2 Presentations

• 
  03:30 PM - 04:00 PM

  Time Dependent Waiting Time Computation at Dutch Blood Collection Sites

  • Sem van Brummelen, presenter, University of Twente
  • Wim de Kort, Sanquin Research
  • Nico van Dijk, University of Twente

  The Dutch blood bank, Sanquin, collects blood from donors on a non-remunerated basis. It is therefore important to keep waiting times as short as possible. Standard analytic models to calculate waiting times use some form of steady state distribution. If arrivals are nicely spread out over the day, this would be reasonably justified. But, as Dutch blood collection sites allow free walk-ins, the arrival process of donors is highly time dependent. Peaks in arrival patterns show up early in the morning, around noon and just before dinner time. Accurately approximating waiting times is especially important during these peak times. A transient rather than steady state approach is therefore required.
  To closely approximate the waiting times and queue length distributions in time dependent tandem queues, a method that splits the opening hours of the collection sites into short intervals is proposed. Within these intervals we assume arrival and service rates to be constant. An iterative algorithm is developed based on Continuous Time Markov Chains. The algorithm starts with some probability distribution of donors when the collection session starts, i.e. donors that arrived before the collection session started. The approximate algorithm computes the probability distribution of the donors at each phase of the collection process at any given time during the collection session. This enables us to compute various performance measures that are of interest to Sanquin, including queue length distributions, average waiting time and probability of no queue upon arrival of a donor.

• 
  04:00 PM - 04:30 PM

  Balancing the Production of Blood Bags from Donation through Appointment Scheduling

  • Seda Baş, presenter, Yeditepe University
  • Giuliana Carello, Politecnico di Milano
  • Ettore Lanzarone, CNR-IMATI
  • Zeynep Ocak, Yeditepe University
  • Semih Yalçındağ, HEC Montréal and CIRRELT

  Blood is fundamental in several care treatments and surgeries, and plays a crucial role in the healthcare system. It is a limited resource as it can be produced only by donors and its shelf life is short. Blood Donation (BD) system aims at providing an adequate supply of blood bags to transfusion centers and hospitals. Its main phases are blood collection, screening, storage, distribution and utilization. An effective collection of blood bags from donors, through a suitable scheduling of donations, is fundamental for adequately feeding the entire BD system and optimizing blood usage. Indeed, blood collection may represent the bottleneck of the whole BD. Despite its relevance, to the best of our knowledge, donor scheduling is only marginally addressed in the literature, whereas a high number of papers deal with blood storage and distribution.
  We address an appointment scheduling problem for the donations in a blood collection center, with the goal of balancing the production of the different blood products among days, in order to provide a quite constant feeding of the BD system. We consider both booking donors and donors arriving at the center without reservation. The proposed approach consists of pre-allocating donation slots for booking donors, which are then used by the call center for scheduling real donation appointments. The pre-allocation is then refreshed at fixed frequency to account for the newly scheduled appointments. The pre-allocation problem is modeled as an integer linear programming model.
  The approach has been applied to the real case of the Milan Department of the Associazione Volontari Italiani Sangue (AVIS), one of the main BD associations in Italy. Results on small examples based on this real case validate the approach and confirm its applicability and effectiveness.

  Keywords: Blood Donation System; Donor Appointment Scheduling; Balancing Blood Production; Integer Linear Programming Model.