Artificial HIV patients do their part for science

Vanderbilt University Medical Center (VUMC) researchers are using generative artificial intelligence (AI) techniques to create hundreds of thousands of simulated HIV patients to broadly aid and stimulate longitudinal observational studies of this chronic condition.

The project is supported by a five-year, $4 million grant from the National Institutes of Health (NIH). Synthetic data created by the VUMC team will be made public, available to researchers everywhere.

“We think simulated data can greatly benefit HIV research, particularly in international settings where data sharing is becoming more complicated,” said one of two principal investigators for the grant, HIV researcher Bryan Shepherd, PhD, professor of Biostatistics and Biomedical Informatics. “The rate of discovery is being impeded by sensitivities around HIV and legitimate privacy concerns.”

The project’s other principal investigator is Bradley Malin, PhD, Accenture Professor of Biomedical Informatics and Computer Science, who is an expert on computational aspects of patient privacy.

“Sharing observational datasets is essential to enabling new hypotheses leading to potential cures,” Malin said. “Findings from our lab and others point to patient simulation as an alternative to typical patient de-identification methods that reduce data fidelity and research utility.”

The VUMC team assembled by Shepherd and Malin has expertise in HIV, biostatistics, AI, epidemiology, bioethics, and social science. They will attempt to mimic patients from two large multinational HIV research cohorts in South America and East Africa, numbering more than 500,000 people living with HIV. Evaluation will include replication of research done on the real cohorts. The project includes development of an ethics framework for the responsible use of synthetic data, creation of novel use cases, and development of open-source software toolkits to allow others to generate synthetic HIV cohort data.

“This project aims not only to accelerate HIV research,” Shepherd said, “but to advance data science methods, aid study reproducibility more generally, and further understanding of the uses of synthetic data.”

Vanderbilt co-investigators for the project include Jessica Castilho, MD, MPH, Stephany Duda, PhD, Jessica Perkins, PhD, Chao Yan, PhD, and Peter Rebeiro, PhD. They are joined by researchers at Indiana University, ETH Zurich, Moi University College of Health Sciences, Kenya, Instituto Nacional de Ciencias Médicas y Nutrición, Mexico, Fundação Oswaldo Cruz, Brazil, and Makerere University, Uganda.

The project is supported by NIH award R01MH139379.