Machine learning-assisted paper-based chemiluminescence biosensor for choline quantification in infant milk: toward portable nutritional quality monitoring

Abstract Choline is a critical micronutrient that is involved in neurotransmission, lipid metabolism, and neurodevelopment of infants. Poor consumption of choline in the initial stages of growth might lead to severe cognitive and developmental defects. Hence, there is a need to constantly monitor choline level in infant food. To meet this requirement, we present a low-cost paper-based luminol cobalt chemiluminescense (PLC-CL) biosensor with the potential to analyze choline in milk and in real-time using smartphone imaging and machine learning (ML) algorithms. The sensing platform works for the detection of the hydrogen peroxide (H 2 O 2 ) which is generated through the use of choline oxidase (ChOx), and then the produce a luminol-cobalt chemiluminescent reaction, where the intensity of the emitted light is directly proportional to the concentration of choline. ML models quantify light intensity, and are highly accurate without the need to use standard methods of calibration. This device has a linear dynamic range of 0.5–10 mM and a detection limit of 257.12 µM, thus it is useful in quantification in real time and can also be used in the real field envirnoment. This point-of-care testing (PoCT) biosensing tool, enriched with an ML algorithm, therefore, increases nutritional surveillance functions, especially in resource-limited environments, and contributes to adherence to the infant food safety protocols..