Search:

A summary of presentations from the weekly Summit partner webinars

March 12, 2026 – The latest Summit Summary

AI for Strain Selection and the Importance of Vaccine Match for Vaccine Effectiveness – Colin A. Russell, Professor, Ph.D., Department of Medical Microbiology, Amsterdam University Medical Center, University of Amsterdam

Colin Russell, Ph.D., described proposed efforts to use AI to improve seasonal influenza vaccine strain selection and current understanding about the relationship between vaccine match and effectiveness.

There is not always a strong match between the circulating influenza strain and the vaccine, and the quality of the match is often not known until after the virus circulates. Decades of influenza virus mapping demonstrates that vaccines continue to provide protection when the circulating antigens vary slightly, but new vaccine formulations are needed when significant antigenic changes occur. The World Health Organization (WHO) uses global data to recommend which circulating strains to include in the vaccine for the upcoming influenza season. The recommendations are made many months in advance to allow time for manufacturers to develop the vaccines. Questions have arisen about whether modern computing technology can better assess the data informing strain selection or provide information to delay the selection until closer to the anticipated start of the influenza season, so that a closer match can be incorporated in vaccines.

Several studies have demonstrated that computer models using genetic sequencing data can outperform WHO in terms of predicting the likely circulating antigen for the coming season. However, WHO recommendations take into account much more information and much more complex data, such as human serology data. A study that, on the surface, suggested that an AI-driven model performed better than WHO in selecting a strain was flawed; ultimately, the AI technology contributed no new information. Dr. Russell pointed out that there are currently no good data sets on which to train AI to make better selections. The judgement of WHO experts is still better than that of technology alone.

Notably, even in seasons when the match is poor, populations experience different levels of effectiveness, some lower and some higher than the average. One study suggested that when the vaccine match was poor, people who were never previously vaccinated experienced higher effectiveness, and previously vaccinated people experienced lower effectiveness. The findings reveal that, while vaccine match is important, it is probably not the key factor driving effectiveness. This contention was supported by recent evidence that although this year’s influenza vaccine did not include subclade K, effectiveness against subclade K (as compared to unvaccinated persons) was about 40%.

QUESTIONS & ANSWERS

Q: Regarding evolutionary changes: do you pretty much see that as going in one direction?
Colin Russell (University of Amsterdam): Yes, there is a real directionality to evolution, where evolution basically is always moving away from some point in the past, right? Because if the virus were to evolve back towards things that had circulated previously, it’s at a profound disadvantage in terms of escaping immunity in the population. However, one of the things that the map I showed early on hides is that the hemagglutination inhibition (HI) assay, which underpins those antigenic measurements that inform that map, only really captures a very small subset of the influenza hemagglutinin (HA) protein. And because of that, there can be a huge amount of diversity that gets collapsed down into this very simple two-dimensional map that’s being missed because we’re focusing on measuring antigenicity by HI. When we get into the human serology data, the sort of much more realistic population, that diversity goes up massively, which really undermines our ability to make real predictions about where the virus is likely to go next.

Q: How much is HI testing still being used to make that strain selection? And also, what impact might there be from the Centers for Disease Control and Prevention (CDC) or the United States no longer being part of WHO, in terms of the quantity of virus isolates available, or the HI data, or other parts of those huge packages [of data] that go into strain selection by WHO? Are we still using HI tables?
Colin Russell (University of Amsterdam): So, HI is still routinely used as part of the vaccine strain selection process. However, where HI was usually the first point of investigation into looking at a virus’s phenotype, or its evolution, it’s now typically secondary to genetic sequencing. This is particularly true at the Francis Crick Institute in London and at CDC in the United States, where there’s a sequence-first protocol. They sequence everything that comes in, because sequencing is cheap, and it’s not particularly labor-intensive. HI, by comparison, is relatively expensive, and it’s relatively labor-intensive. And so, it’s still an important component of the antigenic characterization, but it’s not as widely practiced as it was 20 years ago.

Now, when we look at the situation globally, in terms of CDC, its withdrawal from WHO, its uncertain participation in vaccine strain selection meetings moving forward, and how that change in data availability potentially undermines the landscape—we could spend a whole hour talking about that. But in the spirit of being brief, from a global perspective, CDC had an important role to play, in large part because, well, America’s an incredibly well-resourced country, but it also contributed valuably to the surveillance mission in, I think, roughly 40 other countries around the world, many of them low-income settings, where that surveillance is now absent. We are, unquestionably, worse off without U.S. CDC’s participation in this process. That said, I would argue that that value is potentially greater in the context of pandemic preparedness, where we need eyes in as many places as possible, rather than in seasonal influenza vaccine strain selection. Because seasonal influenza viruses move globally all the time, as long as all of the other collaborating centers maintain their current roles, we can probably do a similarly good job of selecting vaccine strains without the protection and participation of U.S. CDC. But I think for pandemic preparedness, we’re appreciably worse off.

