Abstract
Using data from cardiovascular surgery patients with long and highly variable post-surgical lengths of stay (LOS), we develop a modeling framework to reduce recovery unit congestion. We estimate the LOS and its probability distribution using machine learning models, schedule procedures on a rolling basis using a variety of optimization models, and estimate performance with simulation. The machine learning models achieved only modest LOS prediction accuracy, despite access to a very rich set of patient characteristics. Compared to the current paper-based system used in the hospital, most optimization models failed to reduce congestion without increasing wait times for surgery. A conservative stochastic optimization with sufficient sampling to capture the long tail of the LOS distribution outperformed the current manual process and other stochastic and robust optimization approaches. These results highlight the perils of using oversimplified distributional models of LOS for scheduling procedures and the importance of using optimization methods well-suited to dealing with long-tailed behavior.
Similar content being viewed by others
Notes
Surgeries that took longer than 15 hours in reality were treated as 15 hours for feasibility.
Alternatively, we also estimated each patients’ surgical duration using the average surgical duration for each procedure type in the empirical data. This approach produces similar results to that using the actual surgical durations.
Note that we still enforce the constraint that each patient is scheduled no later than \(d^{max}_p\) through \(Q^{op}\).
86 out of the 596 surgeries performed from September 2018 to March 2020 are associated with patient arrivals prior to September 2018. For these patients, the earliest possible surgery dates are adjusted so that their procedures are always scheduled on or after September 1st 2018. This modified constraint is an artifact of the fixed optimization time window and is tighter than what is realistically feasible. The first six months is treated as the warm up period for this reason.
In practice, prediction errors of the test set will not be available at the time of scheduling. The test errors are included in constructing \(\mathcal {S}\) so that, if the LOS of an incoming patient in the test set surpasses all cases in the training set, conditional sampling described in the previous section (see Step 2(b) of Method 1 still works as intended. Although this is not ideal, the scheduling process should not benefit much from it and it helps simplify our simulations. In practice, if the LOS of a patient surpasses all previous cases, one approach is to consult health providers.
References
Beliën J, Demeulemeester E, Cardoen B (2009) A decision support system for cyclic master surgery scheduling with multiple objectives. J Sched 12(2):147–161
Benzaid M, Lahrichi N, Rousseau LM (2020) Chemotherapy appointment scheduling and daily outpatient-nurse assignment. Health Care Manag Sci 23:34–50
Bertsekas DP, Castanon DA (1999) Rollout algorithms for stochastic scheduling problems. J Heuristics 5:89–108
Bertsimas D, Sim M (2004) The price of robustness. Oper Res 52(1):35–53
Chan CW, Farias VF, Bambos N, Escobar GJ (2012) Optimizing intensive care unit discharge decisions with patient readmissions. Oper Res 60(6):1323–1341
Chen E, Naylor CD (1994) Variation in hospital length of stay for acute myocardial infarction in ontario, canada. Med Care 32(5):420–435
Davison AC, Hinkley DV (1997) Bootstrap methods and their application. Cambridge University Press
Ettema RG, Peelen LM, Schuurmans MJ, Nierich AP, Kalkman CJ, Moons KG (2010) Prediction models for prolonged intensive care unit stay after cardiac surgery: systematic review and validation study. Circulation 122(7):682–689
Fairley M, Scheinker D, Brandeau ML (2019) Improving the efficiency of the operating room environment with an optimization and machine learning model. Health Care Manag Sci 22(4):756–767
Fügener A, Hans EW, Kolisch R, Kortbeek N, Vanberkel PT (2014) Master surgery scheduling with consideration of multiple downstream units. Eur J Oper Res 239(1):227–236
Guinet A, Chaabane S (2003) Operating theatre planning. Int J Prod Econ 85(1):69–81
Gurobi Optimization LLC (2021) Gurobi Optimizer reference manual. URL http://www.gurobi.com
Hsu VN, De Matta R, Lee CY (2003) Scheduling patients in an ambulatory surgical center. Nav Res Logist 50(3):218–238
