There was the just published Covid- test procedure,too, which was scientifically shredded apart. Just because someone publishes some scientific articles, studies and data does not make it settled science.

My master thesis was building on top of prior 20+ years of BS science and peer reviewed publications. Not even a 100k software could actually work out the bad model procedure and testing. The fundamental assumptions were right, but their modeling and conclusions were pure BS. After long nights I figured the problem out and got to a 95%+ accuracy rate for process modeling. The "accepted standard" process gave us between 65 and 75% accuracy. Totally unacceptable in engineering.