Recommendations for enhancing the usability and understandability of process mining in healthcare
Introduction
The healthcare sector is confronted with severe challenges, most importantly the contention between tightening budgets and increased care needs due to the aging population [1], [2]. To face these challenges while achieving high quality of care standards, healthcare organizations such as hospitals are becoming increasingly aware of the need to improve their processes (i.e. interrelated sets of activities, decisions and events with a particular goal [3]). Processes play a central role in a healthcare organization's daily operations [1], [4]. They can be subdivided in two categories: clinical processes (e.g. the emergency care process or the treatment process of a particular condition), and administrative/organizational processes (e.g. the inventory management process of materials or the billing process) [5], [6].
Healthcare processes in general, and clinical processes in particular, possess some distinct characteristics compared to common business processes such as the order-to-cash process [6], [7]. Healthcare processes can be characterized as loosely framed and knowledge-intensive [8], [9]. A loosely framed process can be performed in a large, but finite and predefined, number of distinct ways [9]. This relates to the observation that healthcare processes typically exhibit high levels of variation [6]. The knowledge-intensive character implies that the execution of healthcare processes heavily depends on knowledge workers, such as physicians, and the knowledge-intensive decisions they make [8]. These complex decisions are made based on a wide range of criteria, including medical knowledge, patient-related characteristics and the experience of healthcare professionals [5], [6], [8], [10], [11]. Healthcare processes are typically also closely intertwined with each other and are multi-disciplinary, requiring cooperation between clinicians with different expertises, which adds to their complexity [5], [6], [8]. Besides their knowledge-intensive, loosely framed and multi-disciplinary character, healthcare processes are also highly dynamic as they typically continuously change over time due to advances in medical knowledge, technology or administrative procedures [5], [6], [10].
To identify opportunities for process improvement in such healthcare processes, a healthcare organization first needs to gain a profound understanding of the process under consideration. To gather insights in how the process is executed, staff members who are familiar with the process can be brought together for a discussion. This discussion can target the development of process models capturing process insights (such as the order of activities in a clinical process), which forms a basis for a process analysis. However, this is very time- and effort-intensive, and the created process model tends to present an idealized view on the process which might have little connection to reality [12].
To uncover the real behavior of an executed process, a solution can originate from data already collected by health information systems such as a hospital information system. Using the data embedded in the databases of these health information systems, an event log can be generated which contains detailed process execution data for a healthcare process of interest. In this way, the event log contains real-life data about which activities were performed, when they were performed, who performed them and for whom (e.g. for which patient) [12], [13]. Process mining is the research field concerned with the development of techniques to retrieve non-trivial information from such an event log [7], [12]. Over the past decade, the process mining community has developed an extensive set of techniques which convey profound process insights based on real-life data [4], [12]. These techniques relate to the discovery of process models from data, the detection of deviations between an existing model and reality, or the enhancement of an existing process model with, e.g., process performance information [12]. Process mining outcomes can be leveraged to instigate evidence-based process improvement initiatives in healthcare.
The systematic use of process mining in healthcare would be consistent with the Learning Healthcare System concept [14] introduced by the Institute of Medicine. One of the basic pillars of a Learning Healthcare System is the implementation of data reuse mechanisms, which allow for learning from data generated during the execution of processes [15], [16]. However, despite the potential of process mining, its uptake in healthcare organizations outside case studies in a research context is rather limited. This observation was the starting point for a two-day international brainstorm seminar.1 The seminar brought together 18 experts from 11 different countries, both researchers and healthcare practitioners, to reflect upon how to enhance the usability and understandability of process mining in healthcare. This position paper synthesizes the conclusions of the brainstorm seminar. It specifies ten key recommendations that process mining researchers and the community are encouraged to carefully consider when developing a new research agenda for process mining in healthcare. While the focus of our work is predominantly on recommendations for researchers and the research community, three additional recommendations are formulated which explicitly target healthcare organizations and health information systems vendors.