BACK TO TOP

2025-26 Respiratory Season Surveillance Impact – Caitlin Newhouse, MD, MPH, Medical Director, Vaccine-Preventable Diseases and Immunization Program, Tennessee Department of Health; Member, Council of State and Territorial Epidemiologists (CSTE)

Caitlin Newhouse, MD, MPH, provided recent findings from national surveillance of the 2025-26 respiratory disease season.

VIEW SLIDES

Across the United States, respiratory viral illness is generally low at present, with a few states experiencing a moderate number of cases and one state with a high number of cases. Data from wastewater nationally indicate that COVID-19 activity is low, respiratory syncytial virus (RSV) is moderate, and influenza A is very low.

RSV

  • RSV is elevated and increasing nationally. RSV activity began rising in November, later than typical.
  • RSV vaccination is given seasonally because of its short half-life. Typically, infants receive RSV vaccine from October through March. Maternal RSV vaccine is given from September through January and recommended from 32 to 36 weeks of gestation, so there is some residual protection of newborns that allows for flexibility in the timing of the infant dose.
  • With the late start of the RSV season, it is possible that RSV activity will continue into April in some areas, so some jurisdictions are considering recommending infant immunization beyond March.

Influenza

  • Influenza remains high nationally, but may have reached its peak recently.
  • Overall, the season has been designated as moderately severe, based on outpatient visits, hospitalizations, and deaths.
  • At present, severity is high for the pediatric population and moderate for others. Severity assessments are revised each week, which affects the overall severity rating.
  • Influenza A subtype dominated across the country; influenza B subtype varied by region.
  • Influenza A H3N2 viruses are the most frequently reported so far. A total of 92% of H3N2 samples tested belonged to subclade K, a new variant this year that emerged in August and spread rapidly around the globe.
    • The 2025-26 vaccine included subclade J but not subclade K, resulting in a mismatch.
    • WHO recently recommended updating all the strains for the 2026-27 vaccine, and the Food and Drug Administration is discussing those recommendations now.
    • This year’s vaccine has been estimated to be 40% effective (see earlier presentation by Dr. Russell). The CDC estimates that even with a poor match, influenza vaccination prevents thousands of hospitalizations and deaths.

COVID-19

  • COVID-19 activity hit a peak in August 2025 and increased again in December but has been trending downward since the end of January.
  • The XFG and NB 1.8.1 variants are the most prevalent types circulating in the United States now, estimated to account for 29% and 21% of cases, respectively.
  • One early study suggests 57% vaccine effectiveness.

Impact of Respiratory Viral Illness

  • Similar to the 2024-25 season, pediatric populations account for the most emergency and urgent care visits, but older adults have a higher proportion of hospitalization. However, adult hospitalization is falling while infant hospitalizations remain high, likely because of the late start and continued spread of RSV this year.
    • Influenza: Older adults have the highest numbers of hospitalization for influenza this year.
    • RSV: Infants have the highest numbers of hospitalization for this year, and it is too early to know whether these hospitalizations have peaked.
    • COVID-19: This year, for the second time since COVID-19 emerged, influenza accounted for more deaths than COVID-19. COVID-19 deaths have decreased dramatically since the first 2 years of the pandemic.
  • To date, 90 influenza-related pediatric deaths have been reported for this influenza season. Of those deaths, 85% occurred in children who were not vaccinated. About half of the deaths occurred in children with no underlying medical conditions.

QUESTIONS & ANSWERS

Q: Who is monitoring for human metapneumovirus, and what should people know about it?
Caitlin Newhouse (CSTE): That’s a great question. Human metapneumovirus is not generally a reportable condition, so we do not track it here in Tennessee. It probably can get picked up in some of our emergency department surveillance systems, where we look at, like, emergency department discharge diagnosis codes, but since it’s not a reportable disease, that kind of lives in a slightly different surveillance space.
Human Metapneumovirus Surveillance (CDC): https://www.cdc.gov/nrevss/php/dashboard/index.html

Q: Where can people go to see what proportion of states are including human metapneumovirus in some of their wastewater surveillance?
Caitlin Newhouse (CSTE): That’s a great question. There’s a good wastewater surveillance site.

