An AuNPs/CNT electrochemical aptasensor for accurate cortisol level discrimination assisted by machine learning

Cortisol levels provide valuable information regarding physiological stress and abnormalities associated with endocrine and mental health disorders. In this study, an AuNPs/CNT-modified electrochemical aptasensor integrated with machine learning was developed for sensitive cortisol detection and accurate discrimination of cortisol levels. A porous CNT network supported a large specific surface area with high conductivity, and density functional theory (DFT) calculations revealed the AuNPs modified the electronic structures of CNTs and further accelerated the charge transfer efficiency. Once cortisol specifically binds to the aptamer, target-triggered conformational rearrangement could destabilize the aptamer/cDNA duplex and release the aptamer from the electrode surface. The reduced surface nucleic acid density alleviated steric and electrostatic hindrance for the diffusion of the [Fe(CN)₆]^3-/4- probe, which resulted in an accelerated interfacial electron transfer and yielded an enhanced electrochemical signal. The proposed cortisol aptasensor exhibited a good linear relationship between the current variation and the logarithm of cortisol concentration in the range of 10-5000 nM, with a limit of detection of 1.61 nM. Simultaneously, supervised machine learning was applied to classify cortisol concentrations with low, normal, and high levels, thus providing a promising strategy for intelligent stress assessment.