The remainder of this paper is structured as follows. Section 2 provides a a primer to process mining in healthcare, encompassing both an introduction to the topic and an overview of some applications in a healthcare context. Section 3 outlines the key recommendations for process mining researchers and the research community. Section 4 presents the key recommendations for healthcare organizations and health information systems vendors. Section 5 presents a brief conclusion.
Section snippets
A process mining in healthcare primer
To provide the required background to appreciate the formulated recommendation, this section provides a primer to process mining in healthcare. Section 2.1 introduces the basic concepts of process mining in healthcare. Building on these concepts, Section 2.2 outlines some applications of process mining in healthcare.
Recommendations for process mining researchers and the research community
Despite process mining's great potential to help healthcare organizations understand how their processes are actually executed, its use in healthcare outside a research context is limited. Starting from this observation, a two-day international brainstorm seminar took place to reflect upon how to enhance the usability and understandability of process mining in healthcare. This seminar brought together 18 experts from 11 different countries, both process mining researchers and healthcare
Recommendations for healthcare organizations and health information systems vendors
The predominant focus of this paper is on providing recommendations to process mining researchers and the research community to enhance the usability and understandability of process mining in healthcare. However, healthcare organizations and health information systems vendors also play a key role in a more widespread use of process mining as they should provide an environment which enables a continuous use of process mining. To this end, three key recommendations are provided to shape such an
Conclusions
Despite the unique potential of process mining to retrieve data-driven insights in healthcare processes, its uptake in healthcare organizations is rather limited. Given this observation, this paper synthesized the outcomes of a two-day international brainstorm seminar on how to enhance the usability and understandability of process mining in healthcare. Based on the discussions of 18 experts, both researchers and healthcare practitioners, a set of key recommendations is specified. Ten
Conflict of interest
There are no conflicts of interest to report.
Acknowledgment
This work has been partially supported by EIT Health Activity Nr. 20238.
References (167)
- et al.
Process mining in healthcare: a literature review
J Biomed Inform
(2016) - et al.
IT support for healthcare processes-premises, challenges, perspectives
Data Knowl Eng
(2007) - et al.
Business process analysis in healthcare environments: a methodology based on process mining
Inf Syst
(2012) - et al.
On local anomaly detection and analysis for clinical pathways
Artif Intell Med
(2015) - et al.
Towards a decision-aware declarative process modeling language for knowledge-intensive processes
Expert Syst Appl
(2017) - et al.
A two-step approach for mining patient treatment pathways in administrative healthcare databases
Artif Intell Med
(2018) - et al.
The learning healthcare system: where are we now? A systematic review
J Biomed Inform
(2016) - et al.
Healthcare pathway discovery and probabilistic machine learning
Int J Med Inf
(2020) - et al.
A multi-dimensional quality assessment of state-of-the-art process discovery algorithms using real-life event logs
Inf Syst
(2012) - et al.
