AbstractThe main subject matter of this thesis concerns radiotherapy patient scheduling subproblems formulated as four separate shop scheduling problem models (i.e. hybrid flowshop, flowshop, mixed shop and multiple identical parallel machine scheduling problems) based on the characteristics of the intricate real-life treatment processes observed at the Arden Cancer Centre in Coventry, UK. Insight into these processes was gained by developing and using a novel discrete-event simulation (DES) model of the four units of the radiotherapy department. By typifying the subproblems as well-known scheduling problem models, it was intended that methods amenable to them such as heuristics be used in the study.
Four novel constructive heuristics based on priority dispatching rules and strategies adapted from some established algorithms have been developed and implemented using the C++ programming language. Further, these heuristics were incorporated into the DES model to create schedules of appointments for the patients generated daily. The effectiveness and efficiency of the constructive heuristics have been tested using the following performance criteria: minimising i) average waiting time to the start of treatment, and ii) average percentage of patients late for their treatment, and iii) the amount of overtime slots used for the patients received in a given period of time. The coordinated constructive heuristics and the DES model have also been tested using possible alternative pathways patients can follow in the treatment unit. The aim of these tests was to compare the efficiency of the radiotherapy department’s current pathway to other possible pathways. Further, strategies for using maximum allowed breaches of targeted due dates, reserved slots for critical treatments and overtime slots was also included in the heuristics.
The results of several tests showed that the heuristics created schedules of appointments whose average waiting times for emergency, palliative and radical treatments improved by about 50%, 34% and 41%, respectively, compared to the historical data. However, their major slack was evidenced by the fact that about 13% of the patients needing palliative treatment were expected to be late for treatment compared to about 1% of those requiring radical treatment.
|Date of Award||2010|
|Sponsors||Engineering and Physical Sciences Research Council & University Hospitals Coventry and Warwickshire NHS Trust|
|Supervisor||Dobrila Petrovic (Supervisor), Colin Reeves (Supervisor) & Olivier Haas (Supervisor)|
- operational research
- healthcare simulation
- patient waiting times
- radiotherapy planning