Machine Learning-Based Rapid Multi-Component Quantification in Danshen Injections Using 1h Nmr

Abstract

A rapid method for quantitative determination of components in Danshen Injection was developed using one-dimensional proton Nuclear Magnetic Resonance (1H NMR). Three modeling algorithms were evaluated: partial least squares regression (PLSR), extreme learning machine (ELM), and Lasso regression. We explored different approaches for constructing independent variables, including full NMR response, principal component extraction, and Grey Wolf Optimizer (GWO) for variable selection. Model performance was assessed using root mean square error (RMSE), correlation coefficient (R), and residual prediction deviation (RPD). Quantitative models for 18 compounds in Danshen Injection were developed, with Lasso regression yielding the best results. The correlation coefficient ranged from 0.8305 to 0.9837, and RPD values exceeded 1.77. Lasso regression combined with GWO generally produced more robust models with better generalization. This study demonstrates that machine learning can reliably process complex data from 1H NMR spectra, significantly reducing workload and processing time without alignment operations.

Keywords: machine learning, 1H NMR, Rapid quantitative model, Multiple components, Traditional Chinese Medicine injection, Danshen Injection