Abstract
Surgery is a cornerstone of modern healthcare, with the operating room (OR) as one of the most resource-intensive units within a hospital. Since surgeries account for a substantial portion of hospital revenue, accurate prediction of surgery durations and efficient OR scheduling are critical to minimizing delays for patients as well as reducing overtime for staff. However, OR scheduling is complex due to variability, including patient history, unforeseen complications, and emergency cases that can disrupt planned schedules. This complexity is compounded by the current OR scheduling approach, which remains a highly manual process driven largely by human decision-making and crude statistical averages. The goal of this project is to develop a model that improves elective surgery duration predictions by identifying patterns from historical data from the UVA Department of Surgery. Given the large number of unique surgeries, we clustered the data to gain greater insights into underlying patterns. Using this processed data, we developed XGBoost and Random Forest models that identify the most influential variables affecting surgery duration and improve time estimates for future procedures. Preliminary models show a 10.65-minute improvement in root-mean-square error for time prediction, a 14% decrease from the current prediction method.
Electronic health records (EHRs) have improved care coordination and enhanced information sharing across the health care system in the United States. Despite these benefits, designers have largely based these systems on the resources and infrastructure of large, urban hospitals systems, which has created barriers for rural providers. As a result, rural health care organizations have adopted EHR systems more slowly and have struggled to fully utilize them. This paper applies the Social Construction of Technology (SCOT) framework to analyze how key actors shape the implementation of EHR technology. It focuses on the roles of the federal government, private companies such as Epic Systems, and non-profit organizations such as OCHIN. It also examines Epic Community Connect and OCHIN as case studies to show how organizations have modified institutional strategies to better align EHR systems with rural contexts. The findings show that stakeholders have primarily driven adaptation through changes in governance and delivery models rather than through fundamental redesigns of the technology itself. These approaches have increased EHR adoption in rural areas, but they also introduce new trade offs related to autonomy, coordination complexity, and control over systems. The study concludes that successful health technology implementation depends not only on innovation but on alignment with the social and economic realities of its users. Achieving equitable access to digital health technologies will require continued coordination across stakeholders and policy approaches that are tailored to the unique constraints of rural health care systems.
Both papers examine approaches to improving the health care system in the United States, but they do so from complementary perspectives that highlight the interaction between technical design and real world implementation. The technical paper focuses on optimizing efficiency with hospitals systems, using machine learning to improve surgical scheduling and resource allocation. In contrast, the STS paper analyzes how broader sociotechnical factors such as stakeholder incentives and organizational structures, shape the adoption and effectiveness of EHR technology, particularly in underserved rural settings. Together, these papers demonstrate that technical improvements alone are insufficient to drive meaningful change in health care systems. While new technology and innovation can benefit health care systems, the success of these innovations depends on the alignment with institutional constraints, resource availability, and user needs. The two papers emphasize that effective health care innovation requires integrating technical optimization with an understanding of the social and organizational contexts in which these systems operate.
