased confidence that the resulting description of the organisation is a reasonable one and is not merely a statement of the analyst's preconceptions and prejudices. The thesis describes in more detail the background to the project, the use of the scientific method to derive a domain theory, the construction of interaction theories, and the engineering of information systems through the use of the four developmental motives. This is done through the use of a large case study which presents, documents, and discusses the theories used in the Directorate Information System project, and describes their evolution.