Process mining using data science is an excellent method to optimize processes. In our study, we tested in a project for improvement of the localization of nonpalpable breast cancer lesions using the SiriusLink approach. The project was conducted in collaboration with OLVG and Sirius Medical.

Process mining

Fig 1. Process mining in Healthcare, Springer 2015


We studied how process mining methods could help in studying breast cancer diagnostic and surgical workprocesses by:

(i) identifying the full real world work processes per individual breast cancer patient from diagnosis till end of surgical therapy

(ii) gaining insights from these processes (the mined pathways)

(iii) making recommendations for improved work processes (the hypothetical work processes).

What data did we use?

We collected data for 360 patients over the 3- year period. Per patient, 60 event variables were collected with time stamps. Next, the raw data was cleaned and event logs were made for the purpose of  analysis and visalization.

Raw data

Event logs


Patients were grouped based on the administrative data (for analysis) in two groups. Patient with no nodal involvement (N0) and no neo-adjuvant systemic therapy and patients with nodal involvement (N1-3) plus neo-adjuvant therapy. Insights showed the specific time and activity pathways of the two groups enabling the team to construct process improvement proposals. In a next phase we intent to apply this approach to get insights in the value of the improved work processes and compare these processes to the baseline.


The SiriusLink approach shows the potential of process mining techniques. With the analysis we show that:

we can derive real world work process insights from administrative hospital data with process mining

there by facilitating business process improvement (creating hypothetical work processes).

Scan here to read more.

Published On: August 24, 2022 / Categories: Machine learning, Prediction, Process Mining /