PFI-TT: Developing an Artificial Intelligence Solution for Accurate Medical Coding to Improve Healthcare Billing

Award ID: 2431840

The broader impact of this Partnerships for Innovation - Technology Translation (PFI-TT) project lies in overcoming the limitations of automated biomedical content annotation systems, which include bias, errors, and a lack of adaptability to new information. By automating and optimizing the medical coding assignment process, the project will alleviate administrative and cognitive burdens on healthcare professionals, allowing them to focus more time and attention on patient care. The system will promote transparency, accountability, and trustworthiness through a sociotechnical-based approach, fostering greater trust and confidence in Artificial Intelligence (AI) driven systems within the healthcare industry. This innovative system will improve healthcare delivery by minimizing coding errors and ensuring precise reimbursement claims, leading to better patient outcomes. This project has the potential to advance scientific understanding of AI applications in healthcare and position itself as a leader in the growing healthcare technology market, fostering broader acceptance and adoption of AI-driven solutions. This project is at the intersection of AI, medical informatics, and socio-technical systems. The project will focus on creating robust algorithms for context building from electronic medical records (EMRs) and a novel AI framework to enhance accuracy. A key objective is the seamless integration of a socio-technical feedback loop that incorporates input from healthcare professionals to refine and optimize the system's recommendations. The research and development will involve the integration of heterogeneous EMR data into a comprehensive knowledge graph and the development of an AI framework that iteratively refines code recommendations through a dynamic retriever-generator interaction. The anticipated technical results include a highly accurate and trustworthy medical coding recommendation system. This use-inspired research aims to leverage the collective intelligence of humans and machines, ultimately improving the performance and reliability of automated medical coding. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Program Director: Mara E. Schindelholz