SBIR Phase II: NLP Driven Automation for Optimizing New Patient Referral Pathways

About This Grant

The broader impact and commercial potential of this Small Business Innovation Research (SBIR) Phase II project lie in potentially improving the efficiency, accuracy, and equity of new patient referral triaging. Inefficient referral management leads to treatment delays, poor clinical utilization, and logistical bottlenecks— challenges that are problematic nationwide when transferring patient information across hospital systems. This project streamlines referral workflows, reducing administrative burdens and optimizing healthcare resource allocation. Beyond operational improvements, this innovation addresses incomplete work-ups, missing records, and inaccurately scheduled appointments. By analyzing referral inflows and outflows, the technology identifies inefficiencies, alerts coordinators, and helps ensure access to care. Commercially, this solution meets the growing demand for data-driven referral management in hospitals, clinics, and healthcare networks, helping institutions reduce costs when physicians are operating at the “top of their license” and improve patient outcomes. As healthcare systems shift toward value-based care, this project has the potential to become a scalable, industry-leading solution. By enhancing care coordination and accessibility in a highly fragmented healthcare system, this project advances both scientific and technological understanding while offering a commercially viable tool to reshape referral management nationwide. The proposed project addresses the need for a high-performing, cost-effective solution to triage new patient referrals. Based on pilot data from Johns Hopkins and UCSF, the project has demonstrated over 20% improvement in accuracy for triaging and provider assignment compared to the current standard of care at these institutions. Phase II will focus on expanding the accuracy and capabilities of its algorithms by incorporating over 10,000 data points from multiple institutions. This will enhance model performance and fairness and extend capabilities to high-revenue, time- sensitive subspecialties such as neurosurgery. To achieve these objectives, the project will employ federated learning, enabling multiple decentralized systems to collaboratively train a shared machine learning model while preserving data privacy. Additionally, seamless integration with electronic medical records will allow for automated tracking of operational impact and return on investment (ROI). These advancements will demonstrate measurable post-interventional benefits, ensuring high renewal rates and financial transparency for hospital stakeholders. The anticipated technical outcome is an advanced AI- driven algorithm capable of physician-level review of complex cancer referrals, surpassing the accuracy and efficiency of referral processes at leading national healthcare centers. 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.