Work in process: Unearthing Meaning using Process Mining

Process mining
Picture of the cover of Jelmer Koorn's PhD dissertation. Image:, pch.vector

“Work in process” is the witty title of Jelmer Koorn's PhD dissertation. He recently obtained his PhD as a computer scientist at Utrecht University for his research into Process Mining: discovering, analyzing and improving work processes. He did this in two care organizations, a hospital and an institution for people with intellectual disabilities. Process mining can help all kinds of organizations to discover how they actually work now and whether things can or should be done differently in the future. Jelmer Koorn's research is strongly linked to the Future of Work research of Utrecht University. 

Jelmer and his colleagues developed new techniques to discover and analyze work processes from a causal perspective. They investigated cause and effect relationships within work processes, for example aggressive behaviour of residents and the actions of care providers. And the actions of doctors in a hospital in patients with sepsis, blood poisoning. In addition, Koorn has established guidelines for his dissertation to support qualitative evaluations of process mining projects in which collaboration between academics and business experts takes place.

How do you best involve business experts in this very technical “process mining”?

One example is presenting your findings to the people you interviewed. Can they confirm that you have correctly written down and interpreted what they have said? Is your analysis correct and authentic? That seems like a simple guideline, but it is something that can be forgotten by researchers who are mainly technically trained. Resubmit your data to the people you have studied. They know their own work processes and the working conditions best. A second guideline is that you should have an open and honest conversation with experts from the field (for example healthcare) in which you evaluate your research results together. Carefully pick a few of these experts and discuss with them whether you have collected good quality data. Have you determined the work process at the right time and in a truthful manner, so that you can really say something valuable about it and conduct an analysis?

Ask interviewees to check the correctness and authenticity of a summarized report of the interview results and interpretation.

Why did you choose to study work processes in healthcare?

I have always been interested in the societal application of technical topics, I love to know how we can support doctors, patients or residents with the help of computer science. What works and what doesn't work? How do errors sneak into a process and what do we learn from them? How do you involve practitioners as a scientist? Moreover, I find healthcare very interesting because the stakes are high: the interests of many different people have to be weighed up against each other. It's not about being the fastest, or the cheapest, it's about the quality of what you do. And yes, at the same time you have to pay attention to the costs and time, but always with the concerns of patients in mind. A very complicated and fascinating process.

Part of your research focused on aggression in a care institution for people with mental disabilities. Another part was on dealing with sepsis in a hospital. Those situations seem very different. What was the overlap?

They were both focused on action and effect. A caregiver does something that affects the patient. For example: there is aggression in the resident group, the caregiver intervenes, what is the effect of his or her action? Is the aggression less? We investigated this ourselves through reports that were already available at Lunet Zorg in Eindhoven. The research into sepsis involved studying an already existing dataset of a researcher from the TUe in Eindhoven, who had collected data in a hospital. There we looked, for example, if and when the patients with sepsis had to return to the hospital. It is an expensive treatment and it is also not pleasant for the patient when he or she is back to square one. Is it better to keep the patient in the hospital longer? Are there situations where the patient should not have been discharged? Doctors look at the white blood cells and run tests on them. Is that the best way to see what to do?

Cause and effect patterns, that's what my research was focused on. The effect of certain actions in the workplace: can and should it be done differently or better?

After your analysis, you naturally hope that your result will “land” in an organization. How do you do that?

That is the part that motivates me the most. I think it is very good to involve experts from the domain your studying early on and to show a real interest. They will probably comment on your recommendations: “this cannot be done in our organisation” or “we've already tried that”. These comments should be included in your research. Look at what is going on in the organisation, check whether topics are sensitive and why. I can give an example, not from my own research: suppose you advise: “It would be good to let residents who are aggressive live elsewhere.” Perhaps the organization thinks that this is not possible because they want to provide all residents with the same care. That may be a reason why your advice will not be accepted.

In any case, I recommend that you always include context in your analysis. What were the circumstances in which you conducted research? If you don't include context, there is a looming danger of "concept drift". What used to be the norm might no longer the case, so your research is then not tenable. What we labeled as “aggressive” 5 years ago, we may now see differently. The rules and habits in a workplace can change quickly.

Congratulations on your PhD, will you continue your career in scientific research?

I don't know exactly what the future will bring. I strongly suspect that I will work outside of academia. I've always loved looking beyond the confines of my field, beyond computer science, into social sciences, psychology. It was very nice to hear from people who were present at my PhD defense, who came from other fields of work, that they were able to follow the defense well. I take that as a compliment.

Want to know more?

Omslag van het proefschrift van Jelmer Koorn

There is an open-access digital version of Jelmer Koorn's complete dissertation. He has also recorded a laymen's talk summarising the dissertation, that can be provided upon request.

The PhD dissertation of Jelmer Koorn was (partly) funded with the financial support of NWO in the TACTICS project, Techniques for the Analysis of Client-Team InteraCtionS.