You are reading Flux Capacitor, the company weblog of Fluxicon. Here, we write about process intelligence, development, design, and everything that scratches our itch. Hope you like it!
A 60-day professional ticket for Nitro, so that you can get started on your own data right away.
The lucky winner of the process mining book is Jianmin, a researcher from China!
However, as an academic, Jianmin can already have free access to Nitro through our academic program, so the Nitro ticket part of our prize was not really useful. Therefore we decided to draw a second winner for the 60-day Nitro ticket, and Karl was the lucky one.
Congratulations to Jianmin and Karl! All participants will receive a summary of the results in the coming days. Thanks a lot to everyone who participated!
In a previous post, I had written about the challenges of applying process mining in the healthcare domain. And I had promised to follow-up with a summary of existing case studies in this area. Here is the first one.
often based on independently developed IT systems.
As a result, there is little insight into what happens in a healthcare process for a group of patients with the same diagnosis. But process insight is needed to deliver high quality care while at the same time reducing costs.
Process mining can be used to extract process knowledge from event logs. The goal of the study was to obtain meaningful information about so-called careflows (typical paths followed by particular groups of patients).
The subject of study was a gynecological oncology process. The raw data contained information about a representative group of 627 gynecological oncology patients treated in 2005 and 2006, for which all diagnostic and treatment activities had been recorded in a billing system for financial purposes. Different departments are involved in this process, including gynecology, radiology, and several labs.
Challenges
One of the challenges the researchers faced was that for each activity it was only known on which day the service had been delivered. So, there was no information about the time of the start and completion of activities, and therefore events which happened on the same day could not be ordered properly.
Furthermore, the log contained 376 different event names for activities. This shows that they were dealing with a non-trivial careflow process. This high number of different, low-level activities was reduced by a pre-processing step in which:
The log was filtered by removing all the detailed lab activities and only keeping the initial lab event, where the samples were offered to the lab, as a representative.
The log was aggregated by summarizing events that happened within one department, so for example ‘echo abdomen’, ‘thorax’ and ‘CT brain’ were just represented as ‘radiology’.
The result was an event log with less than 60 different activities across multiple departments, which was then used for the process mining study. No previous knowledge about the care process was used. This means that no existing process model was used to guide the discovery, and no workshops or interviews were performed. Just the preprocessed data from the billing system of the hospital were used for the analysis.
Process mining results
First, a process model was discovered using the Heuristic miner, because it is able to focus on the main process flow. Because of the complexity of healthcare processes it would be difficult to show every detail of the behavior appearing in the process log.
The figure below shows the discovered process model, which is still fairly complex.
To reduce the complexity, trace clustering techniques were used. Trace clustering breaks up the log of all 627 patients into several, more homogeneous sub groups. So, those patient flows that followed a similar path were grouped together.
The picture below shows a visualization of the used SOM (Self Organizing Map) clustering algorithm. The nine cells represent the nine clusters obtained from the log. Each dot represents one instance (one patient), and all instances in the same cell belong to the same cluster. The figure also shows a contour map based on the number of instances in each cell: Clusters with many similarities are visualized as high land, and there are clusters with exceptional cases (sea).
The process model below shows the process flow for all patients in the biggest cluster (with 352 cases). The result is much simpler than the model above. A closer inspection of this main cluster by domain experts confirmed that this is indeed the main stream followed by most gynecological oncology patients.
Also the organizational perspective was explored to gain insight into how the departments interact with each other in this process. Below, a social network is shown, in which each bubble represents one department. The more activity that took place in the department for the 627 patients, the larger is the bubble.
An arc indicates that patients have frequently moved from one department to the other in subsequent activities. Only the most frequent transfers between the departments shown to highlight the most dominant interactions.
The picture reveals that the general clinical chemical lab is highly involved in the process and interacts with many departments.
When the results were presented to the people in the hospital, they were surprised about the strong collaboration with the dietics department. In fact, patients who undergo several chemotherapy sessions often need to visit the dietician. However, this was not immediately clear to everyone and illustrates the value of creating transparency using process mining.
Lessons learned
The unstructured and complex nature of healthcare processes poses a challenge for process mining techniques. Additional methods such as filtering or clustering need to be used in a pre-processing step.
The study shows that it is possible to derive understandable models for large groups of patients. This was also confirmed by people of the hospital.
In the hospital, a manually created flowchart for the diagnostic trajectory of the gynecological oncology healthcare process was available. The process mining results results were comparable to this flowchart, but despite the preprocessing a lot less effort was needed to obtain them.
