TEACHING / RESEARCH PERSONNEL ACADEMIC ACHIEVEMENT
MATHEMATICS AND INFORMATION SCIENCES
Nurse rostering is a difficult combinatorial problem, due to its size and complexity. It has got many difficulties because of the specific functions that each employee has to perform at a specified number of shifts per working day.
Furthermore, tasks that arise in everyday life are hardly manageable, making it necessary to use specialized software for their coordination. The aim is for the best management of scheduled procedures as well as for a more effective, overall function of the hospital.
The suggested approach was presented at the 8th International Conference "Practice & Theory of Automated Timetabling", 2010, by the research team of the Department of Electrical and Computer Engineering, University of Patras, in cooperation with TEI of Messolonghi and TEI of Epirus. It won first place in all three competition tracks of the same problem as far as its size and the available computation time was concerned.
THE WAY IT WORKS
The software developed solves the problem of nursing rostering, separating the original problem into sub-problems. Each sub-problem can be solved easily and quickly through mathematical programming.
The team has implemented a two phase strategy. In the first phase, the workload for each nurse and for each day of the week has to be decided, while in the second phase the specific daily shifts have to be assigned (one shift per day). Afterwards, an integer programming problem is formulated in order to solve the problem for the first week of the year under study. All possible combinations of nurses day on / day off are examined, and the solution that satisfies the majority of flexible restrictions is selected. For this purpose, another integer programming problem is met, covering as many restrictions as possible regarding the different types of shifts that each employee can perform. The problem is repeatedly solved until an optimal solution is reached.
Finally, cost evaluation has been implemented for calculating the cost of each solution. The cost of the first phase depends on the placement of the nurse on a specific day, while in the second phase the cost takes under consideration the specific kind of shift attributed. The proposal was implemented programmatically using integer programming models combined with heuristic techniques.
Practice and Theory of Automated Timetabling
8th International Conference on the Practice and Theory of Automated Timetabling 2010, Ireland
The team of the Department of Electrical and Computer Engineering, at the University of Patras, in cooperation with the TEI of Messolonghi and the TEI of Epirus, won the first place in the 1st International Nurse Rostering Competition which was hosted by the 8th International Conference on the Practice and Theory of Automated Timetabling 2010 and was organized by CODeS, SINTEF and SaTT.
TEI OF WESTERN GREECE
Dr. Panayiotis Alefragis, Professor of the Department of Telecommunication Systems and Networks, Dr. Christos Valouxis, Research Fellow.
TEI OF EPIRUS
Dr. Christos Gogos, Professor of Department of Finance & Auditing.
UNIVERSITY OF PATRAS
Dr. George Goulas, Post-doctoral Researcher, Dr. Efthymios Housos, Professor of Department of Electrical and Computer Engineering.http://www.kuleuven-kulak.be https://www.kuleuven-kortrijk.be/nrpcompetition/competitor-ranking