As medical knowledge expands, the expetise of each individual physician and surgeon, which takes so long to acquire, is necessarily relatively limited in scope and is lost for ever at his death. Serious consideration is therefore now being given to the feasibility of using electronic digital computers for the automation of clinical diagnosis. Data input, in the form of symptoms derived from a standard questionnaire and signs derived from a standardised physical examination, would be collated with information already stored previous patients, and on the basis of pattern recognition, a diagnosis (and perhaps even advice about treatment) would be supplied as data output. However, diagnostic perspicacity cannot simplybe reduced to machine logic, because it depends so much on the doctor's own subjective reactions, and on his skill in editing each individual case history as he elicits it and in designing each physical examination accordingly. Also, a case history is accurately described only in spacetime narrative, which makes scoring impossible without preliminary abstraction. I conclude, therefore, that the automation of clinical diagnosis is impossible, and that efforst should now be concentrated on other uses of computers in medicine which are already known to be feasible.
ASJC Scopus subject areas
- Medicine (miscellaneous)