A framework for harmla alkaloid extraction process development using fuzzy-rough sets feature selection and J48 classification

Medicinal plants as the pivotal source of alternative and complementary medicine have recently supported some hopes in alleviating of symptomatology associated with many diseases. The optimization and development of an efficient method for extracting effective medical substances from wild plants have great importance from both medical and economic prospectives. Therefore, the growing significance of using machine learning algorithms has become an influential positive factor in pushing exploration the pharmacological activities from medicinal plants. Peganum harmala is a widespread species growing as a wild plant in Egypt. It is proved to be useful as an anti-hemorrhoid, anthelmintic, and central nervous system (CNS) stimulating agent in folk medicine. Alkaloids, mainly harmine, harmaline, harmol, and harmalol, represent the major active constituent of the seeds of Peganum harmala. In this paper, a real-world case study of Peganum harmala involving extraction of alkaloids from its seeds using machine learning algorithms is presented. Therefore, dried powdered seeds of Peganum harmala were extracted using 70% methanol by the conventional maceration method. The extraction process was carried out 80 times for three runs using 11 variables, including the volume and concentration of organic solvent, HCl, temperature, and PH. This study proposes a fuzzy rough technique with J48 classification model to find the best extraction procedure for the Peganum harmala. The accuracy is evaluated using 10-fold cross-validation. The experimental results of this proposed intelligent model showed a better understanding tool to present the scientific rule for increasing harmala alkaloid yield range to be around 5%.