Machine Learning Turns Up Covid Surprise

Not denying that secondary infections is a big killer, 2 groups of 50 patients is not a big number.

Also, maybe the patients initially were different that allowed them to recover from a secondary infection or not. It doesn't say much whether a different approach treating these infections would have helped. It might but still remains to be demonstrated. A proper RCT is needed.

Besides, it seems that covid has changed, now almost everyone has been vaccinated or infected, so people have some seroprotection, a different strain is less damaging. The results may not be applicable to potential future outbreaks of covid or other respiratory infections.