Huang, Yu-Li

Background/Objectives: Appointment volume is a primary measure of provider workload and a key contributor to provider burnout in primary care settings. It is highly driven by the size and patient mix of patient panels assigned to providers. To ensure that workload is balanced among providers, an effective patient panel assignment strategy is essential. Methods: A patient panel assignment framework was developed and demonstrated using a year of data from a large family medicine practice to assign patients evenly among providers with the following steps: 1. Predicting patient appointment needs using patient characteristics and Adjusted Clinical Group (ACG) scores; 2. Placing patients in appointment groups based on predictive appointment needs; 3. Determining the patient cap of each appointment group for each provider group; 4. Allocating patients from each appointment group to each provider group without exceeding the patient cap. Results: The results were demonstrated across 100 simulated reassignments to a retrospective dataset. The variations in the number of appointments among providers within a provider group were compared between current practice and the proposed framework. For the largest provider group with 14 providers, the proposed framework with appointment prediction using all features reduced the standard deviation and range by 85% and 84%, respectively. While the proposed framework improved appointment workload balance among providers, 11.7% of patients required reassignment. Conclusions: The proposed framework demonstrates its utility in supporting workload balance. Future work is needed to minimize patient reassignments while accounting for patient care continuity and to incorporate other contributing factors into the framework.

Effective healthcare referral triage poses a significant challenge, especially extracting referral notes on time for scheduling. To offset this issue, this study aims to develop an optimized Random Forest classifier to recreate the referral notes content and identify patients requiring Spinal Cord Stimulation procedures. A performance index was introduced to improve accuracy, while a consistency index was designed to prevent overfitting. The optimized classifier, with the optimal selection at the number of estimates being 5 trees and tree depth being 4, achieved an accuracy of 0.74 with a lower standard deviation, demonstrating its reliability and effectiveness.

Goal: The purpose of the research is to explore, through the lens of organizational performance and staff satisfaction, the characteristics of administrative leaders working as dyad partners with physician leaders. Methods: All 54 administrative leaders from 71 clinical departments at the three US Mayo Clinic sites were invited to participate in the study. We used an unsupervised cluster analysis machine learning method to group the leaders based on their characteristics, as measured by the 32-dimension Occupational Personality Questionnaire (OPQ 32r), and we used a three-cluster model to explore the relationships between the clusters and the performance outcome. We took the department performance data from the previous year and compared the percentage of departments with the upward changes among the clusters. For staff survey data, we calculated the percentage of departments with scores that were above average among the three clusters for both physician and administrative staff responses. Principal Findings: Analysis of personality data revealed three different clusters. Cluster 1 leaders were caring and democratic, forward-thinking, strategic, optimistic, and trusting of others. Cluster 2 leaders were extremely hardworking and authoritative. Cluster 3 leaders were caring, modest, and rule-following. Cluster 1 leaders showed the best financial performance and sense of belonging among their followers, cluster 2 leaders elicited high engagement from their departments, and cluster 3 leaders encouraged lower burnout among staff members. Practical Applications: From this study, we obtained empirical evidence of administrative leaders’ characteristics that showed positive relationships with financial and staff-satisfaction metrics. The results showed that distinct types of leaders influence administrative staff and physician staff differently and that different situations require different styles of leadership. We can also conclude that implementing robust, scientifically validated tools to assess leadership traits and tendencies can positively affect leadership and organizational performance for healthcare organizations.

An algorithm for the two-stage robust optimization surgery-to-operating room allocation problem is presented. The second-stage problem is an integer linear program whose convex hull is approximated using three types of specialized valid inequalities and Chvátal-Gomory cuts. The resulting linear relaxation of the second-stage problem is then dualized and integrated into the first-stage problem. The resulting mixed integer linear program, which is an approximation of the original problem, is then solved using a commercial solver. If the solution of this model is not optimal for the second-stage problem, valid inequalities for the second-stage problem are generated, yielding a type of column-generation based approach that we refer to as the sequential follower refinement (SFR) algorithm. Data from an academic medical center are used to compare the computational performance of SFR with the constraint and column generation (C&CG) algorithm, which is the only exact approach that has been specifically applied for this problem in the literature. An extensive numerical study of SFR and its computational characteristics is presented that shows that SFR yields better-quality solutions compared with C&CG, even as the termination criterion of SFR is met much sooner, especially for problems involving higher number of surgeries.History: Accepted by Paul Brooks, Area Editor for Applications in Biology, Medicine, & Healthcare.Supplemental Material: The software that supports the findings of this study is available within the paper and its Supplemental Information ( https://pubsonline.informs.org/doi/suppl/10.1287/ijoc.2022.0191 ) as well as from the IJOC GitHub software repository ( https://github.com/INFORMSJoC/2022.0191 ). The complete IJOC Software and Data Repository is available at https://informsjoc.github.io/ .