Ledley and Lusted: logic and probability of diagnosis
On 3 July 1959 Robert Ledley and Lee Lusted published 'Reasoning Foundations of Medical Diagnosis' in Science. They broke diagnosis into symbolic logic linking symptoms to diseases, probability by an adaptation of Bayes' formula, and value theory for choosing a treatment, and showed the computation on edge-notched cards sorted by hand.
Why it matters
Diagnosis was described as a computation that could be handed to a machine: from medical knowledge and the patient's symptoms to the probability of each disease and the expected value of each treatment. The authors say plainly that the method in no way implies a computer can take over the physician's duties; rather, the physician's task may become more complicated.
Ledley was a mathematician in the Data Processing Systems Division of the National Bureau of Standards and taught at George Washington University; Lusted was a radiologist at the University of Rochester School of Medicine. The examples were simplified from the Massachusetts General Hospital clinicopathological exercises printed in the New England Journal of Medicine. For the choice of treatment the authors propose von Neumann's theory of games. An edge-notched card is made for every diagnosed patient; the share of a disease is the thickness of the sorted deck, and the authors reckon that a clinic diagnosing over a hundred patients a day will have a thousand cards in ten days. The paper mentions a large electronic computer as a possibility but names no machine. What the record does not claim: any measurement, since the paper has no patients, accuracy or comparison with physicians, and marks its numbers as hypothetical or illustrative. That later diagnostic programs rested on this paper: it is absent from the reference lists read, those of the Leeds paper of 1972 and the MYCIN paper of 1975.