SBIR Phase II: AI-Driven Visualization of Rehabilitation Documentation to Support Decision-Making Across Care Settings

The broader impact /commercial potential of this Small Business Innovation Research (SBIR) Phase II project is to improve rehabilitation progress understanding and use. This project will advance a technology that converts rehabilitation documentation into clear, visual representations of patient functional ability over time, enabling a shared, accessible understanding of progress for all stakeholders. By converting unstructured clinical narratives into visually accessible information, the technology can enhance interpretation, reduce ambiguity, and support more consistent, data-informed decision-making. The initial market segment includes acute care and inpatient rehabilitation settings, with planned expansion to outpatient rehabilitation and home health. The proposed commercialization model is a business-to-business software licensing model sold per healthcare facility. This Small Business Innovation Research (SBIR) Phase II project addresses the technical challenge of extracting, structuring, and interpreting rehabilitation progress data from highly variable, unstructured clinical documentation in operational healthcare. Building on Phase I results, the project will refine and validate a rehabilitation-specific data framework that visually represents patient functional ability in speech, occupational and physical therapy over time and across care settings. Phase II research and development objectives include expanding structured data representations of rehabilitation goals and progress indicators, validating measurement and aggregation methods across larger, varied datasets, and advancing visualization techniques that accurately reflect functional change, stability, and variability. The proposed research will employ natural language processing, small language models, ontology-based structures and expert-in-the-loop validation to transform narrative rehabilitation notes into standardized, analyzable data elements. Methods will include iterative refinement, performance evaluation using real-world datasets and actuarial principles, and feasibility testing within clinical workflows. Anticipated technical outcomes include improved accuracy and consistency in representing patient functional ability, validated visual outputs that support clinical and discharge decision-making, and a scalable technical foundation for future predictive modeling. Phase II results are expected to advance scientific understanding of how rehabilitation data can be systematically structured and applied to improve care coordination and outcomes. 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.