QSAR STUDY OF ARTEMISININ (QINGHAOSU) DERIVATIVES USING NEURAL-NETWORK METHOD

On the basis of the semiempirical quantum chemical calculations by Aus tin Model 1 (AM1) and Parametric Method 3 (PM3) methods, we have carri ed out quantitative structure-activity relationships (QSAR) study on 2 1 artemisinin derivatives (3) using neural network method. Using 5 qua ntum chemical indices as input parameters and antimalarial activity as output values, a QSAR model with reliably predictive capability has b een constructed from both AM1 and PM3 calculation results. According t o this model, we have suggested an oxidative mechanism of artemisinin derivatives acting on the malarial parasite and analyzed the differenc e between AM1 and PM3 methods.