Deadline: June 5, 2025
Notice Number: NOT-DA-26-005
Purpose
The proliferation of large-scale databases and biomedical repositories, accelerated by the transition to electronic health records and calls for increased data sharing, marks a significant shift in biomedical research. The new NIH policy on Data Management and Sharing will further amplify this trend, increasing the quality and accessibility of high-value datasets available for reuse and secondary analysis. Concomitantly, advances in artificial intelligence/machine learning algorithms (AI/ML) enable integration and analysis of vast, unstructured multimodal datasets. Computational modeling of complex relationships between small molecules, genes, proteins, neural circuits, and behaviors can be used to advance any stage of the medication development pipeline, including identification of medications suitable for repurposing, design and optimization of new chemical identities, design of regiments and prediction of side effects, to name a few. This data-driven technology has become ripe for full adoption in the development of novel treatments for substance use disorders (SUDs). For example, data from a genes-disease database, a genome encyclopedia, and a database of protein – chemical interactions have been integrated by a knowledge-driven AI-based system with more than 90 million electronic health records to identify ketamine, an FDA-approved medication, as a potential treatment for cocaine use disorder (PMID: 36792381).
The purpose of this Notice of Special Interest (NOSI) is to elicit projects that would leverage the power of generative AI/ML and predictive models to accelerate medication discovery and development for treatment of SUDs while reducing the risk of failure.
Areas of Interest:
- De novo design of new synthesizable molecules targeting addiction circuitry using generative AI tools, molecular dynamics simulations, and other advanced computational approaches
- Identification of medications that can be repurposed for treatment of SUDs by constructing models based on multiple datasets such as chemical, genetic, and disease databases, electronic health records, clinical trials and scientific literature
- Optimization of promising compounds for SUD treatment to enhance their selectivity, affinity, and efficacy at the target receptor using structural evolution or other AI/ML approaches
- AI-enabled virtual screening of medication candidates for those with high bioavailability, optimal pharmacokinetics, efficient blood brain barrier permeability, minimal toxicity, and desired behavioral responses
- Chemical synthesis design for the most effective synthetic route and ease of manufacturing by application of generative AI trained on databases of chemical reactions and structures
For more detailed information, please see the opportunity webpage.