Keyvanshokooh E, Shi C, Van Oyen MP (2021) Online advance scheduling with overtime: A primaldual approach. Manuf Serv Oper Manag 23(1):246–266
Keyvanshokooh E, Kazemian P, Fattahi M, Van Oyen MP (2022) Coordinated and prioritybased surgical care: An integrated distributionally robust stochastic optimization approach. Prod Oper Manag 31(4):1510–1535
Kramer AA, Zimmerman JE (2010) A predictive model for the early identification of patients at risk for a prolonged intensive care unit length of stay. BMC Med Inform Decis Mak 10(1):1–16
LaFaro RJ, Pothula S, Kubal KP, Inchiosa ME, Pothula VM, Yuan SC, Maerz DA, Montes L, Oleszkiewicz SM, Yusupov A et al (2015) Neural network prediction of icu length of stay following cardiac surgery based on pre-incision variables. PLoS One 10(12):e0145395
LeDell E, Poirier S (2020) H2O AutoML: Scalable automatic machine learning. ICML Workshop on Automated Machine Learning
Maharlou H, Kalhori SRN, Shahbazi S, Ravangard R (2018) Predicting length of stay in intensive care units after cardiac surgery: comparison of artificial neural networks and adaptive neuro-fuzzy system. Healthc Inform Res 24(2):109
Min D, Yih Y (2010) Scheduling elective surgery under uncertainty and downstream capacity constraints. Eur J Oper Res 206(3):642–652
Nassar AP Jr, Caruso P (2016) Icu physicians are unable to accurately predict length of stay at admission: a prospective study. Int J Q Health Care 28(1):99–103
Neyshabouri S, Berg BP (2017) Two-stage robust optimization approach to elective surgery and downstream capacity planning. Eur J Oper Res 260(1):21–40
Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, Blondel M, Prettenhofer P, Weiss R, Dubourg V et al (2011) Scikit-learn: Machine learning in Python. J Mach Lear Res 12:2825–2830
Pham DN, Klinkert A (2008) Surgical case scheduling as a generalized job shop scheduling problem. Eur J Oper Res 185(3):1011–1025
Rahimi I, Gandomi AH (2021) A comprehensive review and analysis of operating room and surgery scheduling. Arch Comput Methods Eng 28(3):1667–1688
Rohleder TR, Klassen KJ (2002) Rolling horizon appointment scheduling: a simulation study. Health care Manag Sci 5(3)
Shehadeh KS, Padman R (2021) A distributionally robust optimization approach for stochastic elective surgery scheduling with limited intensive care unit capacity. Eur J Oper Res 290(3):901–913
Tabbutt S, Schuette J, Zhang W, Alten J, Donohue J, Gaynor JW, Ghanayem N, Jacobs J, Pasquali SK, Thiagarajan R et al (2019) A novel model demonstrates variation in risk adjusted mortality across pediatric cardiac intensive care units after surgery. Pediatr Crit Care Med: J Soc Crit Care Med World Federation Pediatr Intensive Crit Care Soc 20(2):136
Tsai PFJ, Chen PC, Chen YY, Song HY, Lin HM, Lin FM, Huang QP (2016) Length of hospital stay prediction at the admission stage for cardiology patients using artificial neural network. J Healthc Eng 2016
Tsai PFJ, Chen PC, Chen YY, Song HY, Lin HM, Lin FM, Huang QP (2016) Length of hospital stay prediction at the admission stage for cardiology patients using artificial neural network. J Healthc Eng 2016
Vicente FG, Lomar FP, Mélot C, Vincent JL (2004) Can the experienced icu physician predict icu length of stay and outcome better than less experienced colleagues? Intensive Care Med 30(4):655–659
Weissman GE, Hubbard RA, Ungar LH, Harhay MO, Greene CS, Himes BE, Halpern SD (2018) Inclusion of unstructured clinical text improves early prediction of death or prolonged icu stay. Crit Care Med 46(7):1125–1132
Whellan DJ, Zhao X, Hernandez AF, Liang L, Peterson ED, Bhatt DL, Heidenreich PA, Schwamm LH, Fonarow GC (2011) Predictors of hospital length of stay in heart failure: findings from get with the guidelines. J Card Fail 17(8):649–656
Yang CS, Wei CP, Yuan CC, Schoung JY (2010) Predicting the length of hospital stay of burn patients: Comparisons of prediction accuracy among different clinical stages. Decis Support Syst 50(1):325–335
Zhang J, Dridi M, El Moudni A (2019) A two-level optimization model for elective surgery scheduling with downstream capacity constraints. Eur J Oper Res 276(2):602–613
Zhu S, Fan W, Yang S, Pei J, Pardalos PM (2019) Operating room planning and surgical case scheduling: a review of literature. J Comb Optim 37(3):757–805
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Competing Interests
The authors have no competing interests to declare that are relevant to this paper. J. Blanchet acknowledges support from NSF grant 2118199.
