Monadic linear regression analysis and nature orthonormalization function as well as stepwise regression were integrated to establish forecast model in hot and cold seasons for the young in Nanjing and cerebral hemorrhage as well as cerebral infarction for the old in Jinhua with apparent temperature referred as independent variable.And forecast ability was compared between the models built by selecting factors as well as orthonormalization and the other models built by stepwise regression analysis directly.The result shows that forecast ability of model by monadic linear regression analysis and nature orthonormalization function as well as stepwise regression was better than the model by direct stepwise regression analysis.