B. Spyropoulos et G. Papagounos, A THEORETICAL APPROACH TO ARTIFICIAL-INTELLIGENCE SYSTEMS IN MEDICINE, Artificial intelligence in medicine, 7(5), 1995, pp. 455-465
The various theoretical models of disease, the nosology which is accep
ted by the medical community and the prevalent logic of diagnosis dete
rmine both the medical approach as well as the development of the rele
vant technology including the structure and function of the A.I. syste
ms involved. A.I. systems in medicine, in addition to the specific par
ameters which enable them to reach a diagnostic and/or therapeutic pro
posal, entail implicitly theoretical assumptions and socio-cultural at
titudes which prejudice the orientation and the final outcome of the p
rocedure. The various models -causal, probabilistic, case-based etc. -
are critically examined and their ethical and methodological limitatio
ns are brought to light. The lack of a self-consistent theoretical fra
mework in medicine, the multi-faceted character of the human organism
as well as the non-explicit nature of the theoretical assumptions invo
lved in A.I. systems restrict them to the role of decision supporting
''instruments'' rather than regarding them as decision making ''device
s''. This supporting role and, especially, the important function whic
h A.I. systems should have in the structure, the methods and the conte
nt of medical education underscore the need of further research in the
theoretical aspects and the actual development of such systems.