What you need is a mathematical model of the human body, which you can plug in some data about the chemicals, run the simulation, and then spit out the life-scale effect under different conditions. They don't have such a model because of how highly complex it is and not having the first idea where to start to build one. Hence clinical trials with poor bastards signing death waivers.
In another context, they also didn't know how to plug in data about the chemical and predict the industrial-scale effects in their manufacturing facility. Well until they asked me, and I derived the model, because I'm dantes. So that's why I get to look at these other researchers in their other fields, and spit upon them with disdain.