Process Mining

From Clinfowiki
Jump to: navigation, search

Process mining is similar to data mining, however the target knowledge which is mined is a process definition. Process mining is based on a convergence of process modeling and data mining. More and more commonly, electronic health records gather and store information about business processes in event logs. Analysis of these event logs can detect bottlenecks in workflow and detect issues with conformance.(1)

Overview

The EHR contains many detailed data points on a patient’s care. All of this information is entered by (directly or indirectly through an interface), verified by, and accessed by providers and staff. (2) This information is necessary for accountability on a patient specific level. The aggregation of this accountability and access data across many patient records can comprise the event logs which are process mined.

Types

There are three types of process mining, based on how the information is used. (3)

Discovery

Process mining is used to design a process model based on events. This type of process mining can also be used in other perspectives such as identifying relationships as in Gray et al (2)

Conformance

There is an existing process model, and the model constructed from the process mining technique is compared to the existing model. This comparison will detect deviations from previously defined processes.

Extension

The process model is extended with additional information or aspects, such as performance data.

Characteristics

Van der Aalst et al. in the Process Mining Manifesto (3) states three characteristics of process mining.

Three types

Process mining is not limited to control-flow discovery. Discovery is only one of three types of process mining, the other two being conformance checking and enhancement. Furthermore, organizational and time aspects are revealed along with control-flow.

Comparison to data mining

Process mining is not just a specific type of data mining. Van der Aalst indicates process mining as the “missing link” between data mining and business practice management.

Analysis

Process mining is not limited to offline analysis. In fact, analysis can be processed on “live” cases to develop a prediction model for the future course of the process.

Role in healthcare

Healthcare processes involve many disciplines and departments, are not necessarily structured, and tend to vary. Process mining can help elucidate what is really going on in a workflow. The many providers and actors in the patient centric workflow may not necessarily be aware of all of the direct and indirect processes involved with patient care. The results of process mining can be used to educate members of the health care team regarding the roles of other players in the process.

Obstacles to use in the electronic health record

In order to effectively use process mining, the EHR. needs to be somewhat “process-aware.” The event logs that are mined and analyzed need to be present in the EHR. Transaction data is the main source of the event documentation. Accurate analysis of this data depends on the user entering or updating the information at the time which it occurred. For example, if the vital signs on a patient are taken and written on a slip of paper, then entered at a more convenient time, the event logs will not contain accurate information about that process. If this sort of workaround is employed on a regular basis, the data will be skewed. In the future, smart devices that interface directly with the EHR should be able to assist in keeping event data more temporally accurate.

Examples of process mining

  • Huser V, Starren JB, EHR Data Pre-processing Facilitating Process Mining: an Application to Chronic Kidney Disease. AMIA Annu Symp Proc 2009 link
  • Process mining techniques: an application to stroke care.

Ronny Mans, Helen Schonenberg, Giorgio Leonardi, Silvia Panzarasa, Anna Cavallini, Silvana Quaglini, in Studies In Health Technology And Informatics (2008)

  • Data mining techniques for analyzing stroke care processes.

by Silvia Panzarasa, Silvana Quaglini, Lucia Sacchi, Anna Cavallini, Giuseppe Micieli, Mario Stefanelli

Software

Software which can support process mining is ProM (4) from Einthoven University. ProM is an open source framework for using process mining tools. ProM can receive input in XES or MXML formatted logs. It is available for Windows, Mac OS, and Linux operating systems. ProM can be downloaded from http://www.promtools.org/prom6/.

References

  1. Process Mining: Discovery, Conformance and Enhancement of Business Processes by W.M.P. van der Aalst, Springer Verlag, 2011 (ISBN 978-3-642-19344-6). http://www.processmining.org/book/start
  2. Gray JE, Feldman H, Reti S, Markson L, Lu X, Davis RB, Safran CA., Using Digital Crumbs from an Electronic Health Record to identify, study and improve health care teams. AMIA Annu Symp Proc. 2011;2011:491-500. Epub 2011 Oct 22. http://www.ncbi.nlm.nih.gov/pubmed?term=Using%20Digital%20Crumbs%20from%20an%20Electronic%20Health%20Record%20to%20identify%2C%20study%20and%20improve%20health%20care%20teams
  3. Process Mining Manifesto http://www.win.tue.nl/ieeetfpm/lib/exe/fetch.php?media=shared:process_mining_manifesto-small.pdf
  4. ProM software http://www.processmining.org/prom/start
  1. Using process mining techniques to analyze and visualize data: a case study of a large clinical research data repository link


Submitted by Diane Petersen, RPh