Geetha Baskaran Andrzej Bargiela Rong Qu School of ... · W1 to w5 (Complex) Scheduling: W1 to W5...
Transcript of Geetha Baskaran Andrzej Bargiela Rong Qu School of ... · W1 to w5 (Complex) Scheduling: W1 to W5...
3/5/13
1
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN 08.06.12 – 11.06.12
Geetha Baskaran Andrzej Bargiela School of Foundation Studies School of Computer Science
University of Nottingham (Malaysia Campus) University of Nottingham (Malaysia Campus)
Rong Qu School of Computer Science
University of Nottingham (Jubilee Campus)
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
OVERVIEW OF PRESENTATION
Introduction Previous Studies Nurse Rostering Problem Information Granulation Rationale Proposed Studies Example of Nurse Rostering Problem Proposed Solution Computational Results Conclusion
08.06.12 – 11.06.12
3/5/13
2
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
INTRODUCTION
The challenge of healthcare Flexible healthcare staff schedules Need for automated rostering
08.06.12 – 11.06.12
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
Early research – 1970’ties Mathematical programming Heuristics Optimisation …. Data aggregation (granulation)
PREVIOUS STUDIES
08.06.12 – 11.06.12
3/5/13
3
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
EXAMPLE OF NURSE ROSTERING PROBLEM
NRP Scheduling period = 5 weeks
16 nurses in ward
12 (FT) 36h nurses
All HC & SC are satisfied
1 (PT) 32h nurse
3 (PT) 20h nurses
08.06.12 – 11.06.12
ORTEC studies (Dutch hospitals)
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
NURSE ROSTERING PROBLEM
Nurses
E D R N L
GEETHA BASKARAN
One Week Shift(One Nurse)
08.06.12 – 11.06.12
patterns
3/5/13
4
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
NURSE ROSTERING PROBLEM
Nurse rostering problem involves allocating the required workload to nurses subject to a number of constraints.
Constraints are categorized into two groups; Hard Constraints are those that must be
satisfied to obtain feasible solutions. Eg:
3 3 3 1 E D N L
Demands need to be fulfilled
08.06.12 – 11.06.12
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
Shifts and personal cover demand
E D N L
08.06.12 – 11.06.12
3/5/13
5
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
EXAMPLE OF NURSE ROSTERING PROBLEM
1. Demands need to be fulfilled
2. For each day, 1 nurse may start only one shift. 3. Within a scheduling period, a nurse is allowed to exceed the number of hours for
which he/she is available for his/her department by at most 4 hours. 4. The maximum labor time per week is on average 36 hours over a period of 13
consecutive weeks if this period does not include work during night shifts. 5. The maximum number of night shifts is 3 per period of 5 consecutive weeks. 6. A nurse must receive at least 2 weekends off duty per 5 week period. A weekend
off duty lasts 60 hours including Saturday 00:00 to Monday 04:00. 7. Following a series of at least 2 consecutive night shifts, a 42 hours rest is required. 8. During any period of 24 consecutive hours, at least 11 hours of rest is required. A
night shift has to be followed by at least 14 hours rest. An exception is that once in a period of 21 days for 24 consecutive hours, the resting time may be reduced to 8 hours.
9. The number of consecutive night shifts is at most 3. 10. The number of consecutive shifts (workdays) is at most 6. 11. One of the full-time nurses requires not receiving any late shifts
Hard Constraints – ORTEC studies (Dutch hospitals)
08.06.12 – 11.06.12
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
NURSE ROSTERING PROBLEM
GEETHA BASKARAN
Feasible (Nurse 1)
Infeasible(Nurse 2)
3 Nights
08.06.12 – 11.06.12
HC = 9
3/5/13
6
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
NURSE ROSTERING PROBLEM
Soft Constraints are typically time related constraints. They are desirable but not compulsory, and thus can be violated.
Eg:
One Week Shift(One Nurse)
COST = 5
An early shift after a day shift should be avoided. 5
08.06.12 – 11.06.12
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
EXAMPLE OF NURSE ROSTERING PROBLEM
1 For the period of Friday 23:00 to Monday 0:00, a nurse should have either no shifts or at least 2 1000 shifts (Complete Weekend).
2 Avoid sequence of shifts with length of 1 for all nurses. 1000 3a For nurses with availability of 30-36 hours per week, the length of a series of night shifts should be 1000
within the range [2, 3]. It could be part of, but not before, another sequence of shifts. 3b For nurses with availability of 0-30 hours per week, the length of a series of night shifts should be 1000
within the range [2, 3]. It could be part of, but not before, another sequence of shifts. 4 The rest after a series of day, early or late shifts is at least 2 days. 100 5a For nurses with availability of 30-36 hours per week, the number of shifts is within the range [4, 5] 10
per week. 5b For nurses with availability of 0-30 hours per week, the number of shifts is within the range [2, 3] 10
per week. 6a For nurses with availability of 30-36 hours per week, the length of a series of shifts should be within 10
the range of [4, 6]. 6b For nurses with availability of 0-30 hours per week, the length of a series of shifts should be within 10
the range [2, 3]. 7 For all nurse, the length of a series of early shifts should be within the range [2, 3]. It could be within 10
another series of shifts. 8 For all nurse the length of a series of late shifts should be within the range of [2, 3]. It could be within 10
another series of shifts. 9a An early shift after a day shift should be avoided. 5 9b An early shift after a late shift should be avoided. 5 9c A day shift after a late shift should be avoided. 5 10 A night shift after an early shift should be avoided. 1
Soft Constraints – ORTEC studies (Dutch hospitals) Soft Constraints Cost
08.06.12 – 11.06.12
3/5/13
7
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
Penalties of the violations within blocks
Shifts and personal cover demand
E D N L
Succeeding shifts
N E D L Preceding ok n/f n/f n/f Shift E ok ok ok ok
D ok 5 ok ok L ok n/for 5* n/for 5* ok
9a An early shift after a day shift should be avoided. 5 9b An early shift after a late shift should be avoided. 5 9c A day shift after a late shift should be avoided. 5 10 A night shift after an early shift should be avoided. 1
08.06.12 – 11.06.12
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
E
D
N
L
Reducing the number of shift types to be considered (2-stage approach)
Demand
N
d
d
Granulation of Shift Types
08.06.12 – 11.06.12
R
3/5/13
8
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
N N R R E E E E R R E D R R E E E D R R R N N R R E E E E R R E D R R E E D D R R R … …. N N R R E E D D R R E E R R E E E D R R R ….. ….. N N R R D D D D R R E E R R E E E D R R R ….. N N R R D L L L R R E E R R E E E D R R R ….. N N R R L L L L R R E E R R E E E D R R R …. ….
