The "Multimodal Artificial Intelligence to Accelerate HIV Clinical Care" funding opportunity, offered by the National Institutes of Mental Health (NIMH) and the National Institute on Drug Abuse (NIDA), seeks to support innovative research leveraging multimodal artificial intelligence (AI) models to advance HIV diagnosis, prevention, and treatment. The initiative encourages applications focused on developing, adapting, and evaluating multimodal AI models integrating diverse data modalities and sources, including text, images, and audio. Researchers are tasked with enhancing the interpretability and explainability of these models through the integration of HIV-specific knowledge graphs and addressing ethical considerations such as data privacy, transparency, and security. The initiative requires a multidisciplinary team approach and active stakeholder engagement to ensure that the models are human-centered, fair, and unbiased.

The program focuses on three key objectives: advancing multimodal AI models for diverse HIV-related applications, developing knowledge graphs to enhance model interpretability, and synergistically integrating these graphs into AI frameworks. Proposals must demonstrate the ability to impact areas such as early HIV detection, personalized prevention and treatment plans, epidemic surveillance, and clinical care optimization. The research also aims to improve AI models' domain-specific explainability, allowing end-users, such as clinicians and patients, to better understand and trust the outputs. Additionally, NIDA emphasizes applications addressing substance use disorders co-occurring with HIV to enhance clinical care and treatment outcomes for this population.

The funding opportunity provides up to $750,000 in direct costs annually, with a project period of up to five years. NIMH intends to commit $2 million for two awards, while NIDA allocates $2 million for three to four awards. Eligible applicants include public and private higher education institutions, nonprofits, small businesses, and local or state governments. Applications must include a Plan for Enhancing Diverse Perspectives (PEDP), which outlines strategies to incorporate inclusivity into the research. Failure to include a PEDP will result in administrative withdrawal of the application.

Applicants must submit a detailed plan for meaningful stakeholder engagement, specifying how community members, public health leaders, patients, and other relevant stakeholders will be integrated into model design, testing, and implementation. Applications must also describe metrics for measuring stakeholder engagement's impact. Additional required documents include data management and sharing plans consistent with NIH policies. Submissions must address the ethical framework guiding data use and participant consent processes.

Applications are due by March 27, 2025, and must be submitted via Grants.gov. Letters of intent, though not mandatory, are encouraged and due by February 27, 2025. The review process evaluates applications based on scientific and technical merit, rigor, feasibility, innovation, and alignment with program priorities. Proposals must demonstrate how multimodal AI offers a superior approach compared to traditional methods in addressing HIV-related challenges. Awardees must comply with annual reporting requirements, data-sharing mandates, and NIH policies regarding data protection and transparency.

NIMH and NIDA will host a pre-application webinar to provide guidance to prospective applicants and answer questions about the program's scope and application requirements. Interested applicants are encouraged to consult the agency contacts listed in the funding announcement for scientific or administrative inquiries. The initiative aligns with national goals to end the HIV epidemic and advance the use of AI technologies in healthcare innovation.