Project Grant R01MH143269

The National Institute of Mental Health awarded Duke University $1.221 million under the Mental Health Research Grants program (CFDA 93.242) on September 1, 2026, to develop an AI-driven multimodal model for HIV pre-exposure prophylaxis (PrEP) candidacy assessment in women in the South, where 52% of new HIV infections occur nationally but PrEP use remains lower than in other regions and women account for a disproportionate share of new infections.

The award funds development of a clinically facing PrEP Candidacy Assessment Tool (PCAT) that integrates multimodal data from unstructured clinical notes and structured electronic health record variables with neighborhood-level social drivers of health data to improve identification of women at elevated risk of HIV acquisition. Existing HIV risk prediction models have failed to accurately identify at-risk women because they rely on structured data and do not capture hidden narratives about risk. The funded approach incorporates knowledge graphs informed by women living with HIV, HIV-focused community advocates, and providers to add relational context to determinants of HIV risk and enhance model explainability. Aim 1 develops a community-informed knowledge graph through iterative engagement with affected populations and providers.

The project period extends through June 30, 2031, with work performed in Durham, North Carolina. Duke University's Office of Research Administration is the recipient organization.