To make any new method clinically relevant, it is crucial that the method's reliability (accuracy and repeatability) be established. There is currently no non-invasive method for blood glucose measurements that qualifies the above standard.
In order to develop a new method, whatever parameter is to be measured as a representation of blood glucose, it has to be reliable under all (or at least most) circumstances. On experimental level, embedded sensors capable of wireless communication with an external device used to be an exciting potential solution for blood glucose measurements. However, over a period of time, these sensors become increasingly less accurate in vivo due to normal physiological response to synthetic material used in them.
Perhaps one these days something new will come along and children wouldn't have to poked 2-3 times a day to get a drop of blood, which would be very positive change.