The study chose 18 participants such that 83% of them were just above the cut-off for severe autism according to the CARS rating scale, 16 of whom were boys, fo

The study chose 18 participants such that 83% of them were just above the cut-off for severe autism according to the CARS rating scale, 16 of whom were boys, for a non-blinded, non-randomized, non-placebo-controlled trial with no control group, with symptoms assessed by evaluators who weren't blinded to previous assessments of each participant, and explicitly used those assessments to "calibrate" all future results without further elaboration, and with most of the subjects having changed other treatments and diets during the observation period. This is poor quality evidence of anything.