Increasing Visibility through Process Mining

Increasing Visibility through Process Mining

Kalyan S. Pasupathy (Mayo Clinic, USA) and David A. Clark (University of North Carolina Hospitals, USA & University of Missouri, USA)
Copyright: © 2014 |Pages: 11
DOI: 10.4018/978-1-4666-5202-6.ch110
OnDemand PDF Download:
List Price: $37.50


Healthcare processes have variability embedded within care delivery systems. Healthcare providers and managers need visibility of the process to understand variability, and make process design changes, to improve safety and efficiency. Process mapping has been traditionally used to study processes, and is based on perceptions. However, to better understand and manage variability, there is an urgent need for process mining techniques. Process mining requires data for validation, to increase understanding and/or change perceptions and improve visibility. Radio frequency identification (RFID) can be used for collecting process-related data and can eliminate data quality issues. This paper discusses process mining and its application in healthcare, summarizes process mapping and RFID, and proposes an integration framework and method to improve visibility.
Chapter Preview


To be able to provide safe, effective, and efficient care in today’s competitive environment, hospitals must have process visibility and continuously improve quality (Kemafor, Sheth, Cardoso, Miller, & Kochut, 2003). Process visibility refers to the ability to see and understand all aspects of a process at any point in time (Klotz, Horman, Bi, & Bechtel, 2008). However, the complex nature and the high degree of variability reduce process visibility (Clark & Pasupathy, 2014). Although complete visibility is unrealistic, it can be improved using a variety of tools and techniques. Process mining can aid in the reduction of this gap.

Process mining involves the extraction of data from “event logs” stored in various hospital information systems (Mans, Schonenberg, Song, & van der Aalst, 2008). Following extraction, these events are linked, clustered, and examined for flow patterns. Identifying both patterns and flow can assist in the verification of conventional process maps and improve visibility (W. M. P. van der Aalst, Weijters, & Maruster, 2010). However, process mining has challenges related to mining hidden tasks, mining duplicate tasks, mining loops, dealing with noise, etc. A thorough understanding of the existing information system architecture and work processes is essential. Failure to properly understand and manage system complexities can potentially lead to poor data quality (Watts, Shankaranarayanan, & Even, 2009). Poor quality of data can be harmful to system usability and operational performance, resulting in poor decision making. The strength mining results and the decisions are dependent on the quality of the available data. It has been estimated that approximately five percent of an organization's data is of poor quality. Researchers have identified numerous data quality issues in health care, especially problems with accuracy, completeness, and timeliness (Gray, Orr, & Majeed, 2003; Peabody, Luck, Jain, Betenthal, & Glassman, 2004; Thiru, Hassey, & Sullivan, 2003).

Radio frequency identification (RFID) is an advancement in technology and can collect comprehensive data on events, by tracking patients, providers and equipment and improving efficiency and outcomes (Chien, Yang, Wu, & Lee, 2009; Hsieh et al., 2010; C. Huang et al., 2007; H.-H. Huang & Ku, 2009; Leu & Huang, 2009; Raths, 2008; Stahl, Holt, & Gagliano, 2009; Sun, Wang, & Wu, 2008; Tzeng, Chen, & Pai, 2007; Wicks, Visich, & Li, 2006). Process mapping has been used to improve processes and increase efficiency, for instance to reduce overcrowding (Parks, Klein, Frankel, & Friese, 2008) and streamline workflow in ancillary departments (Nagula, Lander, Rivero, Gomez, & Srihari, 2006). Process mapping provides qualitative information and RFID provides a quantitative perspective (Clark & Pasupathy, 2014). The purpose of this paper is to develop a framework and method to integrate process mining with process mapping and RFID to improve process visibility. This framework supports continual identification and correction of process visibility gaps. The reduction of these gaps should assist in the understanding of complex interactions in health care, variability in the process, and support continuous process improvement efforts.

The next section introduces process mining, and various techniques along with its application in healthcare are discussed. Then, the benefits and drawbacks of process mining are identified and those of process mapping and RFID are summarized. Next, the integration framework and method are discussed and finally, future research directions are identified.

Key Terms in this Chapter

Discovery: Discovery is a class of process mining techniques, where there is no a priori model, and the process model is built solely based on event log data.

Radio Frequency Identification (RFID): RFID is a tool used to track the flow of people and materials through a system.

Process Visibility: Process visibility refers to the ability to see and understand all aspects of a process at any point in time.

Extension: Extension is a class of process mining techniques, where there is an a priori model, this model is compared against the event log, and the process model is enhanced.

Process Mining: Process mining is a technique employed to gain an understanding of a process by extracting data from event-logs to develop a process model.

Mental Model: A mental model is defined as a representation of a system or environment based on experience.

Conformance Analysis: Conformance analysis is a class of process mining techniques, where there is an a priori model, this model is compared against the event log, and any discrepancies that arise are studied.

Complete Chapter List

Search this Book: