Prediction of gall-stone pancreatitis by computer.

The clinical features at presentation of 53 patients admitted with primary acute pancreatitis due to gall stones were compared with those of 31 patients in whom the disease was due to other causes. Between these two groups 10 significant differences existed. By listing the frequency of symptoms and signs for each group a computer data base was prepared and incorporated into a program used in the differential diagnosis of acute abdominal pain. A program written to predict the presence of gall stones in patients with acute pancreatitis was accurate in 92% of the patients studied. A predictive index devised from the presence of three of the significantly differing clinical features correctly identified 82% of patients with gall-stone pancreatitis. Predicting the presence of gall stones on admission by analysing the presenting symptoms and signs with a computer had an accuracy comparable to that of ultrasonography or radiology and may be of value in the management of patients with acute pancreatitis.