Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add confidence intervals cox time varying fitter #1150

Open
wants to merge 3 commits into
base: master
Choose a base branch
from
Open

Add confidence intervals cox time varying fitter #1150

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
b8a4959
baseline hazard and baseline cumulative hazard confidence bounds added
stefan-de Oct 14, 2020
8d5c90f
rename survival fun
stefan-de Oct 14, 2020
118e6b9
remove from .idea etc
stefan-de Oct 14, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
42 changes: 34 additions & 8 deletions lifelines/fitters/cox_time_varying_fitter.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -80,8 +80,12 @@ class CoxTimeVaryingFitter(SemiParametricRegressionFitter, ProportionalHazardMix
the strata provided
standard_errors_: Series
the standard errors of the estimates
baseline_hazard_: DataFrame
baseline hazard with confidence bounds
baseline_cumulative_hazard_: DataFrame
baseline cumulative hazard with confidence bounds
baseline_survival_: DataFrame
baseline survival with confidence bounds
"""

_KNOWN_MODEL = True
Expand Down Expand Up @@ -225,7 +229,8 @@ def fit(
normalize(X, self._norm_mean, self._norm_std), events, start, stop, weights
)
self.confidence_intervals_ = self._compute_confidence_intervals()
self.baseline_cumulative_hazard_ = self._compute_cumulative_baseline_hazard(df, events, start, stop, weights)
self.baseline_hazard_ = self._compute_baseline_hazard(df, events, start, stop, weights)
self.baseline_cumulative_hazard_ = self._compute_baseline_cumulative_hazard()
self.baseline_survival_ = self._compute_baseline_survival()
self.event_observed = events
self.start_stop_and_events = pd.DataFrame({"event": events, "start": start, "stop": stop})
Expand Down Expand Up @@ -788,14 +793,16 @@ def plot(self, columns=None, ax=None, **errorbar_kwargs):

return ax

def _compute_cumulative_baseline_hazard(self, tv_data, events, start, stop, weights): # pylint: disable=too-many-locals
def _compute_baseline_hazard(self, tv_data, events, start, stop, weights): # pylint: disable=too-many-locals

with warnings.catch_warnings():
warnings.simplefilter("ignore")
hazards = self.predict_partial_hazard(tv_data).values

unique_death_times = np.unique(stop[events.values])
baseline_hazard_ = pd.DataFrame(np.zeros_like(unique_death_times), index=unique_death_times, columns=["baseline hazard"])
baseline_hazard_ = pd.DataFrame(np.zeros_like(unique_death_times), index=unique_death_times, columns=["baseline_hazard"])
variance_baseline_hazard_ = pd.DataFrame(np.zeros_like(unique_death_times), index=unique_death_times, columns=
["variance_baseline_hazard"])

for t in unique_death_times:
ix = (start.values < t) & (t <= stop.values)
Expand All @@ -807,15 +814,34 @@ def _compute_cumulative_baseline_hazard(self, tv_data, events, start, stop, weig

deaths = events_at_t & (stops_at_t == t)

death_counts = (weights_at_t.squeeze() * deaths).sum() # should always be atleast 1.
death_counts = (weights_at_t.squeeze() * deaths).sum() # should always be at least 1.
variance_baseline_hazard_.loc[t] = death_counts / hazards_at_t.sum()**2
baseline_hazard_.loc[t] = death_counts / hazards_at_t.sum()
# klein and moeschberger, 2013, p.283
z = inv_normal_cdf(1 - self.alpha / 2)
ci_labels = ["%s_upper_%g" % ("baseline_hazard", 1 - self.alpha), "%s_lower_%g" % ("baseline_hazard", self.alpha)]
upper = baseline_hazard_.values * np.exp(+z * np.sqrt(variance_baseline_hazard_.values) / baseline_hazard_.values)
lower = baseline_hazard_.values * np.exp(-z * np.sqrt(variance_baseline_hazard_.values) / baseline_hazard_.values)
baseline_hazard_[ci_labels[0]] = upper
baseline_hazard_[ci_labels[1]] = lower

baseline_hazard_.loc[0] = [0,0,0]
baseline_hazard_ = baseline_hazard_.sort_index()

return baseline_hazard_

return baseline_hazard_.cumsum()
def _compute_baseline_cumulative_hazard(self):
baseline_cumulative_hazard = self.baseline_hazard_.cumsum()
baseline_cumulative_hazard.columns = baseline_cumulative_hazard.columns[:].str.replace("baseline_hazard",
"baseline_cumulative_hazard")

return baseline_cumulative_hazard

def _compute_baseline_survival(self):
survival_df = np.exp(-self.baseline_cumulative_hazard_)
survival_df.columns = ["baseline survival"]
return survival_df
survival = np.exp(-self.baseline_cumulative_hazard_.baseline_cumulative_hazard)
survival.name = "survival"

return survival

def __repr__(self):
classname = self._class_name
Expand Down