八：logistic logit回归分析.pptVIP

Medical Statistics;Logistic Regression ;simple regression model ; positive and negative;Logit transformation logit 变换 Maximum likelihood 最大似然法 Odds 优势，比值 Odds ratio (OR) 优势比，比值比 Likelihood ratio test 似然比检验 Score test 得分检验 Wald’s test Wald’s 检验 Dummy variable 哑变量;Logistic Regression; Main Content§1. Structure of logistic Regression model;§1. Structure of logistic regression model 某医院对53个接受前列腺癌手术治疗的病人，观察了肿瘤是否已扩散至邻近的淋巴结。下表列出的是手术时直接观察到的结果和手术前的一些指标观察值。其中， y =1为扩散，y =0为未扩散；x =X线，s =触诊肿瘤分期，g =活组织检查肿瘤分级，age =诊断时患者年龄，acid=血清磷酸酶水??。试用手术前观察到的各项指标对前列腺癌是否扩散做出预报。;Y---NODES: 1 lymphatic metastasis; 0- no metastasis;No;P(Y=1)=? P(Y=0)=1-?;Predicting model for ?;§2. Parameter estimation and Hypothesis test;A mathematical method called Newton-Raphson’s method of iteration is available for solving the set of equations, and therefore the partial regression coefficients are obtained.;2. Hypothesis test;(3) Standardized partial regression coefficient b’ ;Wald;Review; ;§3. Interpretation of regression parameters;When exposure factor X takes continuous values. Exposure factor X=(-∞,+∞) (AGE,ACID);;How much Y would change for a unit change in Xj if others were held constant.;The value obtained in ,on the average, with a unit change in X.;Wald;STAGE ;95%CI of OR on STAGE;§4. Goodness of fit for logistic model;2) Predicted probability and ;§5. Notes;4-categorical variable ;AGE Levels 4 40 40~ 50~ 60~;2. Selection of independent variables;Table 16-5 Parameter estimates of final model after selection for independent variables by different selective methods;Logit model in SAS为了了解冠心病与种族的关系，某研究所调查了100个样本，数据列在下表中。试估计各种族间患冠心病的相对危险度。; data a; do y=1 to 0 by -1; do race=b , w , o; input wt@@; output; end; end; cards; 20 5