This is interesting, but so far a weak first step on flu. Each year experts stare at numbers and perhaps read some tea leaves to choose 4 strains to cover, often missing somewhat.
The innovation I'd like to see: ignore the "choose 4" process, and instead aim to cover all known strains.
I don’t think experts are reading tea leaves and it’s pretty wild to claim that. There are literally billions of dollars at stake. To bring it bag to software, that’s like me claiming I think machine learning people are just using fancy excel formulas.
I mean... you can do it in excel it is just a lot... slower - you can do like anything in excel
The real problem they have is the flu mutates. They know what strains they see now and can add them all easily enough. In the few months between deciding on strains and manufacturing them the flu can mutate to some new strain that wasn't even seen before.
https://www.nature.com/articles/s41591-020-1118-7 (“A chimeric hemagglutinin-based universal influenza virus vaccine approach induces broad and long-lasting immunity in a randomized, placebo-controlled phase I trial”)