Prediction of Infrequent Prolonged Waiting Time for Outpatients in Blood Drawing Process

Waiting time from reception to patient-call is recognized as a key factor in outpatient satisfaction. While several studies have attempted to predict waiting times, our study specifically focuses on waiting time for outpatients from reception to blood drawing process. The waiting time for blood drawing can exceed typical values due to irregular events occurring during the blood drawing process. The primary objective of this study is to accurately predict these prolonged waiting times. A significant challenge is that such prolonged waiting times are infrequently observed in historical data. In this study, we identify situations where the waiting time takes large values through quantitative data analysis, and propose a method utilizing Deep Local Linear Regression for the prediction. This method incorporates the tendency that the waiting time increases with the number of patients waiting. The effectiveness of the proposed method is evaluated by using actual data collected from a hospital. We demonstrate that for predicting waiting times exceeding 30 minutes, the proposed method improves prediction accuracy by at least 10% compared to the random forest.