Impact of a system of computer-assisted diagnosis. Initial evaluation of the hospitalized patient.

Computer assisted diagnosis was evaluated for its ability to hasten the time to initial diagnosis and reduce laboratory studies. The time to diagnosis averaged 2.8 days for a control group and 1.9 days for the group in which the computer printout was made available. In a small group in whom the diagnosis was not established until after 48 hours, there was a further trend to reduction in number of days to diagnosis and irrelevant laboratory studies. The ability of the computer to include the final diagnosis in the initial list of differential diagnoses remained unchanged for patients whose conditions were diagnosed before or after 48 hours, while the physicians' ability decreased. In a university hospital setting, computer-assisted diagnosis may be most beneficial for patients whose conditions have not been diagnosed within 48 hours of admission.