According to Wikipedia, glucose monitors have sensors under the skin. Seems a little invasive to do as a screening test.

I wonder if you could use spectroscopy to analyze glucose content in blood? For example, take an existing oxygen meter design and tune it to a different wavelength to detect glucose instead of oxygen.