Description
Event Date: Thursday, April 16, 2026
Time: 2:00pm–3:00pm
Speaker: Carlos M. Perez-Velez, MD
Location: The Lister Hill Center Auditorium (LHC, building 38A), and virtual via MS Teams
Abstract:
Tuberculosis (TB) in childhood remains a difficult infectious disease to diagnose because, in children with active disease—especially those with early non-severe disease—clinical manifestations are often nonspecific, radiologic abnormalities may be subtle, respiratory specimens are difficult to obtain, and the mycobacterial burden is often low, limiting the sensitivity of currently available microbiologic tests. Clinicians must therefore rely on multiple sources of evidence—clinical features, imaging, laboratory abnormalities, immunologic indicators, and exposure history—to guide treatment. These complexities make pediatric TB a valuable model for biomedical data science. Dr. Perez-Velez will discuss how AI may support clinicians across the pediatric TB diagnostic pathway, including recognizing findings suggestive of intrathoracic TB, estimating the likelihood that TB is the cause of disease, and assessing risk of progression to inform treatment decisions. He will emphasize AI’s potential to improve timely, clinically grounded decision-making when microbiologic confirmation is not possible. He will also highlight recent work in pediatric chest-radiography AI, illustrating both promise and limitations.
Speaker Bio:
Dr. Carlos M. Perez-Velez is an infectious diseases physician, hospital epidemiologist, and academic clinician whose work spans pediatric and adult tuberculosis, healthcare epidemiology, infection prevention and control, respiratory infectious diseases, and global health. He serves as Hospital Epidemiologist and Medical Director of the Infection Prevention and Control Program at the Raymond G. Murphy VA Medical Center in Albuquerque, New Mexico, and is Associate Professor of Clinical Medicine at both the University of New Mexico and the University of Arizona. Dr. Perez-Velez is internationally recognized for his work in pediatric and drug-resistant tuberculosis, particularly in the development of innovative specimen-collection methods, radiologic approaches, and artificial intelligence tools for diagnosis in children. He is Principal Investigator of an NIH-supported study developing AI-based methods for detection of pediatric tuberculosis on chest radiographs. He has published extensively in leading journals, including The New England Journal of Medicine, The Lancet, and JAMA Pediatrics, and has served as an invited speaker and consultant for NIH, WHO, CDC, and international tuberculosis networks.
The DIR invites speakers with expertise in bioinformatics and biomedical discovery, including computational biology research, computational health research, and related areas of interest, to present in our NLM Colloquia series. Visit our NLM DIR Lectures webpage or reach out to NLMColloquia@nih.gov to learn more.