Process Mining with Celonis - Potential in Ophthalmology?

The analysis of large amounts of data to optimize clinical processes is a great challenge, but essential to save costs and to reduce the length of stay and waiting time of patients. However, there is often the problem that the optimization of "patient paths" is based on only a non-representative sample and thus does not reflect reality. Process mining describes a new method of process optimization and has already arrived in industry. To what extent it also has potential in medicine and especially in ophthalmology, we discuss here with Mr. Endres from the company Celonis.


Hi Mr. Endres,

You work for the company Celonis, which specializes in so-called process mining. Celonis is Germany's first decacorn as a start-up, so it was recently valued at around 10 billion euros. What exactly does Celonis do and what is process mining?

Yes, exactly. As an Account Executive at Celonis, I am responsible for acquiring new customers. My focus is on the industry segment "healthcare", which includes hospitals, clinics and care facilities.

This new rating was of course great news and makes us at Celonis very proud. The rating is also a great confirmation from the market and reflects the need for our Execution Management System.

Process mining is an analytical discipline and provides companies with a completely objective, data-driven, real-time picture of how their processes actually run. It involves extracting valuable data from so-called event logs that are readily available in your information systems.

The technology is a key component of the Celonis Execution Management System, which enables organizations to measure and maximize their performance (we call it "execution capacity").

By visualizing processes, you can identify execution gaps - such as bottlenecks, manual rework, manual processes that should be automated, and so on. - and proactively and automatically eliminate them. The result? Smooth processes, higher productivity, happier employees - and maximum execution capacity.

Can you give examples where process mining has improved in Which areas How?

What is special about our solution is that it is used by companies for a wide variety of areas and processes. A majority of our customers start by using Celonis in administrative processes, as the transactional volume is highest there. One of the most impressive examples of the successful use of Celonis is Deutsche Telekom. By using process mining in its own shared service center, Telekom has built a completely data-driven control of its accounting, reporting, purchasing and HR processes. By optimizing the incoming invoice process alone, the company was able to save €40 million annually through discounts taken.

This is just one of many examples of how Celonis delivers significant value to businesses.

Not only administrative processes can be analyzed with Celonis. Here are a few examples:

  • Insurance claims management
  • Patient admission in hospitals
  • Lending at banks
  • Laboratory diagnostics
  • Production
  • and many more

Are the improvements then measurable? What exactly is being improved?

Definitely! The impact of process mining is usually noticeable within the first few weeks. Whether through increased efficiency, cost savings or increased productivity. The return on investment is positive for all customers already in the first year.

For which constellations is process mining particularly suitable?

The basic prerequisite for process mining are the digital footprints in systems. This means that basically all processes that are captured in a system can also be analyzed by Celonis. The more granular the data capture, the better the analyses can be run. It does not matter whether everything takes place in one system or whether several systems are involved in one process. We find the greatest added value in the processes with a high volume. Firstly, this is where inefficiencies most often creep in, and secondly, these processes can rarely be reconstructed in a truly objective manner using conventional approaches (e.g., employee interviews).

Is it also conceivable to apply process mining in the field of medicine?

The short answer is "yes. Even if we are still somewhat in our infancy here, the potential is huge. After all, the ultimate goal of medical care is to deliver the best treatment for the patient as efficiently as possible. Celonis can help, above all, to optimize the processes surrounding treatment. Be it the timely procurement of materials and medications, the admission and discharge processes, as well as the entire process around OR management. For the latter, for example, we have set up a use case together with one of our partners. This involves setting up real-time control of the occupancy of operating rooms and implementing an early warning system for delays. The goal is to optimize the utilization of operating rooms and thus increase hospital revenues.

Conventional consulting firms often spot-check processes, measure average times, and then change process flows. What does Celonis do differently here? Why is process mining better here?

In fact, Celonis was born out of just such a project. At that time, the three founders were commissioned to analyze the IT service management process of the Bavarian Broadcasting Corporation. Instead of interviews and questionnaires, the three used the data from the source system to reconstruct the ACTUAL processes. The success of the idea can be seen today in the 10 billion euro valuation.

Essentially, I see three major advantages to the approach:

Subjective opinions vs. objective data: When recording processes via interviews, the actual picture is often distorted by the subjective perception of the employees. Processes that are particularly painful or lengthy have burned themselves into the memory and come to the fore in the perception. However, it is difficult to assess whether these processes are objectively frequent or actually relevant from the company's perspective. With the data-driven approach, this assessment does not have to be made. Instead, a quantified statement is obtained directly from the data and the impact on the company can be read off very easily. This helps significantly in the prioritization and implementation of measures.

Completeness: It is not uncommon to discover that there are several thousand process variants when analyzing processes with Celonis. The reason for this is often a complex system landscape and different requirements. Celonis reads this out within seconds and thus creates 100% transparency. One could almost speak of a digital process twin. With manual process recording by consultancies, you get maybe 5 to 10 variants of process steps that have been highly aggregated. This is exactly the reason why many of our partners, such as the Big 4, have included Celonis as part of their consulting portfolio and carry out the ACTUAL recording of processes with project licenses.

