Optical and embedded ML driven computational optimization of portable micro-volume spectrophotometer for biochemical analysis in resource-limited environment

Abstract

Comprehensive biochemical analysis is an essential part of clinical diagnostics, pharmaceutical development, and biomedical research. The gold-standard technique for such analysis is spectrophotometry, which has demonstrated excellent reliability, quantitative accuracy, and broad biomolecular sensitivity. Commercial micro-volume spectrophotometers, which are instrumental to applications with limited sample volumes, are often restricted to well-equipped laboratories due to the high cost and portability limitations especially in resource-limited settings. To address this gap, we propose a proof-of-concept, 3D-printed micro-volume spectrophotometer that integrates precision optics, microfluidic sample handling (8–15 µL), combined with spectral reconstruction and machine learning-based concentration estimation algorithm in a compact and affordable format. Preliminary validation of the proposed system based on bovine serum albumin (BSA) protein quantification via Biuret assay demonstrates strong agreement with commercial spectrophotometers (R² > 0.91). With a production cost of approximately $36 and rapid measurement, the architecture provides a framework that can be extended to other absorbance-based biochemical assays ensuring accessible spectrophotometric systems for point-of-care and resource-limited applications, overcoming traditional limits of cost, sample volume, and portability.

Keywords

tiny ML / portable spectrophotometry / regression AI / embedded neural networks / point of care devices / optoelectronics

References

[1] Understanding Cuvette Volume, Material, Path Length Etc. | Cuvet.Co Cells. Available from: https://cuvet.co/understanding—cuvette—volume—material—path—length—etc/

[2] Jenway 73 series. Available from: https://www.wolflabs.co.uk/documents/Jenway_Spectrophotometers_73_Series.pdf?srsltid=AfmBOop3NWKNA6mA3mWQbxVjAQP_OXzPlVKdV_OwQmMYOoiR2CyduyKv&

[3] Desjardins P and Conklin D (2010) NanoDrop microvolume quantitation of nucleic acids. J Vis Exp 45: e2565. https://doi.org/10.3791/2565

[4] NanoDrop One Microvolume UV—Vis Spectrophotometers. Available from: https://documents.thermofisher.com/TFS—Assets/CAD/Specification—Sheets/NanoDrop—One—Specifications.pdf

[5] Minhas—Khan A, Ghafar—Zadeh M, Shaffaf T, et al. (2021) UV—Vis spectrophotometric analysis of DNA retrieval for DNA storage applications. Actuators 10: 246. https://doi.org/10.3390/act10100246

[6] Duma Z, Ramsuran V, Chuturgoon AA, et al. (2022) Evaluation of various alternative economical and high throughput SARS—CoV—2 testing methods within resource—limited settings. Int J Mol Sci 23: 14350. https://doi.org/10.3390/ijms232214350

[7] Desjardins P, Hansen JB, Allen M (2009) Microvolume protein concentration determination using the NanoDrop 2000c spectrophotometer. J Vis Exp. https://doi.org/10.3791/1610

[8] Baker A, Ward D, Lieten SH, et al. (2004) Measurement of protein—like fluorescence in river and waste water using a handheld spectrophotometer. Water Res 38: 2934-2938. https://doi.org/10.1016/j.watres.2004.04.023

[9] Di Nonno S and Ulber R (2021) Smartphone—based optical analysis systems. Analyst 146: 2749-2768. https://doi.org/10.1039/D1AN00025J

[10] Koohkan R, Kaykhaii M, Sasani M, et al. (2020) Fabrication of a smartphone—based spectrophotometer and its application in monitoring concentrations of organic dyes. ACS Omega 5: 31450-31455. https://doi.org/10.1021/acsomega.0c05123

[11] Aksoy S, Dulda A, Ertaş G (2023) Emerging technologies for fluorescence—based optical test strip readers. Avrupa Bilim ve Teknoloji Dergisi 49: 16-24. https://doi.org/10.31590/ejosat.1265098

[12] Lebanov L, Mikhail IE, Paull B (2024) Light—emitting diode—based absorbance detectors for flow—through analysis in analytical chemistry: a tutorial. J Chromatogr Open 6: 100191. https://doi.org/10.1016/j.jcoa.2024.100191

[13] Bui DA and Hauser PC (2015) Analytical devices based on light—emitting diodes — a review of the state—of—the—art. Anal Chim Acta 853: 46-58. https://doi.org/10.1016/j.aca.2014.09.044

[14] Zhang C, Bills BJ, Manicke NE (2017) Rapid prototyping using 3D printing in bioanalytical research. Bioanalysis 9: 329-331. https://doi.org/10.4155/bio—2016—0293

[15] Niculescu AG, Chircov C, Bîrcă AC, et al. (2021) Fabrication and applications of microfluidic devices: a review. Int J Mol Sci 22: 2011. https://doi.org/10.3390/ijms22042011

[16] TSL2561 DataSheet. Available from: https://cdn—shop.adafruit.com/datasheets/TSL2561.pdf

