Modeling and Decision Support for Public Health Infrastructure for Emergency Response Eva K. Lee, Ph.D. Director and Associate Professor Center for Operations Research in Medicine & Healthcare School of Industrial & Systems Engineering Georgia Institute of Technology Winship Cancer Institute Emory University School of Medicine
[email protected]
What we do • Operations Research – – World War II – target precision, ground logistics and planning and resource supply-chain, health medical plans and execution, soldiers’ well-being (meal planning) and rescue operations – Name – military mission – Operations X – Optimizing the system under limited resources
• A rich class of mathematics and computational tools (theory and computation) • Some examples of my work – – Information and decision technology/tools for medicine and healthcare, biodefense and public health planning, telecommunication, transportation networks, scheduling ad routing, finance and economics, intelligence and mission-critical logistics
Specific Objectives in My Work • Operations Research – developing realistic models and achieving theoretical, computational and engineering advances for systems, operations, and process improvement; making decision “easier” • Biodefense – large-scale dispensing: strategic and operational planning and decision support tool (~2002, with CDC, SNS)
Mass Dispensing • Requires the rapid establishment of a network of dispensing facilities (PODs) and health facilities that are flexible, scalable and sustainable for medical prophylaxis and treatment of the general population. • Optimizing throughput under limited resources remains a daunting task for efficient POD operations. • Understanding tradeoffs
Large-Scale Medical Dispensing
Establish Efficient Dispensing Network
Legend Worker Movement =
Population Flow
Public Movement = Materiel Movement = Public Assembly Point = Worker Assembly Point = Point of Dispensing =
Population 2
Population =
Population 1
Population 3
Central Assembly Point I 80 School Buses Per Hour (3600 PPH)
Busing of individuals to PODS 80 School Buses Per Hour (3600 PPH)
POD
Population 4
POD
RSS DATA ENTRY CENTER
POD Worker Assembly Point (initial POD for essential workers)
80 School Buses Per Hour (3600 PPH)
Population 5
Central Assembly Point II
POD 80 School Buses Per Hour (3600 PPH)
Population 10
POD 80 School Buses Per Hour (3600 PPH)
POD POD
Population 6
Central Assembly 80 School Buses Point Per Hour (3600 PPH) III
Population 9 Population 7
Population 8
Regional Dispensing – Challenges • Given a regional population, determine where and how many PODs are needed for efficient operations • Determine assignment of individuals to various PODs • Determine the best mode of dispensing (drivethrough, walk-through, postal,.… ), and POD layout design • Determine staffing/resources needed at each POD for required throughput (maximize throughput, equalize utilization, minimize cycle time) • Anthrax: 48-hour window • How to make these decisions?
Metro-Atlanta
400000
Number of Households
350000 300000 250000 200000 150000 100000 50000 0
01-1 01-2 02-0 03-1 03-2 03-3 03-4 03-5 04-0 05-2 10-0 District
About 5.2 million
“Optimal” Facility Locations 400
Number of Facilities Required
350 300 250 200 150 100 50 0 0
5
10
15
20
25
30
Maximum Allowed Travel Distance (miles) Infinite Capacity
Capacity 2000
Capacity 1500
Capacity 1000
Capacity 500
Tradeoffs – no. of facilities vs distance traveled vs facility hourly capacity
Realistic & Cost-Effective Dispensing – Public Health + Private Sector •Heterogeneous mixed of dispensing modalities:
Hourly throughput
Public PODs: PODs: Drive-through (brown) Walk-through (blue) Private/Closed PODs: PODs: University/college campus Assisted living facilities Prisons/Jails Large corporation Offices Airport Mobile PODs: PODs: Deliver to special need population (disabled, home-bound, etc)
Zoom in: Most Cost-effective Public PODs
Heterogeneous mode of dispensing, Variation of throughput
Resource Allocation – Number of Workers Required
Minimizing total number of critical personnel required to man the public PODs for each 12-hour shift 700 600 500 400 300 200 100 0 01-1
01-2
02-0
03-1
03-2
03-3
03-4
03-5
04-0
05-2
District Combined Clinic
Disjoint Clinic
Drive-Through Clinic
10-0
Summary • Systems approach to analyze a problem • A powerful modeling and algorithmic design for decision making/analysis • Computer system allows rapid training, and knowledge expansion (GIS, real-time feedback and data processing, etc…) • Plan and execute and refine (drills, iterative knowledge acquiring) • National databank is critical for knowledge sharing and more • RealOpt© (free for public health preparedness usage) -- http://www.isye.gatech.edu/medicalor • Communication, interoperability, survivability • Require multi-disciplinary collaboration
References 1.
2. 3.
4.
5.
6.
EK Lee, HK Smalley, Y Zhang, F Pietz, B Benecke, 2008, Facility location and multi-modality mass dispensing strategies and emergency response for biodefense and infectious disease outbreaks, International Journal on Risk Assessment and Management. To appear. EK Lee, Ready for Worst-Case Scenarios. Feature article in Operations Research and Management Science, 35(1):28-34, 2008. EK Lee, S Maheshwary, J Mason, W Glisson, Large-scale dispensing for emergency response to bioterrorism & infectious disease outbreak. Interfaces -- OR Applications for Homeland Defense, 36(6): 591-607, 2006. EK Lee, S Maheshwary, J Mason, W Glisson, Decision support system for mass dispensing of medications for infectious disease outbreaks & bioterrorist attacks. Annals of Operations Research, 148: 25-53, 2006. EK Lee, S Maheshwary, J Mason, Real-Time staff allocation for emergency treatment response of biologic threats and infectious disease outbreak. Medical Decision Making 2005. Selected as INFORMS William Pierskalla Best Paper Award on research excellence in HealthCare and Management Science, Nov 2005. RealOpt©, System and user manual, 2003-2008.
The End
