Online Archive of University of Virginia Scholarship
Leveraging Agentic Reasoning and Networks to Advance Conversational Data Interfaces In Context of Flood Risk Management Systems; Unraveling the Sociotechnical Roots of Inefficiency in Prior Authorization161 views
Author
Pasupuleti, Varun, School of Engineering and Applied Science, University of Virginia0009-0008-5210-7324
Advisors
Seabrook, Bryn, EN-Engineering and Society, University of Virginia
Nguyen, Rich, EN-Comp Science Dept, University of Virginia
Abstract
This research portfolio encompasses the exploration of two distinct projects that address key challenges in environmental and healthcare systems, respectively. These projects leverage both technical capabilities and sociotechnical analysis to drive meaningful change.
The technical capstone project focuses on Floodwatch, a predictive platform for flood risk management. It improves upon an AI-powered conversational assistant by utilizing agentic reasoning and networks to improve data accessibility and interaction quality. By integrating multi-agent systems, memory-enhanced architectures, and personalized responses, the assistant demonstrates significant advancements in handling complex queries and adapting to user preferences. Additionally, a user study revealed positive feedback from participants, underscoring the potential of intelligent conversational interfaces for improved data accessibility across a variety of sectors.
The sociotechnical research uncovers inefficiencies in the prior authorization (PA) process within healthcare systems, analyzing organizational tensions, misaligned incentives, and technological inadequacies through Actor-Network Theory and Technological Determinism frameworks. The study identifies barriers such as outdated technologies and fragmented provider-payer relationships. To improve efficiency and foster trust, the study proposes strategies like automated PA systems and mechanisms for transparent collaboration between payers and providers. These findings emphasize the need for both systemic changes and new technologies to improve healthcare outcomes.
Together, these projects show how combining technical innovation with sociotechnical insight can solve complex problems in critical systems. This approach provides a practical path to improving environmental disaster management and healthcare systems.
Pasupuleti, Varun. Leveraging Agentic Reasoning and Networks to Advance Conversational Data Interfaces In Context of Flood Risk Management Systems; Unraveling the Sociotechnical Roots of Inefficiency in Prior Authorization. University of Virginia, School of Engineering and Applied Science, BS (Bachelor of Science), 2024-12-31, https://doi.org/10.18130/dj2m-pa10.
