Optimization of Lab-On-a-CD by Experimental Design and Machine Learning Models for Microfluidic Biosensor Application

In order to ensure the optimal functionality of biosensor devices across a diverse range of applications, it is crucial to accurately predict their detection times. This study delves into an in- depth exploration of the centrifugal and Coriolis effects that emerge due to the angular alignment and radial displacement of a rotating microfluidic biosensor specifically designed for detecting complex reactive proteins (CRPs). To address this challenge, we introduce an innovative hybrid model known as PSO-ANN, which combines the power of an artificial neural network (ANN) with the particle swarm optimization (PSO) algorithm. This pioneering model is aimed at predicting the response time of a lab-on-a-CD device by utilizing critical input variables, such as rotational velocity (
$\omega $
), biosensor position (
${X}_{S}$
), angular alignment (
$\theta $
), and radial displacement (
${R}_{D}$
). Our research also involves a comprehensive performance evaluation of the PSO-ANN model, comparing it to an alternative multilayer perceptron (MLP) model, ANN. This evaluation seeks to assess the impact of these models on improving the accuracy and reliability of predictions related to biosensor detection times, with potential applications spanning a wide spectrum of practical fields. Key metrics utilized in our evaluation include mean absolute error (MAE), root mean square error (RMSE), variance accounted for (VAF), and coefficient of determination (
${R}^{{2}}$
). The results highlight the remarkable predictive capabilities of the hybrid PSO-ANN model. This research carries significant implications for enhancing the performance of biosensor devices and advancing their utility in various domains, promising advancements in the field of biosensor technology.