Rapid quality evaluation of MXQLM based on bionic sensory and mathematical separation

Maxing Qinlong Mixture (MXQLM) is a widely used hospital formulation for treating chronic bronchitis (CB). Current quality standards rely only on thin-layer chromatography (TLC) for qualitative identification, which cannot rapidly or accurately monitor the preparation's core active ingredients. To address this gap, we developed a fast, accurate quality-control workflow for MXQLM. We performed chemical separation and component analysis of 127 batches of MXQLM using liquid chromatography-mass spectrometry (LC-MS). Network pharmacology identified five potential active compounds, including baicalein. We quantified two key active ingredients and five representative components by high-performance liquid chromatography (HPLC) and used the CRITIC objective weighting method to assign quality grades to MXQLM batches. Next, we applied bionic sensory technologies - Electronic Tongue (ET), Electronic Nose (EN), and Electronic Eye (EE) - to capture sensor response profiles reflecting the preparation's chemical composition. After multivariate feature extraction, we built qualitative and quantitative prediction models for MXQLM quality using mathematical separation techniques. The qualitative model for MXQLM quality grading (ET + EN; PCA-DA) achieved 89.76% accuracy, and the sample-type classification model (ET + EN; BPNN(Training Set: Independent Test Set =7:3)) reached 100% accuracy. For quantitative prediction, the support vector regression (SVR) model for baicalein content yielded R² = 0.8772 with RMSE = 0.0018 on the validation set (ET & EN & EE), while the SVR model for wogonin content gave R² = 0.8650 with RMSE = 0.0011 on the validation set (ET & EE). Compared with traditional TLC, the method developed here markedly reduces detection time, improves MXQLM quality control, and provides a reliable technical reference for rapid evaluation of this preparation.