MYCIN diagnoses infections
Shortliffe's MYCIN uses about 600 if-then rules with certainty factors to recommend antibiotics, matching specialists in blind evaluation but never used on a patient.
what had to happen · 3 events back to 1943
Every event this one built on, transitively, in order. Direct influences are marked.
I · Foundations · 3
MYCIN was Edward Shortliffe's Stanford doctoral project, finished in 1974, and it is the expert system that most later ones copied. Given a patient's symptoms and laboratory results, it asked questions, applied several hundred rules of the form "if the organism is gram-positive and grows in chains, then there is suggestive evidence it is streptococcus", and recommended an antibiotic and dose. Because medical knowledge is uncertain, each rule carried a certainty factor, a number between minus one and one, and the program combined them with an algebra Shortliffe invented for the purpose.
In a 1979 evaluation, MYCIN's recommendations were judged acceptable by a panel of experts more often than those of the infectious-disease specialists it was compared with. It was never used in a clinic. Liability was unresolved, the program needed a mainframe, and physicians did not want to type answers into a terminal.
MYCIN's real product was its architecture. Separating the rules from the engine that applied them produced EMYCIN, the "empty MYCIN", and the idea of the expert-system shell, sold by the dozen in the 1980s. Its certainty factors were later shown to be a rough approximation of Bayesian reasoning, which Judea Pearl put on firm footing in 1988.
what it led to · 3 events downstream, through 1987
Built on it directly:
And, through them, by era: