Except OP has noticed that when they go out of ketosis everything gets worse again. They've effectively run numerous trials on themselves.
If society wants to determine if X is generally an effective treatment for Y, there need to be robust experiments that involve multiple people as subjects.
If, however, one guy feels a lot better doing X, and feels worse they stop doing X, then sure, that person should probably keep doing X. Why does X work for that person? Who knows?