Scientific progress has a lot to do with having powerful tools. If you get more powerful tools you can easily get a bunch of new insight. As a result it's worthwhile to focus most of our efforts on building better tools instead of funding specific insights.

Secondly the standard way to judge whether something is poisonous is to see how big of a dose you need to kill rats. If the substance doesn't kill rats or reduce their lifespan but it just reduces their IQ our tox tests don't pick it up. That means that we might have pesticides in use that do things like reducing IQ at doses that are legally allowed without our testing procedure being able to tell.

Practically we could measure variables like the heart rate of a labrat via infrared for it's whole lifespan and see whether it shows unusual patterns. Every action that the rat does in their lifetime could be measured. The rat lives in a controlled environment where a lot of information besides whether or not the rat dies can be gathered.

Of course we already do some additional tests such as a test for whether a substance might cause cancer that has nothing to do with rats but in general we could do much better at measuring toxicology.

On the one hand that means that we could regulate harmful substances where we currently don't know they are harmful. On the other hand drug development also get's cheaper if you catch bad drugs while you do your tests with rats and before you run expensive trials with humans.