Artificial Intelligence Helps in Selecting Influenza Strains for Vaccine Production
Vaccination is one of the most important and effective preventive measures in the fight against seasonal epidemics caused by influenza viruses each year. For vaccines to work, they must match the currently circulating influenza strains. The timely selection of appropriate influenza strains, which will be the most widespread in the population in the upcoming period, and their yield are crucial for vaccine development. The new MAIVeSS (Machine-learning Assisted Influenza Vaccine Strain Selection) method, which uses artificial intelligence and machine learning algorithms for direct analysis of clinical samples, could help with this process.
In February 2024, a team from the Center for Influenza and Emerging Infectious Diseases (CIEID) at the University of Missouri published a study in the journal Nature Communications describing the MAIVeSS method, aimed at simplifying the selection of influenza strains. Its main advantages lie in machine learning algorithms that can quickly and accurately evaluate the molecular characteristics of antigenicity and yield of currently circulating influenza viruses, thus shortening the production time of seasonal influenza vaccines.
Mutations of Influenza Strains and Their Monitoring
The mutations of influenza viruses are primarily driven by the membrane protein hemagglutinin (HA), which contains the virus's main surface antigen. This undergoes various antigenic changes, allowing the virus to adapt to the host's immunity, which is often a result of responses to previous infections and vaccinations. Therefore, the composition of vaccines is annually adjusted to match the currently circulating virus strains.
The monitoring of influenza and its mutations is coordinated through the Global Influenza Surveillance and Response System (GISRS) of the World Health Organization (WHO). The GISRS comprises an international network of influenza laboratories from 129 WHO member states. These laboratories monitor influenza viruses, mutations, and pandemic risks year-round and assist in designing preventive measures. All verified data are immediately available to the public via the FluNet web application and reporting system.
Based on verified data, the GISRS system evaluates and recommends the most suitable strains for vaccine production. It provides its recommendations, needed data, and samples to pharmaceutical companies, which then optimize the antigenic and growth properties of the selected strains. The entire system is reliable but requires global coordination of the private and public sectors and is very costly and time-consuming.
Why Does Vaccine Production Take So Long?
For the effective development of a vaccine, it is also crucial that the chosen strain has sufficient yield. This is determined by the amount of virus that can be cultivated in tissue culture or on chicken embryos for the production of inactivated or attenuated vaccines. Optimizing the yield of chosen strains using biotechnology currently takes up to 6 months, which can sometimes lead to vaccine production delays.
An example is the 2009 pandemic caused by the A(H1N1)pdm09 strain. At that time, vaccine deliveries were significantly delayed due to the low yield of the initially chosen influenza strain. The global vaccination campaign thus started only after the second wave of the pandemic, when strains with sufficient yield were finally obtained.
The MAIVeSS Method Could Shorten Influenza Strain Selection
The MAIVeSS method, based on machine learning algorithms, can accurately predict the antigenic and growth phenotypes of a virus strain based on the hemagglutinin sequence. The algorithm was tested on a library of 189 mutant HA variants and its reliability was verified on a dataset of more than 11,000 sequences of the A(H1N1)pdm09 virus detected in patients from 2009–2020. Variants with good antigenicity and yield, which could serve as candidate viruses for vaccine production, were identified.
Data from initial experiments suggest that the MAIVeSS method allows for the rapid selection of antigenically matching influenza strains with high yield directly from clinical isolates. This could significantly accelerate the process of selecting suitable strains from months to days.
(jko)
Sources:
1. Gao C., Wen F., Guan M. et al. MAIVeSS: streamlined selection of antigenically matched, high-yield viruses for seasonal influenza vaccine production. Nat Commun 2024; 15 (1): 1128, doi: 10.1038/s41467-024-45145-x.
2. Global Influenza Surveillance and Response System (GISRS). World Health Organization, 2024. Available at: https://who.int/initiatives/global-influenza-surveillance-and-response-system
Did you like this article? Would you like to comment on it? Write to us. We are interested in your opinion. We will not publish it, but we will gladly answer you.
- Technology allows children in palliative care to escape from the hospital room for a while
- Artificial Intelligence Helps Dentists Find Cavities or Create Implants
- The Role of Inflammation in the Cardiovascular Continuum and the Aging Process
- Talking without vocal cords? A new AI-supported technology may help
- A device the size of a matchbox will help treat obstructive sleep apnea