Comparative accuracy of optical sensor-based wearable system for non-invasive measurement of blood glucose concentration

Dmitry Rodin*, Michael Kirby, Natallia Sedogin, Yair Shapiro, Albert Pinhasov, Anatoly Kreinin

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review


Non-invasive biosensors for indirect evaluation of routinely-measured blood components by sweat analysis have broad potential clinical applications. This trial tested a wrist-borne non-invasive glucose monitor (NIGM) to measure blood glucose (BG) levels using photoplethysmographic (PPG) optical sensors. Our aim was to determine the accuracy of the device in comparison with a standard, invasive clinical method for blood glucose monitoring. Adult participants (n = 200) of both sexes from 18 to 75 were recruited for the study. Exclusion criteria: hemophilia and other serious coagulation disorders, impaired venous access, other serious medical conditions. A biosensor was placed on the right wrist of each participant for a non-invasive indirect BG measurement. In parallel, blood from the antecubital vein was collected and glucose levels were assessed with YSI 2300 Bioanalyzer. The measurements were performed twice: before and after food intake, with a 1-h interval between measurements. There were no limitations to food type and quantity. In both anteprandial (ρ = 0.8994, p < 0.0001) and postprandial (ρ =0.9382, p < 0.0001) glucose measurements, NIGM correlated with values obtained by the YSI 2300 reference device – there was no significant difference between the two methods. Plotted on a Parkes Error Grid for Type II diabetes, NIGM readings did not deviate from those of the YSI 2300 in any clinically-significant way, with the majority of correlated readings falling within Parkes zone A. Very few readings fell within Parkes zone B. In anteprandial measurements, the mean bias between methods for all patient volunteers was 3.705 ± 7.838. In postprandial measurements gave a mean bias of 1.362 ± 10.15 for all patient glucose data. The mean absolute relative difference of currently available glucometer models ranges from 5.6% to 20.8%. The NIGM falls in the lower end of this error range at 7.40–7.54%, indicating that PPG-chemochrome sensors are capable of producing results comparable with those of direct measure glucometers. Data presented here demonstrates the reliability and accuracy of the NIGM system as an adjunctive, and perhaps substitutive, non-invasive tool for blood glucose monitoring.

Original languageEnglish
Pages (from-to)15-20
Number of pages6
JournalClinical Biochemistry
StatePublished - Mar 2019
Externally publishedYes


  • Diabetes
  • Glucose level
  • Non-invasive glucometry
  • Optical biosensor


Dive into the research topics of 'Comparative accuracy of optical sensor-based wearable system for non-invasive measurement of blood glucose concentration'. Together they form a unique fingerprint.

Cite this