Surgical diagnosis assisted by a computer

Abstract

This paper describes a study of surgical diagnosis assisted by a computer. The study was carried out in the University of Leeds, involved the Department of Surgery and the Centre for Computer Studies and made use of an English Electric KDF9 computer. A real-life real-time controlled trial was carried out on a consecutive unselected series of 552 further patients. This trial included all patients who presented to the Department of Surgery between January 1971 and August 1972 with abdominal pain of less than 1 week’s duration. The pre-operative diagnosis of the most senior clinician who saw each case was accurate in some 81% of the patients studied. Using the same information (and producing its ‘diagnosis’ before the operation), the computer-aided system proved to be accurate in 91.5% of the cases. The cost of each computer ‘diagnosis’ was around 10 new pence per case. On these grounds it is suggested that the results of this initial study justify further modestly extended trials of the system.