A miniature optical spectrometer has demonstrated the ability to measure blood glucose, lactate, and alcohol levels through the skin, potentially enabling laboratory-grade biomarker monitoring in wearable health devices.
Developed by researchers at the University of Cambridge and GlitterinTech Limited, the technology, described in Nature Photonics, uses near-infrared light to detect molecular signatures without collecting blood samples.
Conventional spectrometers provide detailed chemical measurements but are generally too large and expensive for wearable applications. Miniaturized alternatives often sacrifice the resolution needed to distinguish biomarkers present at low concentrations.
The new device, called a convolutional spectrometer (ConvSpec), uses a mathematical approach to reconstruct detailed spectral information from light passing through optical components. The fully packaged system measures approximately 4.8 × 6.2 cm, weighs 16.5 g, and has an estimated component cost of $10. Spectral acquisition and processing take less than 0.4 seconds.
To evaluate its biomedical potential, the researchers developed a wrist-worn optical probe and collected approximately 6,000 spectra across 126 participant sessions. Measurements were compared with conventional reference methods, including blood-based glucose and lactate testing.
The device achieved mean absolute errors of 0.36 mmol/L for blood glucose, 0.81 mmol/L for lactate, and 1.70 mg/dL for alcohol. A personalized model also tracked one participant's glucose fluctuations over several days, achieving a mean absolute error of 0.16 mmol/L.
These findings suggest that miniaturized spectroscopy could eventually support frequent, noninvasive biomarker measurements outside clinical laboratories. Potential applications include metabolic monitoring and tracking physiological changes over time.
However, important challenges remain before clinical implementation. The researchers found that glucose prediction models did not transfer reliably between individuals, partly because skin characteristics and other physiological differences influence optical measurements. Larger, more diverse datasets will be needed to improve generalizability.
The technology therefore remains experimental rather than a validated alternative to laboratory testing. Nevertheless, its combination of compact size, rapid analysis, and high spectral precision could help extend quantitative diagnostic monitoring beyond conventional laboratory settings.
