Heidi King, Jon Garibaldi and Simon Rogerson (UK)
This paper discusses the status of intelligent computer systems in medicine. A definition of intelligent systems will be given, and an explanation of how these systems differ from other software. The relationship between a clinician and an intelligent system will then be examined, in order to decide whether an intelligent system should be regarded as a partner or a tool. The question of whether intelligent systems provide better care for patients than clinicians will be analysed. Finally, an ethical analysis will be made of whether clinicians should be required to take advice from intelligent systems that could provide better care than they themselves.
Few intelligent systems are actually used routinely in medicine (King et al, 2001). The Vienna General Hospital has within it a group of researchers working in the field of intelligent systems. This research group has produced six fully functional expert systems that are routinely used within the hospital. Previous has shown that this is the most prolific research group world wide in implementing intelligent systems into routine use in a medical environment. This research is concerned with understanding the issues faced by a range of clinicians associated with the Vienna General Hospital with respect to the usage of intelligent computer systems in medicine within their fields of expertise.
Defining Intelligent Systems
Intelligent systems are based upon Artificial Intelligence (AI). A widely accepted definition of AI is one given by Marvin Minsky in 1968: “Artificial intelligence is the science of making machines do things that would require intelligence if done by men”. An ‘ordinary’ computer will follow a set of instructions in a specified order to complete a task. Intelligent systems often have learning capabilities and use computing power to attempt to model human thought processes.
Intelligent systems have been created for many and varied uses. They may be programmed to detect the onset of adverse conditions and alert medical personnel to changes in the patient’s state, for example the “Sentinel” anaesthesia monitor (Jones et al, 2001) provides automated decision support for anaesthesia monitoring. Another use may be to advise on medical practice, e.g. the ICONS system suggests an antibiotic therapy regimen satisfying medical and economic conditions using a process of finding similar cases that have been previously documented and modifying them to meet the requirements of the current patient (Heindl et al, 1997).
It will be necessary, for the purposes of this paper, to specify the characteristics of a ‘typical’ intelligent system. These characteristics can then be examined in order to consider whether this ‘typical’ intelligent system could be considered as a partner or a tool. Future research will examine specific instances of intelligent systems to see how they compare with this ‘typical’ model, and whether these specific examples should be considered as partners or tools. The research concludes by considering whether intelligent systems are different to other types of systems and whether this means they should be treated differently.
Merriam-Webster’s dictionary definitions of a tool include “something (as an instrument or apparatus) used in performing an operation or necessary in the practice of a vocation or profession”. A partnership between two partners is understood to mean a relationship “usually involving close cooperation between parties having specified and joint rights and responsibilities”. The relationship between a clinician and a specified ‘typical’ intelligent system will be examined, in order to decide whether that system should be regarded as a partner, a tool, or a hybrid of the two.
Some studies have shown intelligent systems to provide better care than clinicians for patients (e.g. Hamilton et al, 1996; Narus et al, 1995; Scott, 1999). Johnson & Mulvey (1995) argued that computer decision systems give assistance in coping with complexity, enabling decision making to be more informed. The field of knowledge in medicine is enormous and so support may be essential for clinicians to do their jobs effectively.
Clinicians have a duty to treat and care for their patients to the best of their abilities (the principle of beneficence). The General Medical Council (GMC) in the UK considers that one of the duties of a doctor registered with the GMC is to “work with colleagues in the ways that best serve patients’ interests”. If an intelligent system is to be considered as a partner, and if that intelligent system is able to provide the best possible care for patients, it can be argued that clinicians have a duty to use these systems whenever possible. At present there is no mention made to intelligent, or other, computer systems within the GMC’s standards of practice. This is an example of Moor’s policy vacuum (Moor, 1985) whereby advances in technology provide users with new capabilities without ethical policies having been formulated to guide those users in their conduct.
Hamilton, P. W., Anderson, N. H., Diamond, J., Bartels, P. H., Gregg, J. B., Thompson, D. & Millar, R. J. (1996) An interactive decision support system for breast fine needle aspiration cytology. Analytical and Quantitative Cytology and Histology 18,3:185-190
Heindl, B., Schmidt, R., Schmid, G., Haller, M., Pfaller, P., Gierl, L. & Pollwein, B. (1997) A case-based consiliarius for therapy recommendation (ICONS): computer-based advice for calculated antibiotic therapy in intensive care medicine. Computer Methods and Programs in Biomedicine 52,2:117-127
Johnson, D. G. & Mulvey, J. M. (1995) Accountability and Computer Decision Systems. Communications of the ACM 38, 12: 58-64
Jones, R. W., Harrison, M. J. & Lowe, A. (2001) Computerised anaesthesia monitoring using fuzzy trend templates. Artificial Intelligence in Medicine 21,1-3:247-251
King, H., Garibaldi, J. & Rogerson, S. (2001) Towards the identification of intelligent systems currently in routine use in medicine and dentistry. Proceedings of the Eusflat Conference 5-7 September, 2001
Moor, J. (1985), What is computer ethics? Metaphilosophy 16,4: 266-279
Narus, S. P., Kuck, K. & Westenskow, D. R., (1995) Intelligent monitor for an anesthesia breathing circuit. Proceedings of the Annual Symposium on Computer Application in Medical Care;96-100
Scott, J. A. (1999) Using artificial neural network analysis of global ventilation-perfusion scan morphometry as a diagnostic tool. American Journal of Roentgenology 173,4:943-948