WEEK1 WEEK2 WEEK3
Offline preparation Process
Granulation of Shift Types
• Convert the problem space from {N,E,L,D,R} to a smaller space of {N,d,R}.
08.06.12 – 11.06.12
WEEK1 WEEK2 WEEK3
N N R R d d d d R R d d R R d d d d R R R
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
• The 18 feasible zero-cost shift patterns which were identified for the sample problem of a ward in a Dutch hospital Which satisfies All the constraints for the 36h nurses. These patterns will be used for the 32 h nurse also since it falls in the same category
Granulation of Shifts into Patterns
P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17 P18
08.06.12 – 11.06.12
Offline preparation Process
3/5/13
9
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
• The 15 feasible zero-cost shift patterns which were identified for the sample problem of a ward in a Dutch hospital Which satisfies All the constraints for the 20h nurses
08.06.12 – 11.06.12
Granulation of Shifts into Patterns
Offline preparation Process
NNRRRRR RRNNRRR RDNNRRR RRRRNNN DDRRRRR RDDRRRR DDDRRRR RRDDRRR RDDDRRR RRRDDRR DRRDDRR RRDDDRR RRRRRDD DRRRRDD RRRRDDD
P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
Granulation of Shifts into Patterns
• Use available patterns to satisfy demands.
08.06.12 – 11.06.12
WEEK1 WEEK2 WEEK3
N N R R d d d d R R d d d d d d R R R D D
WEEK1 WEEK2 WEEK3
P3 P17 P12
3/5/13
10
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
Previous research
Granulation Approach vs. Traditional Rostering
GEETHA BASKARAN
Convert dNR
schedule to EDL
Scheduling: W1 to w5 (Complex)
Scheduling:W1 to W5
EDLNR -> dNR->
Patterns in dNR space
EDLNR
Individual shifts
Prune converted patterns using (constraints 7-10)
08.06.12 – 11.06.12
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN 08.06.12 – 11.06.12
Granulation Approach
• Granulation of the shift types (EDLNR dNR)
• Granulation of shifts into patterns
• “zero-cost patterns”
• “non-zero-cost patterns”
• Schedulling of patterns
• Conversion of schedule in dNR into schedule in EDLNR
3/5/13
11
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
The week 1 feasible scheduling sequence generated from the patterns generated on the different types of nurses and which satisfies all the constraints according to the types of nurses. Besides, these patterns satisfy the personnel cover requirements (9999966) for day and (1111111) for night shift.
Week 1 Schedule
08.06.12 – 11.06.12
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
Week 1-5 Schedule
08.06.12 – 11.06.12
3/5/13
12
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN 08.06.12 – 11.06.12
dNR RESULT
Week 1-5 Schedule
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN 08.06.12 – 11.06.12
EDLNR RESULT
Week 1-5 Schedule
3/5/13
13
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN 08.06.12 – 11.06.12
EDLNR RESULT
Week 1-5 Schedule
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN 08.06.12 – 11.06.12
Solution Cost
5-weeks schedule 31-day schedule
Published ORTEC solution
NA 270
Proposed data granulation solution
250 160
3/5/13
14
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
CONCLUSION
Nurse rostering problem - detailed time constraints and different types of day-shifts represents computational challenge.
The complexity of the problem is due to large solution space and the many constraints that need to be satisfied.
The novel approach proposed here involves a simplification of the original problem by a sensible grouping of shift types and the grouping of the individual shifts into weekly sequences that can be used to derive schedules in this reduced problem space.
Subsequently the schedules from the reduced problem space are translated into the original problem space by taking into account the constraints that could not be represented in the reduced space.
The grouping of the shift types into a prototype shift and the grouping of shift sequences into weekly patterns used in scheduling are collectively referred to as information granulation
08.06.12 – 11.06.12
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
QUESTION & ANSWER
THANK YOU
08.06.12 – 11.06.12
3/5/13
15
25th EUROPEAN CONFERENCE ON OPERATIONAL RESEARCH
GEETHA BASKARAN
Information Granulation Rationale
The concept of information granulation was first introduced by Zadeh in the
context of fuzzy sets in 1979 . The fundamental tasks of Granular Computing (GrC) is the construction of Information Granules, a process that is called “information granulation”. Some of the basic notions of granular computing are :
granules hierarchies levels
Example of Granulation Process by
Corrado Mencar
08.06.12 – 11.06.12