Computerized rapid analysis of pulmonary function test: use of a least mean squares correlation for interpretation of data.

Although systems for automated calculation and interpretation of screening pulmonary function tests (PFT) have been discussed by previous investigators, these systems are often unavailable to other users, require large or hybrid computers, or are difficult to modify. The use of a branching decision tree for PFT interpretation may also limit the number of correlations possible and make revision of the program difficult because changes in the proximal branches of the tree require a rewriting of the entire distal program. We have developed a system based on a minicomputer which uses a least mean squares analysis for rapidly analyzing the patient's PFT data. This program considers the following types of PFT data singly and in combination: percentage predicted total lung capacity and/or alveolar volume, forced expiratory volume in one second to forced vital capacity ratio (before and after bronchodilator or 80% helium challenge), and percentage predicted single breath diffusing capacity for carbon monoxide. The interpretation program (IP) then selects the 'best interpretation' for the patient by computing the variance between the patient's data and data from a menu of over 120 interpretation statements. During the past two years we have used this system for screening over 3000 patients. In about 90% of patients the computer selected interpretation statements which were acceptable to a sub-specialty pulmonary physician. The use of this IP for screening purposes has facilitated the efficient use of laboratory personnel and equipment.