Identification of Plants with Anti-Inflammatory Properties from their Leaves Using Machine Learning

The inflammation is a disease to which the humans are facing from ancient times and their various methods of diagnosis, prophylaxis, and cure are mentioned in various traditional systems of medicines throughout the world. Many plants parts such as leaves were used in earlier times against inflammation which are proven by tests performed in laboratories. List of few plants is stated as Basella alba L. leaves ethanolic extract inhibit membrane lysis in HRBC Stabilization Method and leaves methanolic extract give significant activity against formaldehyde, egg albumin and turpentine oil induced paw edema in rat. Psidium guajava L. leaves methanolic extract give inhibitory action against carrageenan-induced paw edema in rats and the essential oil extracted from leaves inhibit 5-LOX causing anti-inflammatory activity. The Piper betle leaves methanolic extract give inhibitory action against carrageenan-induced paw edema in rats and ethanolic extract by inhibiting inflammatory modulators. Hibiscus rosa sinensis L. ethanolic extract inhibit paw oedema, carrageenan-induced in rat and inhibiting in HRBC hemolysis. Murraya koenigii L. leaf ethanolic extract inhibit heat induced denaturation of albumin and methanolic extract give inhibition of carrageenan-induced paw edema in rat. The leaves ethanolic extract and aqueous extract of Hibiscus rosa sinensis L. inhibit paw edema in rat induced by carrageenan. Three models of machine learning were used including Inception-v3 feature extractor using Logistic Regression, Inception-v3 Feature extractor used with Neural Network confusion matrix and Inception-v3 Feature extractor with Random Forest confusion matrix (Orange Classification Tree). A comparison matrix for every model utilized was generated and maximum accuracy of 99.5% was attained. For all used model the ROC curve was drawn for proper comparison and representation.