Smartphone-based colorimetric determination of dopamine using modified plasmonic gold nanoparticles via implementation of U2-net and machine learning algorithms

Dopamine (DA) is a key neurotransmitter involved in numerous mental and motor functions, with abnormal dopamine levels linked to disorders such as parkinsonism, schizophrenia, and attention-deficit hyperactivity disorder (ADHD). This study details the development of a straightforward, precise smartphone-based colorimetric method to quantify dopamine using plasmonic gold nanoparticles (AuNPs) functionalized with N -acetyl-L-cysteine (N-AC). This method utilizes dopamine-induced aggregation of modified AuNPs, causing a color transition from pink to blue detectable by the naked eye. Image acquisition is performed using a smartphone, followed by data preprocessing that incorporates the U 2 -Net deep learning model along with principal component analysis (PCA). To estimate dopamine concentrations, the random forest regression machine learning algorithm is employed. Various conditions, including buffer type, strength, pH, and reaction time, were optimized for effective results. The proposed method demonstrated linearity for dopamine concentrations ranging from 75 to 400 μM, showing high selectivity against interfering substances like homovanillic acid, epinephrine, and ascorbic acid. Validation of the method adhered to International Council on Harmonisation (ICH) guidelines, successfully determining dopamine levels in both pure and spiked plasma samples, achieving a limit of detection of 17.2 μM. This study marks the first implementation of U 2 -Net for smartphone-derived data preprocessing and highlights the innovative combination of smartphone technology with machine learning through a user-friendly Streamlit web application for estimating dopamine concentrations.