Some of you are currently analyzing the BPI challenge log, which is also a healthcare process. Have you seen similarities, and were you able to apply similar methods? Let us know in the comments.
The conference program was amazing, with straight-talking presenters from outstanding companies like Dropbox or Groupon, and lots of actionable, hands-on advice from lean startup practitioners. Here’s the keynote by Eric Ries, who gives a great outline of what the Lean Startup movement is all about:
One thing I love about this conference1 is that they livestream the complete conference program, and that they provide recordings of every talk online, for free2. So if you missed last night’s session, you may want to watch the recordings here.
We’d like to thank Eric Ries and his fellow organizers for putting all this together, and also all our guests for coming and sharing their mind! Special thanks also go to Mathias from UXSuite, who helped us setting up this event!
Thanks for all the fish, and see you all next year!
Besides that they always feature an amazing and relevant line-up of speakers, that is. ↩
Livestreaming, or at least providing recordings of talks online, is something that I wished a lot of conferences did for years. Especially academic conferences, which are essentially sponsored by the public via taxes, have absolutely no reason for not doing so. ↩
Imagine the situation of a manufacturing company which needs to handle the repair and exchange of faulty products with its customers. To save costs, the process should be as efficient as possible. But as a customer-facing process, the speed and convenience for the consumer is also important.
If the customer service process is handled poorly, the consumers might badmouth the manufacturer in the internet and among friends, which can lead to significant brand damage and a loss in future sales. But if repairs and exchanges are handled very well, the consumers might be delighted and actually increase their brand loyalty.
Customer service example
In the picture below you see an anonymized, simplified example of the planned process for the shipment of replacement products (left) compared with the actual process (right). While the goal is to complete each case within 3 days, in a fairly sequential process, it actually takes 6 days on average (and in several instances much longer than that) and the real process is much more complicated.
But the main problem of the company is not the inefficient or complicated process.
The actual problem
The real problem is that the actual process is not visible to the service manager in the company in the first place. It is impossible to improve when there is no insight into what is actually happening.
Process mining
Because you can’t improve what you can’t measure, the biggest benefit of process mining is that it can make the real processes visible based on existing log data in the IT systems. Only when you can see what is happening, you can get to the root causes of problems and take action.
Conformance and Performance
The gained transparency can be used for both improving the conformance and the performance of the process.
Conformance. Deviations from the intended process can reveal hidden activities or process flows (see below). Deviations do not need to be a problem. But if these deviations are not visible in the first place, then it is impossible to detect illegal workarounds that are a serious problem for an organization.
Performance. The same holds for inefficiencies. For example, if the overall product replacement process takes too long and the customers are unhappy, then one needs to be able to see where exactly time is lost, where more resources should be assigned, and so on.
Process mining can help to make that first step: Creating transparency about what is actually happening.
Where do you wish that processes would be more transparent? And how would it help you? Let us know in the comments.
For those of you who want to become creative in the first BPI Challenge, it will be good news that the deadline has been extended to 20th of June. So, there is still more than one month left to do some wild process mining.
Participants of the challenge are provided with a real-life event log and asked to analyze the data using whatever techniques available. Their findings can be documented in one of two ways:
The participants can focus on a specific aspect of interest and analyze this aspect in great detail. Here, one can choose for example to focus on specific models, such as control-flow models, social network models, performance models, predictive models, etc.
The participants may report on a broader range of aspects, where each aspect does not have to be developed in full detail. The report submitted in this category will be judged on its completeness of analysis and usefulness for the purpose of a real-life business improvement setting.
We think that the BPI Challenge is a great initiative and, being among the jury members, we are already very curious about the results!
About ten years ago, when Anne and myself were still studying for our Master’s degree at the HPI Potsdam in Germany, we first heard about process mining in a BPM seminar given by Mathias Weske. We became so fascinated with this new technology, and with the BPM domain itself that, by the end of 2004, we moved to Eindhoven to start working on a PhD in process mining with Wil van der Aalst and his outstanding research group.
With this background, saying that we owe a lot to the academic community is rather an understatement. Also, as former academics we are of course still very passionate about research and education, and we wanted to give something back to the academic community.
Free tickets for our tool Nitro for all researchers and students.
Nitro fills the gap between real-life logs and ProM. It helps researchers to focus on their case study, or a new process mining algorithm. In an education setting, it allows students to experience the whole scope of a process mining project, starting from raw CSV data.
Slides and instructions for a process mining hands-on tutorial based on Nitro and ProM.
We have given this tutorial multiple times, and it has been used successfully at several universities already. It is a great starting point for including process mining in BPM courses or seminars.
