/
demo_coxph.py
151 lines (133 loc) · 6.33 KB
/
demo_coxph.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
#!/usr/bin/env python
import ast
import configparser
import csv
import os
os.environ['QT_QPA_PLATFORM'] = 'offscreen'
import sys
import numpy as np
import pandas as pd
from lifelines import CoxPHFitter
from lifelines.utils import concordance_index
from sklearn.preprocessing import StandardScaler
from npsurvival_models import BasicSurvival
from survival_datasets import load_dataset
from util import compute_median_survival_time, compute_IPEC_scores
if not (len(sys.argv) == 2 and os.path.isfile(sys.argv[1])):
print('Usage: python "%s" [config file]' % sys.argv[0])
sys.exit()
survival_estimator_name = 'coxph' # for this file
config = configparser.ConfigParser()
config.read(sys.argv[1])
n_experiment_repeats = int(config['DEFAULT']['n_experiment_repeats'])
IPEC_percentiles = ast.literal_eval(config['DEFAULT']['IPEC_percentiles'])
cindex_method = config['DEFAULT']['cindex_method'].strip()
datasets = ast.literal_eval(config['DEFAULT']['datasets'])
output_dir = config['DEFAULT']['output_dir']
os.makedirs(output_dir, exist_ok=True)
output_table_filename = os.path.join(output_dir,
'%s_experiments%d_%s_table.csv'
% (survival_estimator_name,
n_experiment_repeats,
cindex_method))
if os.path.isfile(output_table_filename):
print('*** Filename already exists: %s' % output_table_filename)
print('*** Skipping')
sys.exit()
output_table_file = open(output_table_filename, 'w')
csv_writer = csv.writer(output_table_file)
csv_writer.writerow(['dataset', 'experiment_idx', 'method', 'cindex'] +
['IPEC (%.2f)' % q for q in IPEC_percentiles])
for experiment_idx in range(n_experiment_repeats):
for dataset in datasets:
print('[Dataset: %s, experiment: %d]' % (dataset, experiment_idx))
print()
X_train, y_train, X_test, y_test, feature_names = \
load_dataset(dataset, experiment_idx)
num_IPEC_horizons = len(IPEC_percentiles)
sorted_train_times = np.sort(y_train[:, 0])
num_train_times = len(sorted_train_times)
IPEC_horizons = [sorted_train_times[int(q * num_train_times)]
for q in IPEC_percentiles[:-1]]
IPEC_horizons.append(sorted_train_times[-1])
print('Testing...')
scaler = StandardScaler()
X_train_standardized = scaler.fit_transform(X_train)
X_test_standardized = scaler.transform(X_test)
sort_indices = np.argsort(y_train[:, 0])
train_data_df = \
pd.DataFrame(np.hstack((X_train_standardized, y_train)),
columns=feature_names + ['time', 'status'])
surv_model = CoxPHFitter()
surv_model.fit(train_data_df, duration_col='time', event_col='status',
show_progress=False, step_size=.1)
sorted_y_test = np.sort(np.unique(y_test[:, 0]))
if sorted_y_test[0] != 0:
mesh_points = np.concatenate(([0.], sorted_y_test))
else:
mesh_points = sorted_y_test
surv = \
surv_model.predict_survival_function(X_test_standardized,
mesh_points)
surv = surv.values.T
# ---------------------------------------------------------------------
# compute c-index
#
if cindex_method == 'cum_haz':
cum_haz = \
surv_model.predict_cumulative_hazard(X_test_standardized,
sorted_y_test)
cum_haz = cum_haz.values.T
cum_hazard_scores = cum_haz.sum(axis=1)
test_cindex = concordance_index(y_test[:, 0],
-cum_hazard_scores,
y_test[:, 1])
elif cindex_method == 'cum_haz_from_surv':
surv_thresholded = np.maximum(surv,
np.finfo(float).eps)
cum_haz = -np.log(surv_thresholded)
cum_hazard_scores = cum_haz.sum(axis=1)
test_cindex = concordance_index(y_test[:, 0],
-cum_hazard_scores,
y_test[:, 1])
elif cindex_method == 'median':
predicted_medians = \
np.array([compute_median_survival_time(mesh_points,
surv_row)
for surv_row in surv])
test_cindex = concordance_index(y_test[:, 0],
predicted_medians,
y_test[:, 1])
elif cindex_method == 'median_from_cum_haz':
cum_haz = \
surv_model.predict_cumulative_hazard(X_test_standardized,
sorted_y_test)
cum_haz = cum_haz.values.T
predicted_medians = \
np.array([compute_median_survival_time(mesh_points,
surv_row)
for surv_row in np.exp(-cum_haz)])
test_cindex = concordance_index(y_test[:, 0],
predicted_medians,
y_test[:, 1])
else:
raise NotImplementedError('Unsupported c-index method: %s'
% cindex_method)
# ---------------------------------------------------------------------
# compute IPEC score using a few horizon times
#
test_IPEC_scores = compute_IPEC_scores(y_train, y_test,
mesh_points, surv,
IPEC_horizons)
print('c-index: %5.4f' % test_cindex
+ ', '
+ ', '.join(['IPEC (%.2f): %5.4f'
% (q, test_IPEC_scores[horizon] / horizon)
for q, horizon in
zip(IPEC_percentiles,
IPEC_horizons)]))
csv_writer.writerow([dataset, experiment_idx, 'cox', test_cindex] +
[test_IPEC_scores[horizon] / horizon
for horizon in IPEC_horizons])
print()
print()