Ethical Approval
This research uses secondary data collected from standard clinical practice with minimal risks. Ethical approval was not needed. Anonymity and confidentiality of information were assured.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Yuan Shi and Saied Mahdian contributed equally to this work.
Appendices
Appendix A: Mathematical formulations for \(f(u_d),Q^{op}\)
Solving the quadratic objective, \(f(u_d) = u_d^2\) of the mixed-integer program can be slow. To reduce the runtime required to solve the problem, we implement the quadratic term in the objective function, \(f(u_d) = u_d\), using a piece-wise linear approximation,
and add additional constraints
In words, \(u_d^{(m)}\) counts ICU overflow above \(c+m-1\) for \(m=1,\ldots ,5\). Here, \(e_m\) are constant coefficients for the piecewise linear function.
In our formulation, we set \(e_1 = 1, e_2=e_3=e_4=e_5=2\). Since \(u_d\) only takes integer values, this coefficient choice leads to \(f(u_d)\equiv u_d^2\) for any \(u_d\le 5\). The piece-wise linear approximation is 20 times faster than the quadratic form in our numerical experiments.
The above approximation is used in all our algorithms. The formulations for stochastic and robust algorithms are analogous (e.g. copies of \(u_{d}^{(m)}\) are created for different LOS realizations), and we thus omit the details here.
Next, we introduce the full mathematical formulation of \(Q^{op}\) in constraint 1e for offline optimization below.
Sets
-
\(D = \{1,2,\ldots ,N_d\}\): index for days
-
\(P = \{1,2,\ldots ,N_p\}\): index for patients
-
\(P^{par} \subset P\): set of PAR patients
-
K: set of surgeons
Parameters
-
c: number of ICU beds reserved for elective patients, set to 8
-
\(q_p\): operation duration for \(p \in P\).
-
\(l_p\): post-op length of stay in CVICU for \(p \in P\)
-
\(h_{d,k}\): number of hours surgeon k is available to perform surgery on day d, for \(s \in S\). \(h_{d,k} = 15\) if k operates on day d, 0 otherwise.
-
\(par\_day_{d,k}\): indicator variable for PAR days. \(par\_day_{d,k} = 1\) if surgeon k can perform PAR surgeries on day d, 0 otherwise.
-
\(original\_date_{p} \in D\): the original date that a surgery is scheduled for for \(p \in P\)
-
\(actual\_date_{p} \in D\): the actual date that a surgery takes place for \(p \in P\); can be different from \(original\_date_{p}\) if the surgery was rescheduled.
