On the FDA announcement

FDA just announced removal of *absolute requirement* for animal safety testing for antibody development - and it's blowing up the Internet - in a bad way.

Specifically, the FDA are now accepting data from AI/in silico modelling, organs and existing human trials from other countries.

About time. This should have happened years ago.

Animal testing shouldn't be forced in areas where it can be substituted, isn't needed or doesn't make sense. The cosmetics industry, for example, has used human skin explants for decades.

This particular announcement mostly concerns monoclonal antibody testing in *primates*. In context, until now, the FDA required *new primate* safety data from antibodies that had already been in testing or use in humans in other countries - while the cost of a lab monkey is $50,000.

However, that doesn't mean we are anywhere near replacing animals.

Key sentences:

"There is currently no full replacement for animal models in biomedical research and drug development," the National Association for Biomedical Research (NABR) said in a statement.

"While AI promises to speed up many aspects of research, it largely depends on pulling from existing data", said NABR President Matthew Bailey."

Every successful example of AI in bio (e.g. AlphaFold) required a large, structured, high quality dataset to be developed.

This means that we will need *more animal data* to train these models. That is, large high quality, reproducible, standardized dataset of animal - until there's enough data to make the models good enough.

For that, we need automation.

No-one does animal testing because they like to.

We do so because for most drugs, there's no other option.

That is the reality. Let's go and solve it.

Original post (for comments)