Biotechnology HPC Software Applications Institute (BHSAI)

BHSAI Uses Artificial Intelligence (AI) Methods to Assess Vaccine-Induced Immune Responses

March 31, 2019   |  Download PDF

Vaccines are an essential component of Force Health Protection, particularly for Service members deployed overseas where they may be exposed to infectious diseases, such as malaria and dengue fever, or newly emergent outbreaks, such as Ebola or Zika virus. The Walter Reed Army Institute of Research (WRAIR) and the United States Army Research Institute of Infectious Diseases (USAMRIID) have long formed the backbone of the U.S. Army’s vaccine development capabilities, carrying out pre-clinical studies and clinical trials that measure vaccine-induced immune responses and assess vaccine efficacy.

Recent advances in immune assay technologies, from next-generation sequencing to multiplex flow cytometry, enable clinicians and immunologists at WRAIR and USAMRIID to measure highly complex immune responses in vaccine clinical trials at an unprecedented level of detail. However, the large data sets produced by these technologies pose significant challenges of their own: How do we translate this new ‘big data’ in immunology into actionable information on how vaccines work or fail? Scientists from TATRC’s Biotechnology High Performance Computing Software Applications Institute (BHSAI), led by Dr. Jaques Reifman, recently tackled this problem with collaborators at WRAIR. Together they sought to determine whether Artificial Intelligence (AI)-based models could rapidly process immunological data from clinical studies to identify what combination of immune responses is responsible for protection. Isolating these protective responses, also known as correlates of protection, can help researchers improve existing vaccines or design new ones with increased efficacy.


During a vaccine clinical trial, immunological data are collected at several time points following vaccination, prior to infection. Data from each subject are then fed into an AI-based model to make an individualized prediction of protection, which is compared with the clinically observed protection status of that subject. An accurate predictive model can be used to determine what combination of immune responses is most responsible for protection, providing a basis for guiding future vaccine design efforts.During a vaccine clinical trial, immunological data are collected at several time points following vaccination, prior to infection. Data from each subject are then fed into an AI-based model to make an individualized prediction of protection, which is compared with the clinically observed protection status of that subject. An accurate predictive model can be used to determine what combination of immune responses is most responsible for protection, providing a basis for guiding future vaccine design efforts.


Under the Military Infectious Diseases Research Program, Dr. Sid Chaudhury, a staff scientist at BHSAI, developed a computational method to integrate large immune data sets that capture antibody, cellular, and cytokine responses across a wide range of assays, and then applied AI to make individualized predictions of protection from the immune data. At an individual level, this AI-based model enables scientists to determine why the vaccination succeeded or failed to achieve protection for a given subject in a clinical trial. At a study level, this AI-based model can help determine what combination of immune factors is necessary to achieve protection.

Dr. Chaudhury worked with Dr. Elke Bergmann-Leitner, Chief of the Flow Cytometry Center at WRAIR, to analyze data collected at WRAIR on a recent vaccine study in which non-human primates were immunized with a new Army-developed malaria vaccine candidate–the Self Assembling Nano Particle–using different adjuvants. (Adjuvants act as powerful immunostimulators to modify and enhance vaccine-induced immune responses.) In research recently published in the journal ‘Scientific Reports,’ Dr. Chaudhury used AI to determine how these different adjuvants alter vaccine-induced immunity, providing a basis for rationally selecting adjuvants to maximize vaccine efficacy.

By bringing together the capabilities of complex immunoprofiling at WRAIR’s Flow Cytometry Center and AI-based modeling at BHSAI, Dr. Chaudhury and Dr. Bergmann-Leitner have identified potential correlates of protection in clinical trials carried out at WRAIR and at the Naval Medical Research Center for a range of malaria vaccines, including the RTS,S (Mosquirix™, GSK Inc.), PfSPZ (Sanaria Inc.), irradiated sporozoite (U.S. Navy), and FMP2.1 (U.S. Army) vaccines.

Dr. Sid Chaudhury stated, “The ability of AI to integrate the large immunological data sets and provide key biological insights underscores its potential to revolutionize infectious disease research, from vaccine design to disease surveillance.”


This article was published in the July 2019 issue of the TATRC Times.


Links