-
\(arrived\_on_{p}\): the arrival date for \(p \in P\), i.e., when a patient is first being scheduled for surgery; can be negative (i.e. outside the set D) if the surgery was scheduled before September 2018
-
\(lead_{p}\): the lead time of a surgery, i.e.
$$\begin{aligned} original\_date_{p}- arrived\_on_{p} \end{aligned}$$ -
\(m_{k,p}\): binary, 1 if patient p is assigned to surgeon k and 0 otherwise
Decision Variables
-
\(u_d\): integer, the number of additional elective patients in the ICU on day d above c
-
\( x_{d,p}\): binary, \( x_{d,p} = 1\) if patient p is scheduled to have her surgery on day d; otherwise 0
-
\(d_p\): the date when the surgery of patient p is scheduled by the model
-
\(y_{d,p}\): binary, \(y_{d,p} = 1\) if patient p stays in the CVICU on day d; otherwise 0
-
\(z_{d,p}\): continuous, number of hours that the surgery of p lasts on day d. \(z_{d,p} = q_p\) if \(x_{d,p} = 1\), otherwise 0
Indicator Function
-
Assumed feasible window where patient p is available for surgery:
$$\begin{aligned} g(d,p) = \mathbbm {1}\{d^{min}_p \le d<d^{max}_p\} \end{aligned}$$where \(d^{max}_p = \text {min}(actual\_date_p+90,N_d)\),
$$\begin{aligned} d^{min}_p = {\left\{ \begin{array}{ll} \text {max}[1,original\_date_p- \frac{lead_p}{2}], &{} lead_p> 20\\ \text {max}[1,arrived\_on_{p}],&{}lead_p\le 20 \end{array}\right. } \end{aligned}$$(11)
Next we describe the optimization constraints 12-16 that are equivalent to \(x\in Q^{op}.\)
Constraint 12 ensures each patient is scheduled exactly once within that patient’s window of availability.
Constraints 13-16 incorporate daily availability of surgeons. Constraint Eq. 13 includes number of hours a surgeon is available on each day for (non PAR) surgeries.
Constraint 15 captures that PAR surgeries can only be done on pre-specified days.
Constraint 16 ensures each surgeon is not scheduled for PAR surgeries in subsequent days.
1.1 Appendix A.1: \(Q^{op}\) for batch optimization problem
When solving BOP for rolling scheduling, \(Q^{op}\) is adjusted accordingly. In addition to the Sets and Parameters specified in Section A, we use the following sets and parameters.
Sets
-
\(P_b \subset P\): index of batch patients of period b
-
\(P^{past}_b = \bigcup _{k=1}^{b-1} P_k\): set of patients prior to period b
-
\(P_b^{par} = P^{par} \cap P_b\)
Parameters
-
\(s_b\): start date/day of batch b
-
\(d^{min}_{p,b} = \max (d^{min}_p,s_b)\)
-
\(x^*_{d,p}\): solutions obtained from previous periods
Since the definition of \(d^{min}_{p,b}\) may restrict the original time window of availability for some patients, we also adjust the last available date, \(d^{\max }_p\), to at least 90 days after the start of the period, i.e.,
This adjustment allows flexible scheduling as described in Section 3.1. In practice, the definition of \(d^{max}_{p,b}\) will not include \(N_d\); we included it for our simulation runs. Although the latter adjustment could potentially increase wait time, it is necessary in ensuring that the set of feasible scheduling solutions is not too restricted, and any resultant increase in wait time will be penalized by the objective function.
Indicator Function
-
Adjusted feasibility window for each patient.
Constraints 18-22 give the equivalent formulation of \(x\in Q^{op}\) in all deterministic, stochastic and robust BOPs.
Appendix B: Solving AC &CG for robust BOP
In the following, we introduce the AC &CG algorithm used for solving the robust BOP.
1. Main problem. We start with a subset of traces in the uncertainty set, \(\Omega ^t = \{\omega ^{(n)}:n= 1,\ldots ,|\Omega ^t|\}\subseteq \mathcal {U}_b\). Instead of minimizing the worst-case cost over the entire uncertainty set \(\mathcal {U}\), the Main problem of AC &CG minimizes the worst-case cost over the subset, \(\Omega ^t\):
s.t.
