An analytical approach to aggregate patient inflows to a simulation model over the radiotherapy process

Area of Science:

• Medical Physics
• Health Systems Engineering
• Computational Biology

Background:

• System dynamics (SD) models for radiotherapy (RT) require simplified patient care pathways (workflows) to maintain performance.
• A large RT department can have over 100 workflows, posing a challenge for model complexity.
• This study investigates reducing workflow numbers in an SD model of the RT preparatory process.

Purpose of the Study:

• To evaluate the impact of reducing radiotherapy (RT) workflows on the performance of a system dynamics (SD) simulation model.
• To compare different data grouping strategies for inputting patient pathways into RT simulation models.

Main Methods:

• A system dynamics sub-structure for the preparatory RT process was developed using real patient data from 2015-2016.
• Radiotherapy workflow similarity was quantified using utilization rate differences and correlation coefficients.
• Workflow grouping strategies included the 80/20 Pareto rule, merging all data, and a custom algorithm (A1/A2) with specific similarity criteria.

Main Results:

• 128 RT workflows were identified for 3209 patients.
• The 80/20 Pareto rule yielded 14/8/21 groups (curative/palliative/disregarding intent).
• Custom algorithms resulted in 7-82 groups, with Pareto and A2 (r≥85) showing comparable results to the reference (all workflows separate).

Conclusions:

• The strategy for grouping patient input data significantly impacts the performance of radiotherapy simulation models.
• Grouping by the Pareto rule or resource utilization similarity better reflects departmental effects than merging all data.
• The proposed similarity-based grouping approach can simplify complex process models without compromising overall performance.