Shahraki, Narges

Abstract Background: The dynamic nature of surgical demand poses significant challenges in planning and scheduling resources to staff the surgical suite. Uncertainty in workload often leads to conservative upstream resource allocation, resulting in staffing misallocation. We propose a novel metric, the daily maximum number of open operating rooms (ORs), as an interpretable resource demand proxy to support staff planning. Methods: Using historical surgical data from Mayo Clinic in Rochester, Minnesota, from May 5, 2018 through October 8, 2023 (332,904 surgical cases), we developed a deep learning framework based on a wide-and-deep architecture to predict the daily maximum number of open ORs. The model incorporates features capturing historic OR use trends, seasonality effects, and anesthesia division–specific characteristics. Predictions are generated for forecast horizons of up to 90 business days. Results: Our results demonstrate strong predictive performance, with campus-level root mean square error values ranging from 5 to 7 available ORs across all forecast horizons. Validation across 9 anesthesia divisions confirms the frame- work’s capability to capture complex, nonlinear dependencies within longitudinal surgical datasets. Conclusions: This study introduces a novel operational metric and a robust predictive framework for perioperative resource management. By accurately fore- casting daily maximum open ORs, health care managers can better align staff capacity with anticipated demand, potentially enhancing staff use and improving the patient care experience by minimizing delays.

Abstract Background Each training program has its own internal policies and restrictions, which must be considered while developing trainee schedules. Designing these schedules is complex and time consuming, and the final schedules often contain undesirable aspects for trainees. Objective We developed a decision-support system (DSS) to optimally schedule daily assignments and monthly rotations for trainees. The proposed DSS aims to 1) reduce the schedule development time, 2) maximize trainee preferences for desired rotations and vacation times, and 3) ensure adaptability of the DSS across multiple graduate medical programs through a flexible design and intuitive graphical user interface. Methods Using mixed-integer linear programming, we developed a scheduling model that 1) maximized trainees’ preferences on specific rotations and vacation times and 2) ensured fairness by assigning equal numbers of vacation days and a balanced schedule of difficult versus easy rotations among trainees. The model was successfully implemented in the Mayo Clinic Division of Pulmonary and Critical Care for the academic year 2018–2019. Results Using the DSS, it took only a few minutes to produce a schedule versus several days of preparation time required by the manual process. Compared with the manually developed schedule, the DSS schedule satisfied 11% more rotation preferences and improved fairness by 19%. All trainees met duty hours in the DSS schedule compared with 83% in the manually developed schedule. Conclusion The proposed DSS can dramatically reduce the schedule preparation time, accommodate more of trainees’ preferences, and improve fairness in assigning rotations.

Abstract Background We aimed to determine if patient symptoms and computed tomography enterography (CTE) and magnetic resonance enterography (MRE) imaging findings can be used to predict near-term risk of surgery in patients with small bowel Crohn’s disease (CD). Methods CD patients with small bowel strictures undergoing serial CTE or MRE were retrospectively identified. Strictures were defined by luminal narrowing, bowel wall thickening, and unequivocal proximal small bowel dilation. Harvey-Bradshaw index (HBI) was recorded. Stricture observations and measurements were performed on baseline CTE or MRE and compared to with prior and subsequent scans. Patients were divided into those who underwent surgery within 2 years and those who did not. LASSO (least absolute shrinkage and selection operator) regression models were trained and validated using 5-fold cross-validation. Results Eighty-five patients (43.7 ± 15.3 years of age at baseline scan, majority male [57.6%]) had 137 small bowel strictures. Surgery was performed in 26 patients within 2 years from baseline CTE or MRE. In univariate analysis of patients with prior exams, development of stricture on the baseline exam was associated with near-term surgery (P = .006). A mathematical model using baseline features predicting surgery within 2 years included an HBI of 5 to 7 (odds ratio [OR], 1.7 × 105; P = .057), an HBI of 8 to 16 (OR, 3.1 × 105; P = .054), anastomotic stricture (OR, 0.002; P = .091), bowel wall thickness (OR, 4.7; P = .064), penetrating behavior (OR, 3.1 × 103; P = .096), and newly developed stricture (OR: 7.2 × 107; P = .062). This model demonstrated sensitivity of 67% and specificity of 73% (area under the curve, 0.62). Conclusions CTE or MRE imaging findings in combination with HBI can potentially predict which patients will require surgery within 2 years.

Abstract Hospital census prediction has well-described implications for efficient hospital resource utilization, and recent issues with hospital crowding due to CoVID-19 have emphasized the importance of this task. Our team has been leading an institutional effort to develop machine-learning models that can predict hospital census 12 hours into the future. We describe our efforts at developing accurate empirical models for this task. Ultimately, with limited resources and time, we were able to develop simple yet useful models for 12-hour census prediction and design a dashboard application to display this output to our hospital’s decision-makers. Specifically, we found that linear models with ElasticNet regularization performed well for this task with relative 95% error of +/− 3.4% and that this work could be completed in approximately 7 months.