We hope to grow the set of teaching materials over time with the help of our academic partners. We are convinced that it is essential to support education in process mining, a technology that we believe will shape the future of BPM.
Academic Partners
We are excited and immensely grateful that 20 universities are joining us as launching partners of our Academic Initiative, including some of the most excellent institutes for BPM and process mining research and education around the globe:
But we don’t want you to think of this as an elitist circle — if you can’t find your university on this list, please get in touch with us and let’s make it happen!
Good process improvement can achieve both an increase of quality and lower cost at the same time. Efficient healthcare processes are very relevant, because patient treatments pose a significant burden on our aging societies.
But could hospitals be run more efficiently? Or would this necessarily mean a decrease in quality? A few weeks ago, I attended a Lean Six Sigma seminar in which a very interesting case study was presented: A process improvement team in the general hospital Reinier de Graaf groep had reduced the time patients had to stay in the hospital for a hip operation from six to only three days while improving patient satisfaction at the same time.
This example suggests that there is much more room for improvement than one might think.
Process Mining
Process mining is a revolutionary new technology for process improvement. Process mining does not start at the whiteboard to make an existing business process visible. Instead, it leverages existing log data that are collected by the many IT systems that are supporting enterprises around the world: ERP systems like SAP, but also legacy systems, CRMs, and so on, record very detailed information about the activities that have been performed, when, and by whom.
For any process improvement, determining the current ‘As-is’ process is the first necessary step. Using process mining, one can automatically and accurately visualize the actual process flows based on objective data. This transparency allows organizations to continuously monitor and improve their processes in ways that were not possible before.
But is process mining also applicable to processes in healthcare?
Challenges
Healthcare processes are either diagnosis / treatment processes or of organizational nature (such as the scheduling of appointments). The biggest challenges for applying process mining to healthcare processes are their complexity, their multi-disciplinarity, that they are changing often, and the log data from the IT systems.
1. Heterogeneity
A lot of the complexity of healthcare processes comes from the heterogeneity of the patients that are treated. After all, every one of us runs through the same process when we apply for a new passport, but we are unique and complex human beings when it comes to medical conditions.
As a consequence, the individual treatment processes of different patients even with the same illness are often unique. For process mining techniques—which generalize the common process based on the individual process executions that happened in practice—this lack of similarity can be quite a challenge.
2. Multi-disciplinarity
Hospitals departments are highly specialized in their respective fields but need to work together across their disciplines. For example the doctor of a cancer patient might send the patient to the radiology department, which needs to return the results to the oncology department.
While the multi-disciplinary nature of healthcare processes adds to the complexity, it is also an indicator for improvement opportunities. Generally, process inefficiencies often emerge at the boundaries of different functional units because people oversee their part of the job quite well but lack insight into what happens before and after and why.
3. Changing fast
The medical knowledge evolves continuously. As a consequence, the corresponding medical procedures and processes change as well.
Again, this adds to the difficulty of process improvement initiatives (Try to hit a moving target!). But at the same time it increases the attractiveness of process mining because it allows to make the current process visible automatically, just based on the history logs in the IT systems.
4. Data collection
The log data is the basis for process mining. So, of course the availability and quality of data is key to be able to apply process mining techniques.
On the plus side, detailed records are kept in healthcare processes for billing purposes. And new developments such as the Electronic Patient Record will increase the availability and quality of the data.
At the same time, there are still many manual activities that are not observable. Furthermore, data entries are often made manually after an activity has actually occurred. Over time, hospital information systems will evolve and with increasing integration and automation, the data availability and quality will improve.
More to come
To give you an update on the current state of the art in process mining research in healthcare, I plan to write up a few case studies that have been performed in this area in future blog posts.
For now, join the discussion and let me know what you think about process mining in healthcare: Do you feel there is room for improvement in our hospitals? And do you think that mining (anonymized) patient history data for process improvement purposes is legitimate?
The goal for this event is to give practitioners and students of the lean startup methodology the opportunity to hear insights from leaders in embracing and deploying the core principles of the lean startup methodology. The day-long event will feature a mix of panels and talks focused on the key challenges and issues that technical and market-facing people at startups need to understand in order to succeed in building successful lean startups.
We’ll all be getting together on 23 May to watch the live stream of the Startup Lesson Learned conference at the Fluxicon HQ in Eindhoven starting at 18:00. Please register at this website so we know how many people to expect1.
