cox回归分析遇到了点问题

终于做到了cox回归分析，最后返回的结果让人难以接受

> m = coxph(Surv(time, event) ~ weight, data =  temp)
Warning message:
In fitter(X, Y, istrat, offset, init, control, weights = weights,  :
  Ran out of iterations and did not converge
> m
Call:
coxph(formula = Surv(time, event) ~ weight, data = temp)

                coef  exp(coef)   se(coef)      z        p
weight101 -1.598e+01  1.146e-07  1.677e+06  0.000    1.000
weight102 -2.960e-03  9.970e-01  9.953e+01  0.000    1.000
weight103  2.601e-02  1.026e+00  9.381e+01  0.000    1.000
weight104 -2.853e-01  7.518e-01  3.605e+01 -0.008    0.994
weight105 -2.370e-01  7.890e-01  6.525e+01 -0.004    0.997
weight106  1.783e-02  1.018e+00  1.108e+02  0.000    1.000
weight108  1.159e-01  1.123e+00  1.782e+02  0.001    0.999
weight109 -2.673e-01  7.654e-01  4.360e+01 -0.006    0.995
weight110 -2.960e-03  9.970e-01  1.408e+02  0.000    1.000
weight111  5.162e-02  1.053e+00  1.797e+02  0.000    1.000
weight113 -2.937e-01  7.455e-01  5.387e+01 -0.005    0.996
weight122  7.753e+00  2.328e+03  1.011e+00  7.668 1.75e-14

难受，想哭？？？
想了想之后，看了一下输入数据的类型

看看这数据类型，突然意识到，weight是字符串型

那既然如此，我更改一下是不是就可以运行了？

> temp$weight <- as.numeric(temp$weight)
> m = coxph(Surv(time, event) ~ weight, data =  temp)
> m
Call:
coxph(formula = Surv(time, event) ~ weight, data = temp)

            coef exp(coef)  se(coef)     z     p
weight 0.0005208 1.0005209 0.0056254 0.093 0.926
Likelihood ratio test=0.01  on 1 df, p=0.9264
n= 343, number of events= 82 
   (28 observations deleted due to missingness)

果不其然，可以了。

总结：
临床数据的整理不仅要包括全面的数据，还要注意数据类型，连续型数据要以数值型呈现。因此，临床数据的整理最后要检查数据，str()查看一下数据最后。