Two aspects of computerized patient history that improve bedside pulmonary diagnosis

This thesis project involves two aspects of research in the area of diagnostic decision making of cardio-pulmonary disease. This study is concerned with a set of logic frames that have been written to detect a group of 41 pulmonary and cardiology disease. When the diagnostic frames were first developed, there were 58 frames for 29 diseases. There were two frames for each disease. The first dealt with patient history, and the second dealt with physical exam, laboratory, and radiology findings. The a posteriori probabilities generated by the history frames are used as the entrant a priori probabilities of the corresponding clinical data frames. The history frames have evolved to include more than the original 29 disease states, whereas the clinical data frames have continued to include only the original 29 diseases. The first part of the study deals with the problem of gathering a computerized patient history that could then be available for use by the pulmonary diagnostic history frame. Two different computerized data gathering techniques, that captured a patient history directly from the patient using a terminal mounted near the bedside, were compared. One of the collection modes is hierarchical questionnaire that consists of a conditional branching set of fixed frames (GQAP). The other history gathering program utilized DDA, a decision-driven data acquisition system. The goal of this project is to determine if there is a measurable difference between the two data collection modes in 1) time to complete, 2) number of questions asked, or 3) accuracy of decision made form the entered data. Physical exam data by the physician into a computer terminal were the next type of data that was scheduled to be collected at the patient's bedside. These data would then be available for utilization by the clinical data frames that diagnose cardio-pulmonary diseases. Before the collection of physical exams began, the Bayesian statistics used in the clinical data frames for chest X-ray findings needed to be revised. It was felt that revising the statistics for the radiology finding, using a data base of actual patients who had entered LDS Hospital, would create a better set of statistics that the statistics that were originally estimated by the medical experts. The second part of this study involves the revision of chest X-ray finding in the pulmonary clinical data frames, and the evaluation of the effect the revised statistics had upon the accuracy of the diagnostic system.