Based on Hall et al. electrotopological state indices(EK) of atom types, two quantitative structure-activity relationship(QSAR) models were developed to estimate and predict the action strength(W) of D_(OM)(dimethoxy-methyl-amphetamine) for 18 phenyl-isopropyl-amine dopes(PPAD) through linear method(multiple linear regression, MLR) and non-linear method(Back propagation artificial neural network, BP-ANN). On the basis of EK, the optimal three-parameter(E_(14), E_9, E_7) QSAR model of W for 18 PPAD was constructed. The traditional correlation coefficient(R~2) and cross-validation correlation coefficient(R_(cv)~2) are 0.878 and 0.815, respectively. The result demonstrates that the model is highly reliable(from the point of view of statistics) and has good predictive ability by using R~2, R_(cv)~2, VIF, FIT, AIC and F tests. Form the three parameters of the model, it is known that the dominant influence factors of inhibited activity are the molecular structure fragments: =CH–(secondary carbon), =C<(tertiary carbon atom) in aromatic ring and –O–(phenol ether bond). The results showed that the structure parameters E_(14), E_9 and E_7 have good rationality and efficiency for the W of phenyl-isopropyl-amine dope(PPAD) analogues. A BP-ANN with 3-3-1 architecture was generated by using three electrotopological state index descriptors(E_(14), E_9, E_7) appearing in the MLR model, the above descriptors were inputs and its output was action strength(W). The nonlinear BP-ANN model has better predictive ability compared to the linear MLR model with R~2 and R_(cv)~2 of leave-one-out(LOO) to be 0.995 and 0.994, respectively. The regression method gave support to the neural network with physical explanation, which offers a more accurate model for QSAR. Those models can be used in the rational design of higher stimulating extent PPAD, which provide meaningful reference information to improve the detection methods of PPAD. Based on Hall et al. Electrotopological state indices (EK) of atom types, two quantitative structure-activity relationship (QSAR) models were developed to estimate and predict the action strength (W) of DOM (dimethoxy-methyl-amphetamine) For 18 phenyl-isopropyl-amine dopes (PPAD) through linear method (MLR) and non-linear method (Back Propagation artificial neural network, BP- ANN). On the basis of EK, the optimal three-parameter The traditional correlation coefficient (R ~ 2) and cross-validation correlation coefficient (R_ (cv) ~ 2) are 0.878 and 0.815, respectively. result demonstrates that the model is highly reliable (from the point of view of statistics) and has good predictive ability by using R ~ 2, R_ (cv) ~ 2, VIF, FIT, AIC and F tests. Form the three parameters of the model, it is known that the dominant influence factors of inhibited activity are the molecular structure fragments: = CH- (secondary C = (tertiary carbon atom) in aromatic ring and -O- (phenol ether bond). The results showed that the structure parameters E_ (14), E_9 and E_7 have good rationality and efficiency for the W of phenyl- A BP-ANN with 3-3-1 architecture was generated by using three electrotopological state index descriptors (E_ (14), E_9, E_7) appearing in the MLR model, the above descriptors were inputs and its output was the action strength (W). The nonlinear BP-ANN model has better predictive ability compared to the linear MLR model with R ~ 2 and R_ (cv) ~ 2 of leave-one-out 0.994, respectively. The regression method gave support to the neural network with physical explanation, which offers a more accurate model for QSAR. Those models can be used in the rational design of higher stimulating extent PPAD, which provide meaningful reference information to improve the detection methods of PPAD.