On mining clinical pathway patterns from medical behaviors
Artif Intell Med
(2012)
Mining post-surgical care processes in breast cancer patients
Artif Intell Med
Towards comprehensive support for organizational mining
Decis Support Syst
Discovering work prioritisation patterns from event logs
Decis Support Syst
Processprofiler3d: a visualisation framework for log-based process performance comparison
Decis Support Syst
Computer-interpretable clinical guidelines: a methodological review
J Biomed Inform
Comprehensive analysis of rule formalisms to represent clinical guidelines: selection criteria and case study on antibiotic clinical guidelines
Artif Intell Med
Assessment of hospital processes using a process mining technique: outpatient process analysis at a tertiary hospital
Int J Med Inf
Improving structural medical process comparison by exploiting domain knowledge and mined information
Artif Intell Med
Event log imperfection patterns for process mining: towards a systematic approach to cleaning event logs
Inf Syst
Bridging abstraction layers in process mining
Inf Syst
Lean in healthcare: a comprehensive review
Health Policy
Temporal similarity measures for querying clinical workflows
Artif Intell Med
Simulation of emergency department operations: a comprehensive review of KPIs and operational improvements
Comput Ind Eng
Automated discovery of business process simulation models from event logs
Decis Support Syst
Embedding conformance checking in a process intelligence system in hospital environments
Lect Notes Comput Sci
Process mining in healthcare: evaluating and exploiting operational healthcare processes
Fundamentals of business process management
Knowledge-intensive processes: characteristics, requirements and analysis of contemporary approaches
J Data Semant
Healthcare process support: achievements, challenges, current research
Int J Knowl-Based Organ
Process mining: data science in action
Using electronic health record audit logs to study clinical activity: a systematic review of aims, measures, and methods
J Am Med Inform Assoc
The learning healthcare system
Achieving a nationwide learning health system
Sci Transl Med
Process-aware information system development for the healthcare domain-consistency, reliability, and effectiveness
Lect Notes Bus Inf Process
Clinical event
Getting a grasp on clinical pathway data: an approach based on process mining
Lect Notes Artif Intell
Automated discovery of process models from event logs: review and benchmark
IEEE Trans Knowl Data Eng
Discovering social networks from event logs
Comput Support Cooper Work
Discovering social networks instantly: moving process mining computations to the database and data entry time
Lect Notes Bus Inf Process
Mining resource profiles from event logs
ACM Trans Manag Inf Syst
Business models enhancement through discovery of roles
Proceedings of the 2013 IEEE symposium on computational intelligence and data mining. IEEE
Discovering health-care processes using DeciClareMiner
Health Syst
Conformance checking of executed clinical guidelines in presence of basic medical knowledge
Lect Notes Bus Inf Process
Control-flow analysis of procedural skills competencies in medical training through process mining
Postgrad Med J
Conformance checking
Conformance checking: a state-of-the-art literature review
Proceedings of the 11th international conference on subject-oriented business process management
Replaying history on process models for conformance checking and performance analysis
Wiley Interdiscip Rev: Data Min Knowl Discov
Discovering stochastic Petri nets with arbitrary delay distributions from event logs
Lect Notes Bus Inf Process
Decision mining in business processes. Technical Report
Cited by (47)
Discovering interpretable medical process models: A case study in trauma resuscitation
2023, Journal of Biomedical InformaticsProcess mining for healthcare decision analytics with micro-costing estimations
2023, Artificial Intelligence in MedicineCitation Excerpt :In Australia, several studies analysed the behaviour of healthcare processes in pre-hospital, emergency and in-hospital using routinely collected hospital datasets from South Australia and Queensland (e.g., [30–32]). Attempts to apply process mining techniques in healthcare has highlighted several challenges and opportunities due to the unique characteristics of these healthcare processes [1,28,33,34]. Mans et al. [33] address three issues affecting process mining in healthcare, namely data correlation from multiple healthcare systems, typical questions of interest for healthcare stakeholders, and identification of data quality issues.
Mining context-aware resource profiles in the presence of multitasking
2022, Artificial Intelligence in MedicineCitation Excerpt :Additional information about these events can also be logged, e.g. a timestamp indicating when the event occurred or the involved resources (e.g. nurses and physicians) [3,30]. Such an event log constitutes the primary input for further analysis using various Process Mining techniques [3,9]. Interest in research on Process Mining in healthcare has steadily grown in recent years [26,30,31], and a wide variety of use cases have been demonstrated.
CATNet: Cross-event attention-based time-aware network for medical event prediction
2022, Artificial Intelligence in MedicinexPM: Enhancing exogenous data visibility
2022, Artificial Intelligence in MedicineCitation Excerpt :Thus, finding these opportunities is critical [22–24]. While process mining offers a range of techniques to support healthcare experts and processes, the application of techniques is not without challenges [19,25,26]. Challenges faced by process mining practitioners in the healthcare domain are outlined in [26] and focus largely around the process of decision making in healthcare.
Innovative informatics methods for process mining in health care
2022, Journal of Biomedical Informatics