There seems to be a significant incentive to getting data that fits the purpose and run it through these algorithms. It's probably not trivial, but paid experts

There seems to be a significant incentive to getting data that fits the purpose and run it through these algorithms. It's probably not trivial, but paid experts should be able to deal with it pretty quickly. It's probably enough to run confined experiments. I think you don't need all the data to derive the most important conclusions.