A machine learning colorimetric biosensor based on acetylcholinesterase and silver nanoparticles for the detection of dichlorvos pesticides

Abstract

An uncomplicated and rapid colorimetric biosensor for the detection of highly toxic organophosphates (OPs) is developed, using the pesticide dichlorvos as a representative OP. Dichlorvos (DCV) detection is based on its inhibition of the catalytic activity of acetylcholinesterase on acetylthiocholine, which in turn is coupled to a reversible aggregation of citrate-capped silver nanoparticles (c-AgNPs) that gives strong color changes in solution. This color change can be observed by the naked eye for rapid screening. It can also be observed by UV-vis spectrometry, allowing quantification of DCV over a linear range of 1–7 μM with a limit of detection (LOD) and limit of quantitation (LOQ) of 0.65 μM and 3.21 μM. We show further that implementing a trained image processing convolution neural network (CNN) gives a superior quantitative performance, with DCV assay accuracy of 97.6% and assay range from 0–60 μM, using a smartphone for image collection and analysis. This machine learning-based DCV assay may find application in field studies to obtain rapid but also quantitative information about contamination in, for example, water and juices.