A Health Care Domain Setting

We have chosen to use a domain setting from health care.  We draw primarily from the proposal document [Szolovits 94] of a project called  “Guardian Angel” (GA). The document, also referred to as the “manifesto”, envisions a multi-agent information system infrastructure for health care which enables patients to have much greater control over their health information and care processes. [ga.org]

The Guardian Angel project aims to “construct information systems centered on the individual patient instead of the provider, in which a set of “guardian angel” software agents integrates all health-related concerns, including medically-relevant legal and financial information, about an individual. This personal system will help track, manage, and interpret the subject's health history, and offer advice to both patient and provider. Minimally, the system will maintain comprehensive, cumulative, correct, and coherent medical records, accessible in a timely manner as the subject moves through life, work assignments, and health care providers. Each GA is an active process that performs several important functions: it collects patient data; it checks, interprets, and explains to the subject medically-relevant facts and plans; it adapts its advice based on the subject's prior experiences and stated preferences; it performs "sanity checks" on both medical efficacy and cost-effectiveness of diagnostic conclusions and therapeutic plans; it monitors progress; it interfaces to software agents of providers, insurers, etc.; and it helps educate, encourage, and inform the patient. All this serves to improve the quality of medical decision-making, increase patient compliance, and minimize iatrogenic disease and medical errors”. [Szolovits 94]

We believe that the GA project vision description can be fashioned into an exemplar that meets most of the requirements set out in Section 2. The domain setting offers plenty of opportunities for software agents to be introduced.  The GA vision paints a detailed picture of how one might use software agents in this setting.  The technology vision is described at a high level, so there is considerable room for alternative ways of defining requirements. In particular, the notion of software agent is quite open. As an exemplar it highlights the need for the methodologies to specify the appropriate agent notions at appropriate development stages. The primary focus is on the care relationship between the patient and the physician, which is to be enhanced by the GA software agent.  There can be many other agents providing additional supporting, facilitating, mediating, monitoring or administrative roles.  These other agents could be associated with the physician, the parent (of children patients), other care providers (nurses, administrators), or organizations (hospitals, agencies, etc.).  Each of these could itself be made up of multiple agents.  There can be interactions with many remote databases or knowledge bases.  Some of the agents could be mobile.  Numerous design alternatives exist. – how functions and responsibilities are distributed among agents, how they communicate and coordinate, etc.  These are heavily constrained by issues of privacy, security, ownership and control, and desires for autonomy.  Some portions of the project vision have been carried through to design and implementation. 

For our purposes, we will focus on issues arising from the vision description.  We are not interested in the particular systems and designs that might be best for GA.  Instead, we ask what kinds of systematic and principled approach can guide us towards those designs and solutions.  The rich descriptions of the domain setting therefore provide us with the material for our methodological studies. 

The vision provides us with vivid depictions of likely scenarios.  We have selected excerpts and itemized them for reference (in Section 4).  The domain is readily comprehensible to the layperson.  Readers may be able to relate some of the scenarios to personal experiences. Non-technical issues abound, often driving the technical ones.  The core issue is a human social one – the desire for more patient control and autonomy.  The technology could potentially improve health outcomes and lower overall costs.  Legal, socio-cultural, political forces might affect how the technology develops and evolves.

Among software engineering issues, security and reliability will be prominent.  The setting will bring out the capabilities of software agents in addressing them, as well as for many others such as scalability, performance, maintainability, etc. Although the vision is expressed in terms of software agents, much of the vision can potentially be achieved using more conventional software technology and methods.  The extent to which different issues are addressed could be quite different.  The contrast between agent-oriented approaches and non-agent-oriented approaches can then be examined in considerable detail, both in terms of the technologies and the development methodologies.

To take a neutral stance, any mention of software agent should be interpreted more generically as “software unit”, with the choice of agent notions and capabilities to be determined by the methodology in question.  The application setting is illustrative and representative of large-scale information systems, with many stakeholders and many interacting systems and parts. It can be used to cover methodological issues quite extensively.  The premise of patient-centered care is provocative and goes beyond the mere automation of existing practices.  This helps to raise many questions and can lead to innovative solutions that may involve the technological as well as non-technological. We therefore believe it is a good vehicle for examining how well the different methodologies can guide us along these explorations toward realized systems.