> There is no replacement for animal models, except trials in humans, so if the alternative is that we do more, riskier trials on humans, then I'd much prefer animal testing

Totally true, but, at the same time---

Computational models have prevented all kinds of unnecessary animal testing of things that can be shown computationally to be unlikely to work. So, computational models avoid us doing more, riskier trials on animals.

Of course, we've not reduced animal testing-- instead we've just pursued more compounds and ideas. I'd argue it's worth it for speeding up the pace of medicine. But these things are not completely orthogonal.

There is the chance the computational model is wrong.

The animal and human trials don't always go well either.

You only hear about the human trials that go wrong.

Yes, and there is the chance the animal model doesn’t represent human responses well, too.

We still use both to weed out treatments, for both ethical and cost reasons. Better to errantly eliminate a working drug occasionally than to let a lot of unhelpful stuff through.

Just to add to this, the last thing that will be taken over by computers is tox testing. This is the last step that says "this drug is safe enough in primates to test in humans". It's the most complicated type of animal testing, because it is necessarily the interaction of every system in the body that could cause a tox issue.

So sure, we use modeling, and it's getting better, but it's not going to replace toxicology studies any time soon, unless every other human system is replicated in silico first.

A comment by bsdz links a research paper and says that a computer model for cardiotoxicity has higher accuracy than animal models. The paper title mentions in silico as well.

Are those the same things you're talking about or is there something I'm missing. I'm not that familiar with drug development.

https://news.ycombinator.com/item?id=32048391

The paper linked by bsdz checks for one specific cardiac side effect, it is not anywhere close to being a model for cardiotoxicity as a whole, a very important distinction.

Ok, that makes sense.

Is there a meta analysis of such models?

I'm not sure tbh, this really isn't my area of expertise. Sorry, I was hopeful someone more knowledgeable would reply.

Or we halt the research until computational models become viable. You make it seem like we have no choice but to test on animals. I’m sure that is how the researchers rationalize it to themselves.

We could do that, but the amount of preventable suffering that decision would cause, to me, makes that a bad decision.

I'm sure an argument to stop research in lifesaving medicine is easy to rationalize if you're in a privileged position where you don't have to deal with your children dying from malaria or being born with HIV, your elder family members slowly disappearing to Alzheimer's, or your partner's body withering away to cancer

I have had to deal with at least one of those things. It still doesn’t change my opinion.

I feel that it is ethically wrong to enslave and torture animals for humanities benefit. If that means that I will die from a disease that would have been prevented otherwise, I can accept that. I don’t feel it is right for me to impose suffering on another sentient being for my own incremental benefit.

And let people suffer for decades with much slower advancement in drug making and therapy development?

Perhaps you can sell it to patients and their families first before agitating it in the public sphere.

I guess I don’t understand why it is considered morally acceptable to torture countless animals for the benefit of humanity.

I understand it from a purely utilitarian standpoint, but it doesn’t sit well with me ethically.

I wouldn’t want an advanced alien species to enslave and torture me at their own benefit.