Statistically, this paper seems weak.

https://www.science.org/doi/10.1126/sciadv.aax1396