Enhancing Process Mining Accuracy with K-Means Clustering: A Case Study on Sepsis Treatment

Process mining, an emerging discipline, facilitates the extraction and analysis of process data to enhance system understanding and optimization. This study leverages process mining with a machine learning approach to optimize processes, focusing on the treatment of sepsis patients in a hospital setting. By utilizing event log data comprising ۱۵,۰۰۰ events from ۷۵۳ patients, the research employs the K-means clustering algorithm to segment patients into distinct groups based on relevant features, such as age and clinical indicators. The methodology involves data collection, feature selection, normalization, and determining the optimal number of clusters, followed by process extraction using algorithms like Alpha Miner, Heuristics Miner, and Inductive Miner. The results demonstrate that clustering significantly improves the accuracy of process models, particularly with the Alpha Miner algorithm, as evidenced by enhanced log fitness scores. This approach highlights the potential of integrating clustering techniques with process mining to achieve more precise and actionable process models, offering insights into inefficiencies and bottlenecks in healthcare processes. The study suggests further refinement of clustering through expert consultation to enhance feature selection and address data ambiguities, paving the way for improved process optimization in medical settings.

کلیدواژه ها:

Process Mining ، Clustering ، Sepsis ، Process Optimization ، Healthcare