Although seasonal influenza disease spread is a spatio-temporal phenomenon, public surveillance systems aggregate data only spatially, and are rarely predictive. We develop a hierarchical clustering-based machine learning tool to anticipate flu spread patterns based on historical spatio-temporal flu activity, where we use historical influenza-related emergency department records as a proxy for flu prevalence. This analysis replaces conventional geographical hospital clustering with clusters based on both spatial and temporal distance between hospital flu peaks to generate a network illustrating whether flu spreads between pairs of clusters (direction) and how long that spread takes (magnitude). To overcome data sparsity, we take a model-free approach, treating hospital clusters as a fully-connected network, where arcs indicate flu transmission. We perform predictive analysis on the clusters' time series of flu ED visits to determine direction and magnitude of flu travel. Detection of recurrent spatio-temporal patterns may help policymakers and hospitals better prepare for outbreaks. We apply this tool to Ontario, Canada using a five-year historical dataset of daily flu-related ED visits, and find that in addition to expected flu spread between major cities/airport regions, we were able to illuminate previously unsuspected patterns of flu spread between non-major cities, providing new insights for public health officials. We showed that while a spatial clustering outperforms a temporal clustering in terms of the direction of the spread (81% spatial v. 71% temporal), the opposite is true in terms of the magnitude of the time lag (20% spatial v. 70% temporal).

The distributed operating room (OR) scheduling problem aims to find an assignment of surgeries to ORs across collaborating hospitals that share their waiting lists and ORs. We propose a stochastic extension of this problem where surgery durations are considered to be uncertain. In order to obtain solutions for the challenging stochastic model, we use sample average approximation and develop two enhanced decomposition frameworks that use logic-based Benders (LBBD) optimality cuts and binary decision diagram based Benders cuts. Specifically, to the best of our knowledge, deriving LBBD optimality cuts in a stochastic programming context is new to the literature. Our computational experiments on a hospital data set illustrate that the stochastic formulation generates robust schedules and that our algorithms improve the computational efficiency. Summary of Contribution: We propose a new model for an important problem in healthcare scheduling, namely, stochastic distributed operating room scheduling, which is inspired by a current practice in Toronto, Ontario, Canada. We develop two decomposition methods that are computationally faster than solving the model directly via a state-of-the-art solver. We present both some theoretical results for our algorithms and numerical results for the evaluation of the model and algorithms. Compared with its deterministic counterpart in the literature, our model shows improvement in relevant evaluation metrics for the underlying scheduling problem. In addition, our algorithms exploit the structure of the model and improve its solvability. Those algorithms also have the potential to be used to tackle other planning and scheduling problems with a similar structure.

Operating rooms (ORs) play a substantial role in hospital profitability, and their optimal utilization is conducive to containing the cost of surgical service delivery, shortening surgical patient wait times, and increasing patient admissions. We extend the OR planning and scheduling problem from a single independent hospital to a coalition of multiple hospitals in a strategic network, where a pool of patients, surgeons, and ORs are collaboratively planned. To solve the resulting mixed-integer dual resource constrained model, we develop a novel logic-based Benders’ decomposition approach that employs an allocation master problem, sequencing sub-problems for each hospital-day, and novel multistrategy Benders’ feasibility and optimality cuts. We investigate various patient-to-surgeon allocation flexibilities, as well as the impact of surgeon schedule tightness. Using real data obtained from the General Surgery Departments of the University Health Network (UHN) hospitals, consisting of Toronto General Hospital, Toronto Western Hospital, and Princess Margret Cancer Centre in Toronto, Ontario, Canada (who already engage in some collaborative resource sharing), we find that on average, collaborative OR scheduling with traditional patient-to-surgeon allocation flexibility results in 6% cost-savings, while flexible patient-to-surgeon allocation flexibility increases cost-savings to 40%, and surgeon schedule tightness can impact costs by 15%. The collective impact of our collaboration and patient flexibility results in between 45% and 63% savings per surgery. We also use a game theoretic approach to fairly redistribute the payoff acquired from a coalition of hospitals and to empirically show coalitional stability among hospitals. Data and the online supplement are available at https://doi.org/10.1287/ijoc.2017.0745 .

Patients undergoing a bone marrow stem cell transplant (BMT) face various risk factors. Analyzing data from past transplants could enhance the understanding of the factors influencing success. Records up to 120 measurements per transplant procedure from 1751 patients undergoing BMT were collected (Shariati Hospital). Collaborative filtering techniques allowed the processing of highly sparse records with 22.3% missing values. Ten-fold cross-validation was used to evaluate the performance of various classification algorithms trained on predicting the survival status. Modest accuracy levels were obtained in predicting the survival status (AUC = 0.69). More importantly, however, operations that had the highest chances of success were shown to be identifiable with high accuracy, e.g., 92% or 97% when identifying 74 or 31 recipients, respectively. Identifying the patients with the highest chances of survival has direct application in the prioritization of resources and in donor matching. For patients where high-confidence prediction is not achieved, assigning a probability to their survival odds has potential applications in probabilistic decision support systems and in combination with other sources of information.

We view the beam orientation optimization (BOO) problem in intensity-modulated radiation therapy (IMRT) treatment planning as a global optimization problem with expensive objective function evaluations. We propose a response surface method that, in contrast with other approaches, allows for the generation of problem data only for promising beam orientations as the algorithm progresses. This enables the consideration of additional degrees of freedom in the treatment delivery, i.e., many more candidate beam orientations than is possible with existing approaches to BOO. This ability allows us to include noncoplanar beams and consider the question of whether or not noncoplanar beams can provide significant improvement in treatment plan quality. We also show empirically that using our approach, we can generate clinically acceptable treatment plans that require fewer beams than are used in current practice.