Q: What proportion of pediatric influenza deaths were children who died at home, or is that reported yet? [Those cases] maybe could have been mitigated by antiviral treatment. There’s been a lot of hospitalization data showing where we could do better as a country on antiviral treatment, for sure. But are there any more recent data on the proportion of kids dying at home? Those are absolutely tragic deaths, … which makes the recent U.S. Department of Health and Human Services (HHS) changes and the vaccine recommendation just heartbreaking.
Caitlin Newhouse (CSTE): That’s a good question. There is a part of the CDC pediatric death surveillance page where you can filter [data] by location of death. … It looks like in our most recent season, 60% of the children were in the hospital, 20% in the emergency department, and about 20% outside of the hospital. So that could either be home or anywhere outside of the hospital, obviously. Pediatric Flu-Related Mortality Data: https://gis.cdc.gov/GRASP/Fluview/PedFluDeath.html

Q: One of the tricky things about the recommendations for the use of the RSV monoclonals is the seasonality and the deferral to the states for making a decision about extending the season—you know, what clinicians should be doing. We have seen that the National Center for Immunization and Respiratory Diseases has let the states know that there’s a late RSV season, and your state should be considering whether or not to recommend extending the vaccination [period]. What’s unclear to me is how do we direct clinicians to find out what their state is recommending? Because we obviously don’t want everyone just contacting their state immunization programs. Do you have any insights on what we should be advising when we’re writing, say, an IZ Express article to clinicians?
Caitlin Newhouse (CSTE): That’s a great question. The slide that I showed when I was discussing the monoclonal antibodies was actually shared [by] CDC in the email that they sent out to the jurisdictions with that alert that RSV activity is still rising and that jurisdictions should consider this. Here in Tennessee, what we did is look at what our activity is like based on our emergency department discharge data and the local RSV activity that we could find. We could see that, similarly to the others in HHS Region 4, we may have peaked and are starting to go down, and so we will probably not be issuing a recommendation to extend the administration of RSV monoclonal antibody. And to answer your question, that means we probably won’t say anything. I think in many places, if the recommendation isn’t going to change, the health department is probably not going to issue any sort of messaging, although maybe we need to think about that and sort of issue a recommendation that just says, “We don’t recommend you extending administration.” In the past, when we have issued recommendations to change the timing of administration—a couple years ago, we did recommend that providers start administering nirsevimab in September, because we were anticipating an early season—we sent out a health alert to the clinicians who are licensed in the state. So, we did direct-to-provider communication. States may issue a press release or a health alert to reach out to providers. I think it’s always absolutely fine to recommend that providers reach out to their health department. We’re always happy to answer questions, especially if they’re not sure about recommendations like this, because there’s not always an easy place to go, and sometimes the messages that we send out do miss people. But it’s a great question. I don’t think that it’s standardized from one jurisdiction to the next, exactly how they’re getting that messaging out.

Q: I would encourage thinking about proactively communicating to providers across the state, one way or the other, just so they know, so they’re not guessing. Because a lot of those questions come to us as, well, just, “What are we supposed to do?” So that kind of anticipatory guidance: “You should still be doing this through the month of March, but we’re not necessarily recommending changing that, and here’s why,” would be fabulous. I know [the Summit] is an adult-focused program, but I think it does highlight a larger issue that this RSV seasonality is rather fluid, and we need to come up with a better way to inform those taking care of infants how they’d need to do that, with a consistent, standardized process nationwide. Maybe that’s a CSTE leadership area as well.
Caitlin Newhouse (CSTE): Thanks for that question. Absolutely.

Q: It’s not just the monoclonals; when we are seeing these more delayed seasons, how can we think about, extending maternal immunization?
Kelly Moore (Immunize.org): I will say that CDC, as Dr. Newhouse said, specifically said, “Don’t change the maternal vaccination [timing].” So, we need to have clearer guidance, because they were saying, “Do not change maternal vaccination, only change the monoclonals,” but they only told that to the states, and the clinicians haven’t gotten really anything, unless their state sent it.

BACK TO TOP

Announcements
  • There will be no weekly meeting on Thursday, March 19, 2026 (because of the rescheduled Advisory Committee on Immunization Practices meeting, even if that meeting is postponed).
  • Registration is open for the 2026 National Adult and Influenza Immunization Summit, May 19-21, 2026, at the Crowne Plaza Atlanta Perimeter at Ravinia, 4355 Ashford Dunwoody Rd, Atlanta, GA 30346 (https://www.izsummitpartners.org/2026-naiis/).
  • The Summit includes a poster session for scientific abstracts. Posters can be submitted at https://www.izsummitpartners.org/2026-naiis-poster-form/. Attendees interested must submit their abstract for consideration by March 20, 2026. Submitters will be notified if their poster is accepted by April 3, 2026.

BACK TO TOP

Search: