get Cox results (p-value, confidence interval, etc)

[aster94]

Using print_summary() I am able to see the p-value, confidence interval, hazard ratio coefficient, but, funnily, I think that I am doing something cumberstone, this is how I get them into a variable:

import pandas as pd
from lifelines import CoxPHFitter
df = pd.DataFrame({'days':[0,1,2],
                   'event':[3,4,5],
                   'covariable':[6,7,8]})
cph = CoxPHFitter(alpha=0.1)
d = cph.fit(df, duration_col='days', event_col='event')
cph.print_summary()
r = round(d.hazard_ratios_.iat[0],2)
ci = (round(1+d.confidence_intervals_.iat[0,0],2), round(1+d.confidence_intervals_.iat[0,1],2))

And also, I am not able to get the p-value as a variable, can someone enlight me?

[CamDavidsonPilon]

.print_summary is just a nice way to look at .summary:

import pandas as pd
from lifelines import CoxPHFitter
df = pd.DataFrame({'days':[0,1,2],
                   'event':[3,4,5],
                   'covariable':[6,7,8]})
cph = CoxPHFitter(alpha=0.1)
cph.fit(df, duration_col='days', event_col='event')
cph.print_summary()

covariable_summary = cph.summary.loc['covariable']
hz = covariable_summary['exp(coef)']
ci = (covariable_summary['exp(coef) lower 90%'], covariable_summary['exp(coef) upper 90%'])

What do you think?

[aster94]

Far away a cleaner way to get the variables!

I have a question about your library: how do you manage the missing values?
Because my dataset is actually missing some values here and there, something like this:

df = pd.DataFrame({'days':[0,1,2,3,4],
                   'event':[3,4,5,9,7],
                   'cov1':[6,7,8,1,Na],
                   'cov2':[1,2,3,Na,2],
                   'cov3':[Na,2,3,8,2]})

My approach was to make a for loop and remove the whole line where a missing value was present

[CamDavidsonPilon]

Missing values is an entire field of statistics, take a look at pages like this: https://stefvanbuuren.name/fimd/sec-problem.html

I think lifelines will just reject the data if there are nans