Molar Absorptivity Model Powers Near-Infrared Glucose Testing

Introduction

In recent study published in Applied Spectroscopy, a team led by Hadi Barati, alongside Arian Mousavi Madani, Sorena Shadzinavaz, and Mehdi Fardmanesh, from the Department of Electrical Engineering at Sharif University of Technology, Tehran, unveiled a new, simple method to estimate blood glucose using near-infrared (NIR) spectroscopy. Their approach compared a traditional absorbance model, grounded in the known chemical properties of glucose, with principal components regression (PCR) and partial least squares models derived from multivariate spectral data. Many spectroscopic techniques have been applied for the analysis of glucose using both blood samples and non-invasive in vivo measurements of the skin (1–3).

The technique relies on how glucose in blood absorbs NIR light. Blood sample absorbance is measured with a Fourier transform infrared (FT-IR) instrument, while reference glucose levels are obtained using a standard finger-prick glucometer. This combination allows the evaluation of how well the chemically grounded model can replicate glucose measurements without extensive statistical training (1–3).

Molar absorptivity model powers near-infrared glucose testing © chompoo-chronicles-stock.adobe.com

A Different Model: Molar Absorptivity in Action

Rather than depending exclusively on multivariate regression techniques like PCR, the researchers developed a “blind” model that calculates regression coefficients directly from the molar absorptivity of glucose, employing a modified Beer–Lambert law framework. This approach leverages the known spectroscopic properties of glucose—its absorbance across NIR wavelengths—to predict blood glucose from measured spectra, without training on large datasets (1).

A blood sample of 10 mm thickness was measured in transmittance. Blood was taken using the finger-prick method and was immediately encapsulated in a liquid cell with minimal exposure to air. The FT-IR blood transmittance spectra were collected over the wavenumber range of 4000 to 8000 cm⁻¹, with particular attention to the 4700 cm⁻¹ (2128 nm) spectral region, where glucose exhibits a clearly identifiable NIR absorption peak. These spectra were then processed to produce absorbance values, allowing direct comparison between the molar-absorptivity model and PCR predictions (1).

FT-IR Spectra Processing

For each blood sample, an FT-IR transmittance spectrum was measured. The raw transmittance data were processed in a series of steps to generate absorbance values corresponding to glucose concentration (1).

1. Transmittance to Absorbance: Transmittance data (T) were converted to absorbance (A) using the relation A = −log₁₀ T.
2. Water Absorbance Correction: The absorbance spectrum of pure water, measured separately, was subtracted from the sample absorbance to remove the water contribution.

This correction enhanced the visibility of glucose absorption peaks in the spectra, which is critical for accurate glucose estimation. By contrast, PCR derives regression coefficients directly from the patterns within the measured dataset, requiring multivariate analysis to identify the spectral regions most informative for glucose prediction (1).

Findings and Implications

The following findings were reported (1).

• Accuracy of the molar absorptivity model: The blind approach produced glucose estimates with acceptable accuracy relative to PCR, demonstrating that intrinsic chemical constants can effectively guide transmittance measurement predictions.
• Practical advantages: While PCR remains statistically more precise, the molar-absorptivity model simplifies calibration, reduces reliance on large datasets, and can be implemented more quickly in clinical or portable devices.
• Instrumentation and reference data: The study underscores the utility of an FT-IR spectrometer for measuring transmittance and converting it to absorbance, combined with a finger-prickglucometer for ground-truth glucose levels. This pairing provides a reliable framework for validating transmittance measurement spectroscopic methods.
• Future prospects: The approach has potential applications in wearable or point-of-care NIR devices, as it combines chemical specificity with minimal data preprocessing. Continued refinement could improve signal-to-noise ratios and expand its applicability across broader glucose ranges or diverse patient populations.

Conclusion

This research highlights a chemically grounded alternative to multivariate regression methods for transmittance measurement blood glucose estimation. By using molar absorptivity and NIR absorbance, researchers demonstrated that it is possible to achieve results comparable to PCR-based models, potentially simplifying calibration and implementation for real-world applications.

References
(1) Barati, H.; Mousavi Madani, A.; Shadzinavaz, S.; Fardmanesh, M. Principal Component Analysis and Near-Infrared Spectroscopy as Noninvasive Blood Glucose Assay Methods. Appl. Spectrosc. 2025, 79 (7), 1047–1055. DOI: 10.1177/00037028241300535

(2) Sameera, F. M.; Kumar, J. S.; Jeya, P. A.; Selvaraj, J.; Angeline, K. S. Potential of Near-Infrared Optical Techniques for Non-Invasive Blood Glucose Measurement: A Pilot Study. IRBM 2025, 46 (1), 100870. DOI: 10.1016/j.irbm.2024.100870

(3) Song, L.; Han, Z.; Shum, P. W.; Lau, W. M. Enhancing the Accuracy of Blood-Glucose Tests by Upgrading FTIR with Multiple-Reflections, Quantum Cascade Laser, Two-Dimensional Correlation Spectroscopy, and Machine Learning. Spectrochim. Acta A Mol. Biomol. Spectrosc. 2025, 327, 125400. DOI: 10.1016/j.saa.2024.125400