1 / 35

A Decision-Making Tool for a Regional Network of Clinical Laboratories

A Decision-Making Tool for a Regional Network of Clinical Laboratories. Ali Erdem Banak Berk Torun Ladin Uğur. Main Outline. Introduction Process Problem Definition Project Scope Literature Review Directed Graph Mathematical Model Sensitivity Analysis DSS Implementation

owena
Download Presentation

A Decision-Making Tool for a Regional Network of Clinical Laboratories

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. A Decision-Making Tool for a RegionalNetwork of ClinicalLaboratories Ali Erdem Banak Berk Torun Ladin Uğur

  2. MainOutline • Introduction • Process • Problem Definition • Project Scope • LiteratureReview • DirectedGraph • Mathematical Model • SensitivityAnalysis • DSS Implementation • Program Architecture • DSS Utilization • Conclusions • HealthcareResearch in Turkey

  3. Introduction • AndalusianHealthService (AHA) is thegovernmentrunhealthcaresystemforautonomouscommunity of Andalusia

  4. Providesuniversalhealthcare service tomorethan 8 millionpeople • As of 1 January 2007 employs 83,132 professionals, 20,810 of them in primary care and 62,322 in specialized care • Has laboratoriesallaroundtheregionunderthe name RNCL

  5. Process • RNCL collectsbiologicalsamplesfromanylab in itsnetwork • Samplesareanalyzed in RNCL’sownlabsoroutsourcedto a privatelab • Resultsare sent backtothelab in whichtheywerecollected • In 2008, AHA decidedtorun a projecttoefficientlyallocatetheirresources

  6. Problem Definition • Lack of planningprocedureinfluencesthechoice of theprocessinglab • Excessiveshippingcostsbecause of theindividualcontracts • Althoughlabs can handlelargerworkloads, theyprefersendingtheteststotheirpreferredhospitals

  7. Scope of the Project • A newreference model toincreasecooperationbetweenlaboratories • A newplanningproceduretobetterutilizeRNCL’slaboratoriesandreducethenumber of outsourcedtests

  8. Regardless of time needed, DSS must be abletouseinput data toresolvedifferentscenarios • Itmust be abletoanalyzetheworkloadandflowassignments • Itmust be abletocomparescenarios

  9. LiteratureReview • EkşioğluandJin (2006) CrossFacilityProductionandTransportationPlanning Problem withPerishableInventory • Ekşioğlu et al (2007) A LagrangeanHeuristicForIntegratedProductionandTransportationPlanningProblems in a Dynamic, Multi-Item, TwoLayerSupplyChain

  10. AndreattaandLulli (2008) A Multi-Period TSP withStochasticRegularandUrgentDemands • OswaldandStirn (2008) A VehicleRoutingAlgorithmfortheDistribution of FreshVegetablesandSimilarPerishableFood • AmbrossinoandSciomachen (2007)A FoodDistribution Network Problem: A CaseStudy

  11. DirectedGraph • SC: ServingCenter • OC: OutsourcingCenter • POE: Point of Extraction • DTP: Demand Transfer Point Arcslongerthan 250 km areexcludedfromthe model forsamplestability

  12. Mathematical Model • Problem is modelled as multicommodity minimum costflow problem. • Decisionvariablesare link selectionandflowassignment. • Discountfactorsareattachedtotheflow in eachconnection. • 488 verticesand 2477 arcsforAndalusia • 6 differenttypes of samples

  13. Objectivefunctionincludes;shipping, transhipment, processing, outsourcing, penaltycostfor not achieving minimum workloadandpenaltyforexcesstranshipment.

  14. Wehavepenaltycostsforexcesstranshipmentsandlowusage of RNCL laboratories.

  15. Per-unitshipmentcostalongeach link is a discretefunction of theflowalongthat link since it is possibletohave a betterper-unitpricefor a largerbatch. • Firmsofferthesame set of discountfactorforeach link. • Eachrange has lowerbound, upperboundanddiscountfactor.

  16. Objectivefunction;

  17. Constraints • A1 is similartoequalizing C(0k)’s to 0 • A2 regulatesdemand. • A3 is thecapacityconstraint. • A4 is theflowbalanceequation.

  18. A5 is transhipmentbalance. • A6 and A7 is trafficbounds. • A8 forces us touse 1 carrier in a specificrange. • A9 relatesflowvariables.

  19. A9 relatesflowvariables. • A10 decidesdiscountfactors.

  20. Linearization of excesstranshipmentvariables

  21. Linearization of minimum workloadvariables.

  22. Aftergettingthesolution plan fromthe model, AGA (averagegeographicalaccessibility) is found in ordertoestimateexpectedquality of service (QoS) whichindicatesthethenumber of transhipmentseach test needstoreachthelaboratory.

  23. Sensitivity Analysis • Sensitivityanalysisaboutdifferentscenarioswillprovideinsights. • Design of experiments (DOE) is used. • 3 type of parameters. • Parametersrelatedtooptimizationtool (Stop time) • Model parameters (penaltycosts) • Network topology (Size andcomplexity)

  24. 4 responsesaremeasured • Value of objectivefunction • AGA indicator • Total outsourcingcost • Gaptothe optimal solution • Singlefactoranalysis of variance, single-degree-of-freedom ANOVA andanalysis of meanstechnique is used

  25. ConclusionsFrom Analysis • Allparametershavesignificanteffect on theresponses; exceptthethresholdfortriggeringthepenaltycost of excesstranshipment. • Network size is themostinfluentialparameter. • Outsourcingleveldoes not depend on network size. • Quality of service is affectedbyalltheparametersexceptthethresholdfortriggeringthepenaltycost of excesstranshipment.

  26. DSS ToolImplementation • A collaborativeprojectwaslaunched in conjuctionwithnew RNCL planningapproach • Objectivewastodevelop a plan formostresourceconsuming service: provision of laboratories

  27. RNCL authoritiesprovided data andsharedbusinessrules • Theyalsoincreasedawarenesstoeliminateanyresistanceto DSS by RNCL staff.

  28. Program Architecture • An optimization engine coded in AMPL • Inputsaretakenby engine from RNCL database • DSS is web-based • Graphicaluserinterface is embeddedforuser-friendliness • Via GUI differentscenarios can be tested

  29. Resultingroutesaredisplayed on GoogleMaps API • Optimal routes can be savedforfuturecomparisons

  30. DSS Utilization • In 2009, RNCL usedthe DSS todecidewhichsubset of newfacilitiestoactivatefromallalternatives • Costsfellfrom 8 million € to 0,2 million € in 2008-2009 period, primarilyduetoreducedoutsourcing

  31. Conclusions • Recommendedorganizationalchangeswere • A newreference model • Centralizedmanagement of logistics • A huge problem with 47000 variables, 14300 of themarebinaryand 36000 constraints • Solutionsareapproximatedtoreducecomputation time (limitedwith 1800 seconds), upto %10 awayfrom optimum

  32. HealthcareResearch in Turkey • Yücel, E., Salman F. S., Örmeci E. L., Gel, E. S. , Gel A., “Logistics of ClinicalTesting: BicriteriaHeuristics for Routing and Scheduling of Specimen Collection,” 2011. (Research in Progress)

  33. ClinicalSpecimenCollection Problem (CSCP) • Bicriteria problem • Primary: Thenumber of specimencollected is maximizedthrough a MIP model constrainedby an upperbound • Secondary: Total transportationcost is minimizedwithMyopicTourBuildingHeuristicand Tabu SearchHeuristicconstrainedby a lowerbound • Bothlevelsare NP-hard

  34. Thankyouforlistening…

More Related