[17] Chaianantakul N, Wutthi K, Kamput N, et al. (2018) Development of mini—spectrophotometer for determination of plasma glucose. Spectrochim Acta A Mol Biomol Spectrosc 204: 670-676. https://doi.org/10.1016/j.saa.2018.06.107

[18] González—Morales D, Valencia A, Díaz—Nuñez A, et al. (2020) Development of a low—cost UV—Vis spectrophotometer and its application for the detection of mercuric ions assisted by chemosensors. Sensors 20: 906. https://doi.org/10.3390/s20030906

[19] Macka M, Piasecki T, Dasgupta PK (2014) Light—emitting diodes for analytical chemistry. Annu Rev Anal Chem 7: 183-207. https://doi.org/10.1146/annurev—anchem—071213—020059

[20] Keiser G, Xiong F, Cui Y, et al. (2014) Review of diverse optical fibers used in biomedical research and clinical practice. J Biomed Opt 19: 080902. https://doi.org/10.1117/1.jbo.19.8.080902

[21] Shakeri A, Jarad NA, Leung A, et al. (2019) Biofunctionalization of glass‐ and paper‐based microfluidic devices: a review. Adv Mater Interfaces 6: 1900940. https://doi.org/10.1002/admi.201900940

[22] Prairie MW, Frisbie SH, Rao KK, et al. (2020) An accurate, precise, and affordable light emitting diode spectrophotometer for drinking water and other testing with limited resources. PLoS One 15: e0226761. https://doi.org/10.1371/journal.pone.0226761

[23] Moses Kayanda Kiteto and Cleophas Achisa Mecha (2024) Insight into the bouguer—beer—lambert law: a review. Sustainable Chem Eng 5: 567-587. https://doi.org/10.37256/sce.5220245325

[24] Tiago MM, Higuti RT, Segura LE, et al. (2019) Versatile ultrasonic spectrometer for liquids with practical sample handling by using standard cuvettes. IEEE Trans Ultrason Ferroelectr Freq Control 66: 109-118. https://doi.org/10.1109/TUFFC.2018.2875872

[25] Rabbath CA and Corriveau D (2019) A comparison of piecewise cubic hermite interpolating polynomials, cubic splines and piecewise linear functions for the approximation of projectile aerodynamics. Def Technol 15: 741-757. https://doi.org/10.1016/j.dt.2019.07.015

[26] Fritsch FN and Carlson RE (1980) Monotone piecewise cubic interpolation. SIAM J Numer Anal 17: 238-246. https://doi.org/10.1137/0717021

[27] Banbury C, Zhou C, Fedorov I, et al. (2021) Micronets: neural network architectures for deploying tinyml applications on commodity microcontrollers. MLSys Proceedings 3: 517-532. https://doi.org/10.48550/arXiv.2010.11267

[28] Heydari S and Mahmoud QH (2025) Tiny machine learning and on—device inference: a survey of applications, challenges, and future directions. Sensors 25: 3191. https://doi.org/10.3390/s25103191

[29] Hagan MT and Menhaj MB (1994) Training feedforward networks with the Marquardt algorithm. IEEE Trans Neural Netw 5: 989-993. https://doi.org/10.1109/72.329697

[30] Rieger LH, Wilson M, Vegge T, et al. (2023) Understanding the patterns that neural networks learn from chemical spectra. Digital Discovery 2: 1957-1968. https://doi.org/10.1039/D3DD00203A

[31] Urbina F, Batra K, Luebke KJ, et al. (2021) UV—adVISor: attention—based recurrent neural networks to predict UV—Vis spectra. Anal Chem 93: 16076-16085. https://doi.org/10.1021/acs.analchem.1c03741

[32] Doumas BT (1975) Standards for total serum protein assays——a collaborative study. Clin Chem 21: 1159-1166. https://doi.org/10.1093/CLINCHEM%2F21.8.1159

[33] Noble JE, Knight AE, Reason AJ, et al. (2007) A comparison of protein quantitation assays for biopharmaceutical applications. Mol Biotechnol 37: 99-111. https://doi.org/10.1007/s12033—007—0038—9

[34] Train Deep Neural Networks — MATLAB & Simulink. Available from: https://in.mathworks.com/help/deeplearning/train—deep—neural—networks.html?s_tid=CRUX_topnav

[35] Suhendar H, Silambi MS, Pratama MR (2024) Design of a Portable Spectrophotometer Based on Raspberry Pi for Tea Type Classification Using Machine Learning. J Phys Conf Ser 2866: 012041. https://doi.org/10.1088/1742—6596/2866/1/012041

[36] Giavarina D (2015) Understanding bland altman analysis. Biochem Med (Zagreb) 25: 141-151. https://doi.org/10.11613/bm.2015.015

[37] Altman DG and Bland JM (2017) Assessing agreement between methods of measurement. Clin Chem 63: 1653-1654. https://doi.org/10.1373/clinchem.2016.268870