The Dynamic Patient Admission Scheduling Problem under Uncertainty (PASU) extends the Patient Admission Scheduling (PAS) problem, introduced by Demeester et al. , by including several real-world features, such as the presence of emergency patients, uncertainty in stay lengths, and the possibility to delay admissions.
Therefore, we propose the definition of a new version of the problem that comprises the basic notions:
The PASU problem consists in assigning a room to each patient for a number of days equal to her/his stay period, starting in a day not before his/her planned admission. The assignment is subject to the following constraints and objectives (soft constraints):
Constraints are split into hard constraints, that must be satisfied, and soft constraints (or objectives) that can be violated and contribute to the objective function, as shown in the following table. Each soft constraint is associated with a weight, that accounts for its relative importance.
We designed a parametrized generator which receives as parameters the number of departments, rooms, features, patients, and days. It creates a random instance based on predefined distributions concerning various features such as the stay length, the room capacity, the number of specialisms, and so on. You can download the instance generator.
We have created 9 sets of 50 instances each, using the values shown in following table. The datasets correspond to three different sizes in terms of number of patients and planning horizons. When the horizon is doubled, the number of patients is doubled as well so as to maintain approximately the same average bed occupancy.
A description of the input and output format can be found here.
To download all instances and solutions in one shot, the public subversion repository is available:
git clone https://bitbucket.org/satt/pasu-instances.git
You can download the C++ source code of a solution validator.
The table summarizes the features of the benchmark instances of the PAS problem by Demeester et al. in terms of number of beds, rooms, departments, patients and multi speciality patients. The one in parenthesis is the number of patients reported in the instance files, but it also includes patients that stay for zero days (same admission and discharge date) or that have an admission day subsequent to the end of the planning horizon. The last column reports our best results.
|Instance||Beds||Rooms||Depts||Patients||MP Patients||Best results|
Here you can download the original validator implemented by Peter Demeester solution validator.
The Patient Admission Scheduling with Operating Room Constraints is a further extension of the PAS problem, which considers also constraints about the utilization of operating rooms for patients that have to undergo a surgery.
Instances, generator, solutions and validator are published here.
S. Ceschia and A. Schaerf. Local search and lower bounds for the patient admission scheduling problem. Computers & Operations Research, 38(10):1452-1463, 2011.
S. Ceschia, A. Schaerf. Modeling and Solving the Dynamic Patient Admission Scheduling Problem under Uncertainty. Artificial Intelligence in Medicine, 56(3):199-205, 2012.
S. Ceschia and A. Schaerf. Dynamic patient admission scheduling with operating room constraints, flexible horizons, and patient delays. Journal of Scheduling, 1-13, 2014. Online first.