Presentation on theme: "Outpatient Clinical Scheduling"— Presentation transcript:
1Outpatient Clinical Scheduling Research TeamMark Lawley, Principal InvestigatorKumar Muthuraman, University of TexasLaura Sands, Purdue School of NursingDeDe Willis, MD, Indiana University School of MedicineAyten Turkcan, Research Scientist, PurduePo-Ching DeLaurentis, Research Assistant, PurdueRebeca Sandino, Research Assistant, PurdueJi Lin, Research Assistant, PurdueSantanu Chakraborty, Research Assistant, PurdueJoanne Daggy, Research Assistant, PurdueBo Zeng, Post-doc, PurdueFunding: National Science Foundation, $460K, Regenstrief Foundation $395K
2Partnering Clinics Wishard Health Services Cottage Corner Health Center (low income)North Arlington Health Center (low income)Community Physicians of IndianaGiest Family Medicine and Pediatrics (mid. class)
3Project thrust Study and improve internal clinic operations Develop new scheduling theory that accounts for environmental complexitiesSequential schedulingPatient no-showGeneral service time distributionsImplement in real systems and validate impact
4Outpatient Clinical Scheduling In the US, almost 90% of patient care provided in the approx. 200,000 non-psychiatric outpatient clinicsPressures for improving clinic operationsAging populationIncreased chronic diseaseHospitals to reduce LOSImproved patient serviceAccessOutcomesSatisfactionRevenue / ReimbursementNew modes of care
5Why is out-patient scheduling complex? Emphasis on patient satisfaction (low waiting time)Emphasis on staff and physician utilization (low idle time)High patient no-show, cancellation, walk-inTardy arrivals (patients and physicians)Stochastic, patient dependent service timesSequential schedule constructionOn-call physiciansPhysician constraintsMany others …
6Patient no-show Ubiquitous problem in clinical operations Can be 40-50% for some types of clinicsApproximately 20% for our partnersCan be modeled and used in schedulingNo show prob. can be estimated usingpatient history, diagnosis, demographics, medicationslead time to appointment,exogonous factors such as weather, public transp.A patient’s no show probability should not be used to predict whether a given patient will arriveThe no show probability of a group of patients should be used to evaluate the no-show characteristics of a given schedule
7Sequential Scheduling Process Patient calls clinic for appointment with physicianScheduler looks at the current schedule, negotiates with patient, adds the patient to a “slot” (we would add estimate no-show prob.)Couple of days in advance, clinic might call to remind the patientPatient is expected to, but might not, arrive at appointed time.Schedules are built incrementally, patient by patient.Information used is current schedule (plus no-show prob.)No opportunity to “optimally” schedule final set of patients.How can we create good sequential schedule that takes patient no-show into account?
8Slot Model I slots in a consultation day J patient types, pj probability of patient no-showXi denotes the number of patients arriving at beginning of slot iYi number of patients overflowing out of slot iLi number of patients served in slot i, initially assumed PoissonR(Sn) overflow probability matrixQ(Sn) arrival probability matrix
9Slot Model Objective max E[ r i Xi - c i Yi - C YI ]
13Charles Joseph Minard (1781 – 1870), a French civil engineer noted for his inventions in the field of information graphics.
14General Service TimesOverflow implies patient in service overflowing from one slot to next.Must include time in service in previous slotDistribution of Li takes more general form that requires numerical integration
16Non-myopic approaches for sequential scheduling Optimal Sequential Schedule: Dynamic ProgrammingAdd simple forecasting to the previous assignment algorithm
17Improvement over myopic up to 12% Small System with 4 slots and 2 patient types
18Next StepsContinue clinic process mapping, operational data collection, simulation – seeking opportunities to improveMake suggestions to improve clinic operational efficiency, help implementContinue no-show modeling effortsContinue developing sequential patient scheduling theory and algorithmsBegin working with scheduling software vendors
19Submitted and Working papers PublicationsMuthuraman, K., Lawley, M. A stochastic overbooking model for outpatient clinical scheduling with no-shows. To appear in IIE Transactions Special Issue on HealthcareSubmitted and Working papersChakraborty, S., Muthuraman, K., Lawley, M. Sequential clinical scheduling with general service times and no-show patients, Operations Research.Zeng, B., Turkcan, A., Lin, J., Lawley, M., Clinic scheduling models with overbooking for patients with heterogeneous no-show probabilities, Annals of Operations Research.Turkcan, A., Zeng, Muthuraman, K., Lawley, M., Sequential clinical scheduling with moment-based constraints, in preparation.Daggy, J., Sands, L., Lawley, M., Willis, D. The impact of no-show probability estimation on clinic schedules, in preparation.Lin, J., Muthuraman, K., Lawley, M. An Approximate Dynamic Programming Approach to Sequential Clinical Scheduling, in preparation.19