The Bukhari Lab at St. John's University has secured $562,000 from the National Institutes of Health to use artificial intelligence for identifying existing drugs that could treat serious respiratory diseases. The three-year grant, starting July 2026, will fund a project that combines machine learning with biomedical knowledge to surface drug candidates and, critically, to explain the scientific reasoning behind each suggestion.
How the system builds a chain of evidence
The initiative, called RESPOND: Repurposing Drugs for Respiratory Disorders Using a Hybrid Neuro-Symbolic AI Approach, does not just generate predictions. It maps relationships among drugs, diseases, genes, and biological pathways, then presents the evidence trail. "Biomedical researchers need more than a prediction," said Syed Ahmad Chan Bukhari, Ph.D., Associate Professor and director of the lab. "They need to understand the evidence behind that prediction, how different pieces of scientific knowledge are connected, and why a particular candidate should be examined more closely. Trust, transparency, and scientific reasoning are central to the RESPOND project."
Developing a new medicine from scratch can take years of lab work, clinical testing, and regulatory review. Drug repurposing starts with compounds already approved for one condition and investigates whether they might work for another. GLP-1 receptor agonists, for example, were first developed for Type 2 diabetes and later found to produce weight loss, opening a new treatment path for obesity. AI can accelerate this kind of search by narrowing a vast pharmaceutical landscape to a shortlist of candidates that researchers can then test.
The project draws on a hybrid approach, blending machine learning with structured biomedical knowledge. This combination helps the system not only spot patterns but also ground its output in established science. The work aligns with a broader push in AI for Science & Research, where the goal is not just faster computation but more reasoned, auditable results.
Students will work alongside the research team
The NIH funding, distributed annually through June 2029, will support machine-learning engineers as well as undergraduate and graduate students from Computer Science, Pharmacy, and related health disciplines. They will contribute to model development, biomedical data analysis, computational drug discovery, validation studies with collaborators, and scientific communication. "Students will receive firsthand experience at the intersection of artificial intelligence, computer science, pharmacy, and biomedical research, while contributing directly to an NIH-funded effort," Bukhari said.
Luca Iandoli, Ph.D., Dean of the Collins College of Professional Studies, said the grant "demonstrates the importance of interdisciplinary research in addressing complex health-care challenges." He added that the work reflects the university's commitment to research that combines AI, biomedical science, and service to society.
A lab focused on explainable AI for medicine
The RESPOND project is part of the Bukhari Lab's larger effort to build AI systems for healthcare that are accurate, auditable, and grounded in reliable evidence. In 2024, the lab received a $550,000 grant from the U.S. National Science Foundation to develop an AI solution for medical coding and healthcare billing, another area where AI for Healthcare can reduce administrative burden while demanding high accuracy.
Bukhari emphasized that the value of AI in this domain goes beyond processing power. "Its real value comes from helping scientists ask better questions, connect evidence more effectively, and make more informed research decisions," he said. "We hope RESPOND will contribute to a faster, more transparent, and more thoughtful approach to discovering new treatment possibilities."
Why this matters for science and research
For researchers in drug discovery and biomedical informatics, the RESPOND project signals a shift toward AI systems that prioritize explainability alongside performance. The funding validates the idea that scientific AI tools must do more than output a ranked list-they must show their work. As grant agencies invest in neuro-symbolic methods, professionals who can bridge machine learning and domain knowledge will become increasingly central to translational research. The project also creates a pipeline for students to gain direct experience on NIH-funded work, training the next cohort of researchers to build AI that scientists can interrogate, not just trust.
Your membership also unlocks:
