Fuzzy Set-Based Detection of Hypotension Episodes for Predicting Leaks in Sleeve Gastrectomy

This paper utilizes a fuzzy sets approach for the analysis of arterial blood pressure and detection of hypotension episodes during sleeve gastrectomy surgery. Membership of systolic blood pressure measurements to the set of "low systolic blood pressure" is used for feature construction of predictive variables in predicting leakage after a sleeve gastrectomy procedure. The prediction task is posed as a classification problem. Logistic regression and Takagi-Sugeno fuzzy inference systems are used as the classification tools. Results indicate an increase in predictive performance compared to previous studies using the same data set.