Accelerated Green Analysis of Phyllanthi Fructus by One Dimensional Convolutional Neural Network (1D CNN) Modeling of Time-Contracted HPLC Fingerprints

Abstract

To address the lengthy analytical cycles and high solvent consumption of conventional HPLC, this study focused on Phyllanthi Fructus (PF) and, by optimizing gradient elution conditions, established a time contracted chromatographic fingerprint. Combined with a one-dimensional convolutional neural network (1D-CNN), this approach enabled quantitative analysis of multiple components. Compared to conventional methods, the analysis time per sample was reduced by approximately 60 minutes, and mobile phase consumption decreased by nearly 79%. Model results demonstrate high concordance between predicted and measured values for most target compounds, with five exhibiting correlation coefficients exceeding 0.91. Paired t-tests and effect size analysis further validate this reliability. Overall, this method significantly enhances detection efficiency and reduces environmental impact while maintaining accuracy, offering a green and efficient new approach for quality control of natural products with considerable potential for widespread application.