University of Twente Student Theses

Login

Optimising online capacity-to-patient assignment: a case study in the Oncological Centre Deventer

Seydel, MSc. T.J. (2019) Optimising online capacity-to-patient assignment: a case study in the Oncological Centre Deventer.

[img] PDF
14MB
Abstract:Problem description The increasing incidence and prevalence rates of cancer result in a major cost burden for society and requires healthcare providers to utilise their capacity in an optimal way. This is also the case for the Oncological Centre Deventer (OCD) within Deventer Ziekenhuis (DZ) where the number of unique patients has increased from 610 in 2016 to 744 unique patients in 2018 (+22%). This raises the question to expand the current capacity of the pharmacy and/or outpatient clinic or to investigate the possibility to better utilise the current capacities. This research investigates the possibility to better utilise the current capacities of the OCD. The utilisation of the capacities of the OCD is, among other things, determined by the appointment schedule for infusion therapies. Appointment requests are handled in an online First Come First Serve (FCFS) manner. The outpatient clinic assistants schedule the appointment requests based on the planning horizon and the available capacity. The problem is defined as an appointment scheduling problem that is divided into two sub-problems to reduce complexity: I. The determination of an appointment day. II. The determination of an appointment time. The determination of an appointment time is defined as an identical parallel machine scheduling problem since the treatment chairs of the OCD are considered identical. This leads to the following research goal: To design a planning methodology for the OCD in order to optimise the utilisation of the capacities of the pharmacy, chairs and nurses, while maintaining medically acceptable access times, achieving stable workloads and ensuring compliance to planning rules. Approach Solving the online scheduling problem can be divided into two phases: Determination of the appointment day and determination of the appointment time. Two methodologies from the literature were adjusted and combined to solve the two phases of the scheduling problem. The first phase of the online scheduling problem is solved by using an adaption of the methodology developed by Alvarado et al. (2018). The second phase of the online scheduling problem is solved by using an adaption of the methodology developed by Hahn-Goldberg et al. (2014). The resulting solution approach consists of a mixed integer linear programming model which minimises the makespan and the number of chairs used during a day. A comparison between the current scheduling methodology and the solution approach that is developed during this research is made by using Monte Carlo simulation. Results The use of the new scheduling algorithm results, as expected, in similar mean utilisation of the capacities of the OCD compared to the current scheduling methodology (62.42% and 62.94%, respectively). However, the standard deviation of 7.13, using the new scheduling algorithm, is 68.9% lower, indicating a reduction in the variability of the utilisation. The reduced variability in workloads within a day is visible in Figure 1 where two schedules are generated in an online manner using the current scheduling methodology and our proposed 2-phase scheduling algorithm. The mean access times to treatment increase from 6.1 days to 6.6 days when using the new scheduling algorithm instead of the current scheduling methodology. The refusal rate of the scheduling requests negligibly increased from 0.19% to 0.61% when using the new scheduling algorithm instead of the current scheduling methodology. Making use of offline scheduling, compared to online scheduling using the current scheduling methodology, results on average in a makespan reduction of 16.4% and a reduction in the number of used chairs on a day of 30.8%. Conclusions Our new scheduling approach performed better in terms of variability in workload between days with an average reduction of 68.9%. The workload was more evenly spread over the days. This, however, resulted in an increase in mean access times to treatment of 8.2% from 6.1 days to 6.6 days. Note that the increase in access times to treatment is small and remains within medically acceptable bounds. The use of our new scheduling approach results on average in a 41 minutes reduction in makespan per day compared to when the current scheduling methodology is used. Furthermore, the average number of chairs used in a day is reduced, which results in less variability in the workload within a day.
Item Type:Essay (Master)
Clients:
Deventer Ziekenhuis, Deventer, The Netherlands
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:31 mathematics, 44 medicine
Programme:Industrial Engineering and Management MSc (60029)
Link to this item:http://purl.utwente.nl/essays/79870
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page