To yield a slide that can be analysed under a microscope, respectively scanned by a whole-slide imaging solution in the case of digital pathology, the specimen has to undergo an elaborate preparation process in the pathology process comprising several manual activities (see the figure above). The objective of this work package is to investigate means for optimizing the workflow in the pathology laboratory in order to reduce the overall cycle-time, i.e., the time between specimen arrival and sending a diagnostic report in response.
The respective sub-goals in this work package are:
Process Mining (Reporting)
Excerpt of a process map with performance data (mean durations) of the pathology workflow.
In the first stage of the process, the goal is get a clear understanding about “what is actually happening”, e.g. what pathways specific cases take in laboratory, how long specific activities take, and whether there are concrete clusters of cases. Process Mining is relatively young scientific approach, which combines data science with business process management (BPM) methods and forms the methodology for the first stage of the process.
Screenshot of the simulation tool CPNtools.
Based on the results of the first-stage, we are planning to develop formal simulation (e.g. Coloured Petri nets (CPN) or Abstract Behavioural Specification (ABS)) and queueing models, which can be used to predict the number of incoming specimen, cycle-times for concrete cases, and resource utilization both in the short- and long-term perspective.
Optimal solution: for max(3a+2b) s.t. 1a+1b≤4∧1a-1b≤2
In the final stage of the project, we want, based on the results of preceding stages, to investigate means for optimization, e.g. algorithms that automatically determine the sequence of samples to be worked on. Here, methods from operations research (i.e. [linear|integer-linear|non-linear|dynamic] programming), satisfiability/optimization modulo theories, and genetic programming will provide the theoretical foundation.