Constraint 23g sets the variable \(\theta \) to be equal to the maximum cost of overflow among all traces of LOS realization, \(\omega ^{(n)}\). The objective 23a thus minimizes the weighted sum of total wait time and the maximum cost of overflow for all \(\omega ^{(n)}\in \Omega ^t\). The remaining constraints mirror those for BSOP-SAA in optimization problem 6, only replacing the set of sampled traces with set \(\Omega ^t\). The optimal objective value of the Main problem provides a lower bound to the optimal objective value of robust BSOP in optimization problem 7.
2. Recourse problem. At each iteration t, after the Main problem is solved with \(\Omega ^t\), an optimal solution \(x^{t*}\) is obtained. The Recourse problem aims to find a trace \(\omega ^{t*}\in \mathcal {U}_b\) that maximizes the cost of ICU overflow under the given schedule \(x^{t*}\). Since \(x^{t*}\) and \(\omega ^{t*}\) are feasible under the original BOP, solving the Recourse problem finds an upper bound of the objective value of robust BOP.
s.t.
This formulation follows from expressions 7e-7h, and constraints 24e and 24f use the definition of \(\mathcal {U}_b\) in expressions 8 and 9.
One key difficulty of solving the Recourse problem in its current form is that the decision variable \(l_p\) appears in the boundary of the summation in constraint Eq. 24b. We follow the reformulation approach in [22] and extend it to convex, piece-wise linear forms of \(f(u_d)\). The reformulated Recourse problem is provided in Appendix B.1. We present the full AC &CG algorithm in Algorithm 5.
1.1 Appendix B.1: Solving the recourse problem
The Recourse problem, denoted as \(Q(x^{t*})\), involves constraints including \(l_p\) decision variables in the boundary of summations. Here, we introduce our MIP reformulation that is readily solvable by Gurobi. We refer readers to [22] for more details and proofs of its validity.
As in [22] we define the variables \(v_{d,p},w_{d,p} \in \{ 0,1\}\) for \(d \in D, p \in P\). The variable \(v_{d,p}\) is 1 only if patient p is admitted in the ICU by d. Given the temporary solution \(x^{t*}\), \(v_{d,p}\) are constant parameters determined by \(x^{t*}\). The decision variable \(w_{d,p}\) is 1 only if patient p leaves the ICU by day d. So, \(y_{d,p} = v_{d,p} - w_{d,p}\).
The inner minimization problem of \(Q(x^{t*})\) is
where \(e_1 = 1, e_2=e_3=e_4=e_5 = 2\).
We apply strong duality to reformulate the inner minimization as a maximization problem and also substitute for the definition of uncertainty set \(\mathcal {U}_b\). Let \(d_p\) denote the date of scheduled procedure for patient p according to \(x^{t*}\), i.e., \(d_p = \sum _{d\in D} d \cdot x^{t*}_{d,p}\). \(Q(x^{t*})\) is reformulated below with decision variables \(\lambda _d^{(m)}\) and \(w_{d,p}\).
For the optimal solution, we must have \(\lambda ^{(m)}_d\in \{0,e_m\}\).
The formulation above involves a bilinear term, \(w_{d,p}\lambda ^{(m)}_d\). Since \(w_{d,p} \in \{0,1\}\), we can reformulate the problem by using \(q^{(m)}_{d,p} = w_{d,p}\lambda ^{(m)}_d\) for all d, p, m. The final reformulation of the Recourse problem is the following.
Appendix C: Additional numerical results
This section presents additional results on the optimality gap of AC &CG and biweekly scheduling.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Shi, Y., Mahdian, S., Blanchet, J. et al. Surgical scheduling via optimization and machine learning with long-tailed data. Health Care Manag Sci 26, 692–718 (2023). https://doi.org/10.1007/s10729-023-09649-0
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10729-023-09649-0