Machine Learning Decodes Immune Responses in HIV, Paving Way for Personalized Medicine
New research led by York University is harnessing the power of machine learning to unravel the complexities of immune responses in individuals living with HIV, offering a promising path toward personalized vaccination and therapeutic strategies. The study, published as a pre-print in the journal Patterns and slated for print publication on March 13, 2026, demonstrates the potential to predict how individuals with compromised immune systems will respond to vaccines, accounting for factors like age, comorbidities and genetics.
Unlocking the Complexity of Immune Variability
Understanding how individuals with weakened immune systems respond to vaccination is a critical area of immunological research. Researchers analyzed data from individuals with and without HIV who received up to five doses of a COVID-19 vaccine over a 100-week period. All individuals living with HIV were from the Greater Toronto Area and managed their illness with antiretroviral therapy. The team employed a machine-learning method called random forest to analyze 64 immune biomarkers triggered by the COVID-19 vaccine, creating ‘virtual patients’ to model immune responses.
“This study constitutes an important step forward in the potential for personal vaccination intervention strategies,” says lead author Chapin Korosec, formerly a postdoctoral fellow at York University and now an adjunct professor with the University of Guelph.
Key Findings: Distinguishing Immune Signatures
The machine-learning model accurately differentiated between HIV-positive and HIV-negative participants with nearly 100% accuracy, revealing distinct immunogenic signatures. Researchers found that saliva-based antibodies, specifically IgA, combined with white blood cells, were key differentiators between the two groups. This is significant given existing research highlighting altered mucosal immunity in individuals living with HIV.
Interestingly, the study identified outliers within both groups. In a small subset of HIV-positive individuals, vaccine-induced immune responses were indistinguishable from those of HIV-negative individuals, suggesting their immune function had been effectively restored through treatment. Conversely, one healthy control participant exhibited immune markers similar to someone living with HIV, potentially indicating underlying, yet unidentified, immune issues.
The Role of Machine Learning and ‘Virtual Patients’
Professor Jane Heffernan, whose research at York University focuses on infectious disease modelling, emphasized the complexity of the immune response. “Sometimes something can act as an inhibitor of an arm of the immune response, but in other times it might be an activator. There is also a lot of individual variability among people with similar immune system status.” She explained that machine learning, combined with mechanistic modeling and the creation of ‘virtual patients,’ allows researchers to uncover important differences within subgroups and between individuals, even for immune components not directly measured.
Implications for Personalized Medicine
The findings underscore the need for tailored healthcare approaches, particularly for individuals with compromised immune systems. By learning the structure of immune variability, researchers are moving towards a data-driven foundation for personalized vaccination and therapeutic design.
The study was supported by the National Research Council of Canada (NRC)-Fields Mathematical Sciences Collaboration Centre, the National Sciences and Engineering and Research Council of Canada, and Artificial Intelligence for Public Health (AI4PH). Researchers from York University collaborated with colleagues from the University of Toronto, St. Michael’s Hospital, Pennsylvania State University, and the NRC Digital Technologies Research Centre.
Journal Reference: Korosec, C. S., et al. (2026). Modeling of longitudinal immune profiles reveals distinct immunogenic signatures following five COVID-19 vaccinations among people living with HIV. Patterns. DOI: 10.1016/j.patter.2025.101474.