Page 332 -

314 12 Analyzing “Spaghetti Processes”

tems tend to be Spaghetti-like. To simplify diagnosis, the log is often preprocessed
as discussed in [12–14]. Moreover, fuzzy mining, as illustrated by Fig. 12.8,isused
to further simplify the model [51].
Mining processes from the event logs generated by Allura Xper systems is very
challenging. The machines consist of many components and can be used in many
different ways. Moreover, logging is rather low-level and changes with every new
version. Nevertheless, there are various opportunities for process and system im-
provements using process mining. These are listed below. Note that opportunities
also apply to other types of systems that are monitored remotely.
• Process mining provides insight into how systems are actually used. This is in-
teresting from a marketing point of view. For example, if a feature is rarely used,
then this may trigger additional after sales activities. It is also possible that, based
on process mining results, the feature is removed or adapted in future systems.
• Testing can be improved by constructing test scenarios based on the actual use of
the machines. For instance, for medical equipment it is essential to prove that the
system was tested under realistic circumstances.
• Process mining can be used to improve the reliability of next generations of sys-
tems. Better systems can be designed by understanding why and when systems
malfunction.
• Process mining can also be used for fault diagnosis. By learning from earlier
problems, it is possible to find the root cause for new problems that emerge. For
example, we have analyzed under which circumstances particular components
are replaced. This resulted in a set of signatures. When a malfunctioning X-ray
machine exhibits a particular “signature” behavior, the service engineer knows
what component to replace.
• Historic information can also be used to predict future problems. For instance, it
is possible to anticipate that an X-ray tube is about to fail. Hence, the tube can be
replaced before the machine starts to malfunction.

These examples show the potential of remote diagnostics based on process mining.

12.3.2.3 AMC Hospital

Hospitals are particularly interesting from a process mining point of view. By law,
hospitals need to record more and more data in a systematic manner and all event
data are connected to patients. Therefore, it is relatively straightforward to corre-
late events. For example, by Dutch law all hospitals need to record the diagnostic
and treatment steps at the level of individual patients in order to receive payments.
This so-called “Diagnose Behandeling Combinatie” (DBC) forces Dutch hospitals
to record all kinds of events. There is also consensus that processes in hospitals
can be improved. Unlike most other domains, operational care processes are not
tightly controlled by management. This, combined with the intrinsic variability of
care processes, results in Spaghetti.