Patient Journey Optimization using a Multi-agent approach

1
Patient Journey Optimization using a Multi-agent
approach

• Choi Chung Ho

2
Agenda

• Introduction
• Problem formulation
• Scheduling framework
• Agent coordination
• Experiments
• Conclusion

3
Introduction
4
Our goal

• To improve patient journey by reducing undesired
  waiting time for patients

5
How to achieve our goal?

• To schedule patients in such a way that medical
  resources could be utilized in a more efficient
  manner

6
Why using a multi-agent approach?

• Hospitals are found to have a decentralized
  structure
• ? A multi-agent approach is proposed as it favors
  the coordination between geographically
  distributed entities

7
Related works of using a multi-agent approach for
patient scheduling

• T. O. Paulussen, I. S. Dept, K. S. Decker, A.
  Heinzl, and N. R. Jennings. Distributed patient
  scheduling in hospitals. In Coordination and
  Agent Technology in Value Networks. GITO, pages
  12241232. Morgan Kaufmann, 2003.
• I. Vermeulen, S. Bohte, K. Somefun, and H. La
  Poutre. Improving patient activity schedules by
  multi-agent pareto appointment exchanging. In
  CEC-EEE 06 Proceedings of the The 8th IEEE
  International Conference on E-Commerce Technology
  and The 3rd IEEE International Conference on
  Enterprise Computing, E-Commerce, and E-Services,
  page 9, Washington, DC, USA, 2006. IEEE Computer
  Society.

The use of health state as an utility function
has been challenged
Temporal constraints between treatment operations
are not considered
8
Problem formulation
9
Seven cancer centers in Hong Kong
C HKE, HKW, KC, KE, KW, NTE, NTW
10
Treatment operations and medical resources
Treatment plan
Medical resources (A) Radiotherapy planning
unit, Radiotherapy unit, Operation unit,
Chemotherapy unit
11
Patient journey

• We define Patient journey as
• Duration from the date of admission to the
  date of the last treatment operation completed

12
Scheduling framework
13
Two types of agents

• Patient agent
• Resource agent

14
Patient agent

• A patient agent (Pi) is used to represent one
  cancer patient
• Each Pi stores the corresponding patients
  treatment plan

Treatment plan
15
Resource agent

• A resource agent is used to represent one
  specific medical unit, denoted as Rab a A,
• b C

16
Scheduling algorithm
17
Agent coordination
18
Coordination framework
19
Coordination framework (cont.)

• For each request, it includes
• 1) Earliest possible start date (EPS)
• It is the earliest date on which a treatment
  operation could start
• 2) Latest possible start date (LPS)
• It is the latest date on which a treatment
  operation should start such that the treatment
  operation could be performed earlier

20
Earliest possible start date (EPS)
(j 1) th treatment operation
21
Latest possible start date (LPS)
(j 1) th treatment operation
j th treatment operation
1 day
22
Coordination framework (cont.)
23
Coordination framework (cont.)

• In order to compute the bid value, three binary
  variables were defined
• 1) Last
• 2) Noti
• 3) Temp

24
Coordination framework (cont.)

• Last is a binary variable that specifies whether
  the involving treatment operation is the last one
  in PGs treatment plan
• Last 0 if it is not the last one
  otherwise

1 th treatment operation
2 nd treatment operation
3 rd treatment operation
25
Coordination framework (cont.)

• Noti is a binary variable that specifies whether
  there is a weeks time of notification for the
  target patient agent regarding the exchange
• Noti 0 if there is a weeks time of
  notification otherwise

26
Coordination framework (cont.)

• Temp is a binary variable that specifies whether
  the temporal constraints between treatment
  operations are violated for the target patient
  agent after the proposed exchange
• Temp 0 if no violation otherwise

27
Coordination framework (cont.)

• For each target patient agent PG

28
Coordination framework (cont.)
Coordination process for eliminating unnecessary
exchanges
29
Unnecessary exchanges
30
Experiments
31
Data set

• 5819 cancer patients in Hong Kong, with an
  admission period of 6 months (1/7/2007
  31/12/2007)
• The average length of patient journey is 90.7
  days before applying our framework

32
Experiments (cont.)

• Group A The scheduled treatment plans in the
  dataset are used for the initial assignment
• Group B Only the statistics of the scheduled
  treatment plans and the capacities of medical
  units are used for the initial assignment

33
Experiment settings

• Setting 1) All patient agents are willing to
  exchange their timeslots with others whenever
  there is a Pareto improvement
• Setting 2) Only 20 of the patients of each
  center are allowed to exchange their timeslots
• Setting 3) Patients are only be swapped to a
  nearby cancer center
• Setting 4) Timeslots released by deceased
  patients are allocated to those who have the
  longest patient journey

34
Experimental results
Group A
Group B
35
Experimental results (cont.)
Group B
36
Conclusion
37
Conclusion and future works

• A multi-agent framework has been proposed for
  patient scheduling
• In this framework, while no single patient will
  get a lengthened patient journey, all the
  temporal constraints between treatment operations
  would not be violated

38
Conclusion and future works (cont.)

• Experiments show that the average length of
  patient journey could be reduced by about a
  weeks time by using the proposed framework
• In the future, we are going to see how the bids
  submitted by the target patient agent could be
  defined in a more sophisticated way such that the
  overall patient journey could be shortened in
  greater extent

39
The end