Box Benhnken response surface method based on fingerprint spectrum combined with BP neural network for multi index optimization of black pepper extraction process
Pepper is the dried near mature or ripe fruit of Piper nigrum L., a plant in the Piperaceae family. Harvested from late autumn to the next spring when the fruit turns dark green, dried in the sun to produce black pepper. It has the effects of warming and dispersing cold, lowering qi, and eliminating phlegm, and is used for stomach cold vomiting, abdominal pain and diarrhea. Pepper is used in combination with other medicines in a variety of traditional Chinese patent medicines and simple preparations, such as Qingyu Piwen Pill, Qiya Tiantengzi Pill, Pazhu Pill, Compound Geqing Tablet, Huoxue Zhitong Cream, etc., and also used as a separate medicine, such as Xiaoer Fuxin Zhixie Powder. Black pepper contains numerous alkaloids with various pharmacological activities such as pain relief, anti-inflammatory, and antioxidant properties. Piperine is a representative component and a quality indicator of pepper. In the 2020 edition of the Chinese Pharmacopoeia, the content of piperine is often used as a quality evaluation indicator for medicinal herbs or preparations such as pepper and black pepper, or for prescriptions made with pepper. The existing literature on pepper extraction methods mostly uses piperine as a single evaluation indicator, without considering the impact of extraction processes on the content of other alkaloids in pepper. Chinese medicine fingerprint is a comprehensive and quantifiable quality evaluation method for identifying the intrinsic characteristics of traditional Chinese medicine. It has been applied to the quality control of traditional Chinese medicine since the 2010 edition of the Chinese Pharmacopoeia, mainly through HPLC chemical fingerprint, which can effectively overcome the limitations of a single evaluation index. The optimization of existing extraction processes mainly uses response surface optimization and orthogonal experiments, but response surface optimization experiments often have limitations, and the selected solution may not necessarily be the best solution. BP neural network simulates the function and thinking mode of human brain neural network, and uses backpropagation to adjust the weights and thresholds of the network to handle complex nonlinear problems in order to achieve ideal results. To this end, this study intends to use BP neural network combined with traditional Chinese medicine fingerprint, and optimize the effects of factors such as ethanol concentration, solid-liquid ratio, and ultrasound time on black pepper extraction process through Box Benhnken response surface methodology. The process parameters of black pepper extraction will be clarified to provide reference for subsequent production of related formulations.

 

 

Traditional Chinese medicine exerts pharmacological effects through the coordination of multiple components, but existing literature on the extraction process of piperine from pepper mostly evaluates it based on a single indicator, namely piperine, and the results of the extraction process are difficult to reflect the inherent nature of the complex multi-component nature of traditional Chinese medicine. This study used multiple indicators (the content of piperine, i.e. the ratio of piperine peak area to total peak area, the content of common components in black pepper, i.e. the ratio of common peak area of black pepper extract to total peak area, and the similarity of fingerprint spectra) to evaluate the extraction process of piperine in black pepper, which better reflects the inherent nature of the synergistic effect of multiple components in traditional Chinese medicine. It considers not only the main quality marker of black pepper, piperine, but also other unknown components, focuses on individual and overall considerations, is more scientific and reliable, and can provide more scientific theoretical and technical support for the extraction of piperine in pepper.
The 2020 edition of the Chinese Pharmacopoeia and related literature used HPLC method for quantitative analysis of piperine, with methanol water as the mobile phase, UV detector as the detector, and detection wavelength of 343nm. After scanning the entire wavelength of piperine, it was found that there is also an absorption peak at 240nm wavelength, and the absorbance value can meet the analysis requirements. At the same time, the determination requirements of other alkaloids can also be met at 240nm wavelength. Therefore, the determination wavelength of the fingerprint spectrum in this study was selected as 240nm. After changing the mobile phase from methanol water to acetonitrile 0.1% formic acid aqueous solution, the separation efficiency between the components was significantly improved.
There are literature reports. Among them, Wang et al. used orthogonal experimental design to obtain the optimal extraction process of piperine from black pepper by ultrasonic extraction method, which was ethanol concentration of 80%, solid-liquid ratio of 1:20, ultrasonic power of 700W, and ultrasonic time of 80min. Wang et al. designed using response surface methodology and obtained the optimal extraction process parameters for piperine in white pepper using reflux extraction method, which were a material to liquid ratio of 1:12, 3 extractions, and 2.0 hours each time. In addition, the extraction of piperine from pepper can also be achieved through methods such as supercritical CO2 extraction, enzyme assisted supercritical CO2 extraction, and enzymatic extraction. This study considered the use of anhydrous ethanol as the extraction solvent and ultrasonic extraction as the extraction method for further research. Although response surface methodology and orthogonal experimental design are commonly used for optimizing extraction processes, they often have limitations in optimizing experiments, and the selected solution may not necessarily be the best solution. BP neural network can utilize the self-learning ability of neural networks to compensate for the shortcomings in previous optimization experiments and provide new solutions for optimizing extraction methods. This study is based on traditional Chinese medicine fingerprint technology and uses Box Benhnken response surface methodology combined with BP neural network to optimize the extraction process of black pepper. The results show that the extraction process obtained by BP neural network is superior to that obtained by response surface optimization and more stable.