A more flexible single-index regression model is employed to characterize the conditional distribution. For this emiparametric model, a pseudo least integrated squares pproach is developed for the estimation of index oefficients. It is shown in the numerical studies that our estimator outperforms both the pseudo maximum likelihood and semiparametric least squares ones. In addition, we propose the generalized cross-validation criteria for bandwidth selection and the bootstrap implementation for the estimation of asymptotic variance and the construction of confidence intervals. With our defined residual process, a test rule is established to check the adequacy of the considered single-index conditional distribution model. To tackle with the problem of sparse variables, a multiple-stage adaptive Lasso algorithm is developed to identify significant variables and achieve the semiparametric efficiency bound. In this study, a class of simulation scenarios was conducted to assess the finite sample properties of the proposed estimators and inference procedures. Two empirical examples from the house-price study in Boston and the environmental study in New York are further used to illustrate the usefulness of our approaches.