Nemaura:
So, 1) no peer review, 2) sample group of just 25 (apparently no one without diabetes was invited at all?), 3) deviations of about 30% in about 90% of the cases.
Diamontech:
"A total of 59 healthy persons and 41 diabetics between the age of 18 to over 70 years participated in the study. The participants were asked to attend the measurement fasting, with a low blood glucose level. During the test period, blood glucose was increased with a glucose-containing drink for healthy volunteers. For diabetes patients, the usual increase in blood glucose after a meal was used. This ensured that measurement results were available for both, low and high blood glucose levels. DiaMonTech's non-invasive measurement delivered accurate measurements over the entire measuring range from about 50 mg/dL to over 350 mg/dL."
So, 1) no peer review, 2) small sample group, 3) extremely specific scenario.
…it's a start, I guess.