การใช้เหมืองกระบวนการข้อมูลแบบหลายชั้นสำหรับเพื่อวิเคราะห์วิถีการรักษาพยาบาลของผู้ป่วยโรคเบาหวาน

TLDR

This study aims to visualize revisit intervals affecting patient treatment using multi-level process mining and machine learning to identify factors causing missed appointments, and suggests that understanding these factors can improve patient engagement and reduce missed appointments, enhancing care for diabetes patients.

Abstract

This study addresses the challenge of chronic diseases like diabetes, requiring ongoing care. It aims to visualize revisit intervals affecting patient treatment using multi-level process mining and machine learning to identify factors causing missed appointments. Our research shows that patients with improvements have shorter

revisit intervals than those deteriorating, with a P-value of 3.83x10%. A significant challenge is the high number of no-shows. Using PyCaret to compare 14 models, we found that the historical attendance of individual patients is a crucial factor in

predicting no-shows, with an AUC score of 0.80 using the Gradient Boosting Clas-sifier. Our analysis suggests that understanding these factors can improve patient engagement and reduce missed appointments, enhancing care for diabetes patients.