Optimizing microfluidic chip for rapid SARS-CoV-2 detection using Taguchi method and artificial neural network PSO.

Microfluidic biosensors offer a promising solution for real-time analysis of coronaviruses with minimal sample volumes. This study optimizes a biochip for the rapid detection of SARS-CoV-2 using the Taguchi orthogonal table L9(34), which comprises nine groups of experiments varying four key parameters: Reynolds number (Re), Damköhler number (Da), Schmidt number (Sc), and the dimensionless position of the reaction surface (X). Signal-to-noise (S/N) ratios and analysis of variance (ANOVA) are employed to determine optimal parameters and assess their impact on binding kinetics and response time of the detection device. These obtained optimal parameters correspond to Re = 4.10-2, Da = 1000, Sc = 105, and X = 1. Additionally, results highlight Da as the most influential factor, accounting for 91%, while X has a minimal effect of 0.3%. Furthermore, an artificial neural network optimization technique, specifically particle swarm optimization (PSO), was utilized to predict biosensor performance. Derived from the Full L81(34) design experiment, the PSO model demonstrates its effectiveness compared to the conventional multi-layer perception (MLP) model, thus underlining its potential in this innovative optimization context.