Investigation of narrow-band NIR LED spectral response towards a non-invasive glucose sensor

Posted by Josie Dixon on 7 January 2024

The prevalence of type 2 diabetes is increasing. More than 250,000 older New Zealanders live with this disease. Increased insulin resistance leads to high blood glucose and other serious complications such as heart disease, blindness, limb amputations, and early death.

Poor blood glucose management is the primary culprit behind these complications and carries an estimated economic cost of 1% GDP/year (2-3B NZ$/year). The key to good control is easily accessed, low-cost, high adherence, and non-invasive/pain-free glucose measurement. To date, the pin-stick (painful, low adherence) and implanted glucose sensors ($50-100/week) do not meet these needs. There is an urgent need to develop low-cost, non-invasive blood glucose monitoring methods and improve health outcomes for people living with diabetes in New Zealand and internationally. The proposed project meets this goal and will provide significant impact on health care and outcomes for older adults.

Currently, the only way to measure the spectral response of an LED is with expensive, lab-based spectrometer devices, whose high cost and subsequent rarity make this task difficult. To understand the components used in the glucose sensor and develop scientific knowledge of NIR-LEDs, a discrete spectrometer was designed to enable fast and cost-effective quantification of the spectral response curves. The system developed in this work is a low-cost, discrete spectrometer and allows customisation of the emitting and detecting wavelengths. The system consists of electrical hardware for driving the emitting LEDs, including a Teensy 4.0 micro-controller (Teensy) with USB serial connection for power and software connection, and a simple detecting LED connected to a multi-meter. The sensor electrical circuitry is cased in 3D printed housings for protection and consistency of measurement. Figure 1 shows a schematic of the system developed.

Figure 1: Schematic of the set-up for the discrete spectrometer developed.

 This system was used to develop initial spectral response curves for 6 NIRLEDs in the spectrum of interest for detecting blood glucose, as shown in Figure 2. The findings invalidated initial assumptions about the absorption spectra of NIR-LEDs. The literature showed using a narrow-band LED in place of a traditional photodiode for visible light can reduce the noise produced by stray/ambient light. This science was also applied to the NIR-LEDs used on the sensor. However, the absorption by the LEDs is much wider and is shifted, potentially causing inaccuracies with sensor readings. Thus, these results lead to a significant redesign of the detecting circuit in the sensor.

Figure 2: Spectral response curves developed using the discrete spectrometer system.

A HardwareX paper on the system design has been submitted and is in review. The graphical abstract can be seen in Figure 3. I was also fortunate enough to present these results at the Engineering in Medicine and Biology Conference in Sydney in July 2023.

Figure 3: Graphical abstract for the spectral response system HardwareX.