AI Can Predict Body's Response to Vaccines from Antibody Patterns in Blood
The condition of the immune system before vaccination can apparently provide clues about how strongly a person will respond to a vaccine. Researchers are now using artificial intelligence (AI) to read antibody patterns in the blood and estimate that response.
The findings come from a study by the SeroNet programme at the National Cancer Institute in the United States, published in the journal Cell Press Blue. The research was led by Joshua LaBaer and involved more than 4,000 people with a total of over 8,000 blood samples.
In the study, the researchers examined antibodies against 185 antigens derived from various viruses, bacteria, and targets related to autoimmune diseases. The study participants consisted of healthy people as well as groups with weaker immune systems, including patients with HIV, multiple myeloma, and organ transplant recipients.
All participants received the COVID-19 vaccine. Blood samples taken before and after vaccination were then analysed using AI to look for patterns associated with strong or weak antibody responses.
From that analysis, the researchers found a number of antibody patterns that could provide clues about a person’s response to a vaccine. One of them is called ‘sentinel antibodies’, which could potentially be used as an early indicator to estimate the immune system’s response before a vaccine is given.
The researchers also found a relationship between the levels of antibodies already present in the body and the response to a vaccine. Antibodies against Staphylococcus aureus, Respiratory Syncytial Virus (RSV), and human parainfluenza virus 3 (HPIV-3), for example, were found at higher levels in the group that showed a stronger response to the COVID-19 vaccine.
‘We identified a universal antimicrobial pattern that is positively associated with the 25 per cent of participants with the highest COVID-19 vaccine response,’ the researchers wrote.
Differences in vaccine response between individuals are not new. Age, sex, medical history, and genetic factors can all influence it. Immunosuppressive conditions can also cause the body to produce fewer antibodies after vaccination.
For this reason, the researchers used machine learning to develop a model that can identify people who are likely to have a suboptimal vaccine response.
LaBaer said that certain biomarkers analysed using AI could help estimate who is likely to respond well to a vaccine even before receiving an injection.
If these findings can be developed further, the information could potentially help doctors tailor vaccination strategies, such as determining dosing schedules, considering specific vaccine types, or providing booster doses to people who are predicted to have a lower immune response.
However, the AI in this study serves as a tool for reading patterns and making predictions based on biological characteristics, not for confirming an individual’s vaccination outcome.