Related Experiment Video
Updated: Jan 9, 2026
Qualitative Identification of Carboxylic Acids, Boronic Acids, and Amines Using Cruciform Fluorophores
Published on: August 19, 2013
Machine learning-assisted multicolor identification and quantification of antidepressant drugs by waste-derived
Elham Madani-Nejad1, Ardeshir Shokrollahi1, Faezeh Shahdost-Fard2
1Chemistry Department, Yasouj University, 75918-74831, Yasouj, Iran.
Abstract:
Recently, the severe side effects related to the widespread consumption of antidepressants (ADs) have alarmingly created a global challenge for clinics and forensic laboratories. This study introduces a machine learning-empowered multicolor fluorescent sensor array for the rapid and sensitive spectral and visual quantification and discrimination of six common ADs under green chemistry principles. Blue carbon dots (B-CDs) was synthesized by an environmentally friendly method using sweet lemon waste to directly reduce Ag+ to Ag nanoparticles (AgNPs) without any reducing agent, and form core-shell AgNPs@B-CDs. The synthesized core-shell AgNPs@B-CDs and red-emitting CDs (R-CDs) have been utilized as the dual emissive nanoprobes in the absence and presence of cetyltrimethylammonium bromide (CTAB) to create different interactions with the ADs through distinctive color patterns. Notably, vivid concentration-dependent responses akin to a unique fingerprint were discernible both spectroscopically and visually across a broad concentration range, from 5 μM to 100 μM for each AD, in buffer, as well as in human urine and serum samples. The limit of detection (LOD) values of 3.02 μM, 1.81 μM, 2.29 μM, 2.33 μM, 2.07 μM and 1.66 μM and limit of quantification (LOQ) values of 10.09 μM, 6.05 μM, 7.64 μM, 7.78 μM, 6.90 μM and 5.44 μM for SRT, IMI, FLOX, DOX, TFP and PER were calculated, respectively. Statistical analysis of the data set extracted from the spectral profiles was meticulously carried out using two machine learning techniques: PCA reduced the data dimensions, and LDA effectively discriminated ADs and their binary, ternary, and quaternary mixtures with high accuracy. These findings promise the utility and reliability of the proposed green methodology in the fabrication of an AI-based electronic tongue for on-site monitoring of ADs' dosage levels, facilitating routine analysis of human biofluids to aid medicolegal investigations.
