Prediction of morbidity after lung resection in patients with lung cancer using fuzzy logic

Abstract

Background: Preoperative evaluation of patients with potentially resectable non-small cell lung cancer aims to estimate the risk of planned surgery. Evidence of several factors that identify patients at risk for complications from thoracotomy is controversial. The aim of this study was to introduce and implement in medical practice a fuzzy system used in risk assessment of pulmonary resection for lung cancer. Methods: Ninety-one consecutive patients who underwent pulmonary resection for lung cancer were investigated. The overall complication rate was 39.6% (a total of 63 complications were seen in 36 patients). A fuzzy logic model was created with 9 input (presence of chest pain, weight loss, clinical T stage of the tumor, FEV1, serum protein, preoperative arterial partial oxygen pressure and cigarette smoking, erythrocyte sedimentation rate and peripheral blood leukocyte count) and two output classes (high-risk and low-risk groups). The fuzzy classifier's performance was tested. Results: The model was able to predict correctly the occurrence of complications in 22 out of 29 patients in the high-risk group with a sensitivity of 76%, while 9 out of the 52 patients from the low-risk group developed complications (17%). Conclusion: The fuzzy classification system provides an accurate tool to predict complications of resections in patients with non-small cell lung cancer.