This simulcast is brought to you by Fluxicon and our friends from UXSuite. Our location is a private and cozy place, so there won’t be professional catering. We will provide some basic caffeination, but please feel free to bring some snacks, drinks, and whatever you require to make it through a night of non-stop lean startup action2!
Note: your name, company and email address will be provided to the conference organizers (Eric Ries). We won’t send you any spam or share your information with others. ↩
And, yes: we plan to keep this going at least until the end of the conference program (around 3:00). ↩
Yes, of course it’s free. But you better hurry, this place fills up mighty fast… ↩
Together with Prof. Wil van der Aalst, Fluxicon is involved in a Master’s project, in which existing, commercial process mining tools are evaluated for different use case scenarios. It is important to consider the context of use for a process mining tool because, for example, an auditor has quite different requirements than a typical process analyst.
Our Master student Irina has compiled a first list of use cases in a survey here. She did a great job in defining a short but comprehensive list of process mining functionalities. But you can help her to rank them in importance and identify missing use cases.
Please forward and share this invitation with anyone who could contribute to the survey. Thank you!
Everyone participating in the survey will also receive the results upon request. ↩
In case you are wondering: No, Nitro is not part of the evaluation study because we only consider process mining tools that can discover process models. ↩
We had the privilege of reading drafts of this book, and it is really hard not to recommend it for everyone interested in process mining. Wil is one of the fewer academics writing in an accessible and down-to-earth manner, without skimping on clarity or scientific rigor, and without hyperbole. The book covers the fundamentals and basics of process mining, and gives a comprehensive overview about the state of the art of the field.
Wil was kind enough to answer five questions about his new book for us. He explained why BI is not really intelligent, who this book is for, and why you should read it.
Interview with Wil van der Aalst
Anne: This is the first book on process mining. I know that both academics and professionals have been waiting for a book on process mining. For whom did you write this book?
Wil: The initial goal was to write a shorter less technical book primarily focusing on professionals. However, while writing it became clear that the topic cannot be introduced without giving concrete definitions and examples. Therefore, the book does not shy away from technical details. As Einstein said: “Everything should be made as simple as possible, but no simpler”. As a result the book is interesting for both academics and professionals.
Anne: You make the point that most Business Intelligence systems are rather un-intelligent. What do you mean by that?
Wil: The problem of new technologies and tools in the field of Business Process Management (BPM) and Business Intelligence (BI) is that they are presented as silver bullets able to solve notoriously difficult problems with little effort. In reality such technologies seldom live up to their expectations as there is no such thing as a free lunch.
BI tools tend to be data-centric while providing only reporting and dashboard functionality. They can be used to monitor and analyze basic performance indicators (flow time, costs, utilization). However, they do not allow users to look into the end-to-end process. Moreover, despite the “I” in BI, most of the mainstream BI tools do not provide any intelligent analysis functionality.
Anne: You distinguish between ‘Lasagna’ processes, which are more structured, and ‘Spaghetti’ processes, which are unstructured. Where do you find them and how is process mining different for these two types of processes?
Wil: Lasagna processes are relatively structured and the cases flowing through such processes are handled in a controlled manner. Therefore, it is possible to apply all of the process mining techniques presented in the book (also more advanced techniques such as prediction and short-term simulation). Spaghetti processes are the counterpart of Lasagna processes. Because Spaghetti processes are less structured, only a subset of the process mining techniques described in the book are applicable. However, the potential process improvements may be much more substantial.
Spaghetti processes are typically encountered in product development, service, resource management, and sales/CRM. Lasagna processes are typically encountered in production, finance/accounting, procurement, logistics. The structuredness of processes also varies from industry to industry, e.g., processes in healthcare tend to have more variability than processes in manufacturing.
Anne: Which aspect of process mining deserves more space than it gets in your book and why?
Wil: The initial goal was to write a book of 200 pages. In the end the book was more than 350 pages. As a result, the book is comprehensive and self-contained. Although the book shows various examples of process mining results based on numerous real-life event logs, it would have been good to present a few case studies in more detail. Moreover, the relationship to Visual Analytics could have been discussed in more detail.
Anne: If someone is completely new to process mining, what would you hope is the biggest take-away point for that person?
Wil: Event data is omnipresent, thus enabling evidence-based BPM. Process mining combines techniques from data mining and process modeling and analysis. As a result, it is possible to analyze and improve business processes based on facts rather than fictive PowerPoint diagrams.
The threshold to start a process mining project is low. Therefore, it is best to experience the “magic” of process mining using data from your own organization. The book shows how this can be done and provides pointers to the software needed to start discovering and improving processes based on facts rather than fiction.
There are no comments for this article yet. Add yours!