Continuous monitoring: The idea behind Celonis is not only process analysis, but rather operational control and continuous measurement of processes. For this purpose, we have a quasi permanent connection to the source systems and permanently the current status of the processes. With Celonis, it is therefore possible to determine which weak points exist in the process, take appropriate measures and then monitor the change in the long term. This makes a sustainable improvement of the processes much easier than if the process recording is carried out once without measuring the result afterwards.

In the medical field, there is often the problem that many different systems are used and sometimes even paper files are used that have been scanned. Are there solutions to build a usable "data warehouse" here?

We have this challenge in other industries as well and have gained plenty of experience as a result. The first option is to actually connect to each individual system. Celonis already provides over 80 standard connectors for this to simplify implementation. But connecting new systems is also possible via an open interface. The second possibility is the connection of a data warehouse. This option is also often adopted when too many systems are involved. In practice, the two scenarios are evaluated during implementation and then a decision is made together with the customer as to which is the better solution.

In the medical sector, there is often the problem that, unlike in supply chain management, for example in the automotive industry, medical emergencies cannot be planned and some patients simply require significantly more time than others. How can process mining help here?

Even in other industries, like automotive, there are emergency scenarios that you want to prepare for. I know of one of our customers who uses Celonis to simulate emergencies. They use historical data from, say, the last two years and run through different scenarios using our simulation environment. In this way, the dependencies of individual process steps can be determined and potential bottlenecks in the processes can be identified. I can very well imagine a similar application in emergency medicine.

Is it also conceivable to integrate intelligent systems that, for example, record diagnoses or even create tentative diagnoses based on the medical history and initiate examinations independently?

It is definitely conceivable if you look at how our product has changed over the last 10 years. For such further developments, we try to combine the actual needs of our customers with research projects in order to develop the product within the framework of co-innovation. Our partners also play an important role in this, as our platform gives them the opportunity to develop their own sub-products (so-called apps). If you think about the feasibility of such a solution, I also see a large overlap with existing features of Celonis products. To take the example of diagnoses, they are very similar to data analysis. It is about collecting data about the patient, analyzing it and drawing the right conclusions from it. This is a realistic application scenario, especially for standardized clinical pictures and their treatment. For these standardized treatments, there is a so-called SOP (Standard Operating Procedure) in which each step in the treatment process of a patient is precisely specified. However, it is important for me to point out two things in this context: Our EMS system is not designed to replace a human being, in this case a physician. In a possible scenario, it would be able to make a diagnosis and also show possible treatment steps, but these should serve as a suggestion or support for the user (=doctor) and not be implemented unsupervised. The second point concerns regulation and is the reason why Celonis has so far been very reluctant to further develop the EMS in this direction. As soon as, and that would be the case here, our system is actively involved in patient treatment, Celonis needs to be certified as a medical device. This process is very time-consuming and cost-intensive. The extent to which Celonis will develop in this area remains to be seen, and will certainly also depend on how high the effort required to overcome the regulatory hurdles is to be estimated.

In your opinion, how high could the savings potential be in medicine through the use of technologies such as process mining? Are there comparable values from industry here?

Generally speaking, the savings potential is of course enormous. Unlike in industry, where processes and procedures are usually designed as efficiently as possible as soon as they are implemented, the process landscape in medicine has grown historically. As a result, the existing potential is naturally all the greater due to the existing, often convoluted processes. However, medicine is not a sector like any other and one must differentiate when evaluating the savings potential. On the one hand, we have the private hospital associations. These are usually already much better structured than public or non-profit hospitals and have central purchasing organizations, etc. that ensure the supply of the hospitals. It is certainly relatively easy to save a great deal here. However, if you compare this with a public hospital, you will see that each hospital works autonomously and may have implemented completely different solutions. It is therefore not possible to make a general estimate of the savings potential. To counteract this, we usually conduct pilot projects with customers in which we validate the potential for the hospitals based on their processes and data. Within the framework of these pilot projects, we work closely with our partners and the result of these projects then serves as a basis for the decision to use Celonis and ultimately the purchase decision.

Your personal opinion: Do you think such operational optimization systems work more for or more against humane medicine?

100% for humane medicine. I believe that one must not make the mistake of confusing optimization with maximization. My experience has shown me that most hospitals have to provide more and more services with very limited resources. That is exactly where we want to start. I believe process mining should not be about optimizing the time with the patient, but rather the operation around the actual treatment.

In addition, the healthcare sector is facing fundamental challenges that process mining cannot completely solve, but can certainly make a contribution to overcoming. For me, these include first and foremost the shortage of skilled workers in almost all areas, which has been discussed for years. Especially in nursing, this effect has been even more noticeable since the Covid19 pandemic. Coupled with an increasingly aging population - unfortunately, statistically speaking, more diseases go hand in hand with higher age - we are heading for an ever greater burden on our healthcare system. In addition to the aforementioned shortage of personnel, hospitals are also under enormous cost pressure (budget constraints, flat-rate remuneration, etc.). Process optimization, for example by means of process mining, is almost a necessity for me in this context in order to ensure at least constant, if not increasing quality in human medicine.

Who should I contact if I want to try Celonis in my clinic or practice? With which systems is Celonis compatible so far?

We have a very broad network of management consultancies already mentioned that offer pilot projects with Celonis. An overview can be found on our homepage. Another option is to take the first steps independently with our free solution SNAP. Otherwise, I am always available to evaluate the best approach for testing with interested parties!

Thank you so much for your time!


Contact Information: