YAIB-cohorts/Python/kidney_function.py at main · aidh-ms/YAIB-cohorts · GitHub

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
import os
import argparse
import pyarrow as pa
import pyarrow.parquet as pq

from src.cohort import Cohort, SelectionCriterion
from src.steps import (
    InputStep, LoadStep,
    AggStep, FilterStep, TransformStep, CustomStep, DropStep, RenameStep,
    Pipeline
)
from src.ricu import stay_windows, hours
from src.ricu_utils import (
    stop_window_at, make_grid_mapper, make_patient_mapper,
    n_obs_per_row, longest_rle
)

outc_var = "crea"
static_vars = ["age", "sex", "height", "weight"]
dynamic_vars = ["alb", "alp", "alt", "ast", "be", "bicar", "bili", "bili_dir",
                  "bnd", "bun", "ca", "cai", "ck", "ckmb", "cl", "crea", "crp",
                  "dbp", "fgn", "fio2", "glu", "hgb", "hr", "inr_pt", "k", "lact",
                  "lymph", "map", "mch", "mchc", "mcv", "methb", "mg", "na", "neut",
                  "o2sat", "pco2", "ph", "phos", "plt", "po2", "ptt", "resp", "sbp",
                  "temp", "tnt", "urine", "wbc"]

def create_kf_task(args):
    print('Start creating the kidney function task.')
    print('   Preload variables')
    load_kf = Pipeline('Load and process kidney function')
    load_kf.add_step([
        LoadStep(outc_var, args.src, cache=True),
        FilterStep('time', lambda x: (x > 24) & (x <= 48)),
        DropStep('time'),
        AggStep('stay_id', 'median'),
    ])
    kf = load_kf.apply()
    load_static = LoadStep(static_vars, args.src, cache=True)
    load_dynamic = LoadStep(dynamic_vars, args.src, cache=True)

    print('   Define observation times')
    base_patients = stay_windows(args.src)
    base_patients = stop_window_at(base_patients, end=168)

    patients = stop_window_at(base_patients.copy(), end=24)
    patients = stop_window_at(patients, end=24)

    print('   Define exclusion criteria')
    # General exclusion criteria
    excl1 = SelectionCriterion('Invalid length of stay')
    excl1.add_step([
        InputStep(base_patients),
        FilterStep('end', lambda x: x < 0)
    ])

    excl2 = SelectionCriterion('Length of stay < 6h')
    excl2.add_step([
        LoadStep('los_icu', args.src),
        FilterStep('los_icu', lambda x: x < 6 / 24)
    ])

    excl3 = SelectionCriterion('Less than 4 hours with any measurement')
    excl3.add_step([
        load_dynamic,
        AggStep('stay_id', 'count'),
        FilterStep('time', lambda x: x < 4)
    ])

    excl4 = SelectionCriterion('More than 12 hour gap between measurements')
    excl4.add_step([
        load_dynamic,
        CustomStep(make_grid_mapper(base_patients, step_size=1)),
        CustomStep(n_obs_per_row),
        TransformStep('n', lambda x: x > 0),
        AggStep('stay_id', longest_rle, 'n'),
        FilterStep('n', lambda x: x > 12)
    ])

    excl5 = SelectionCriterion('Aged < 18 years')
    excl5.add_step([
        LoadStep('age', args.src),
        FilterStep('age', lambda x: x < 18)
    ])

    # Task-specific exclusion criteria
    excl6 = SelectionCriterion('Length of stay < 48h')
    excl6.add_step([
        LoadStep('los_icu', src=args.src, cache=True),
        FilterStep('los_icu', lambda x: x < 48/24)
    ])

    excl7 = SelectionCriterion('Had no creatinine measurement between 24 and 48 hoursn')
    excl7.add_step([
        InputStep(kf),
        CustomStep(make_patient_mapper(patients)),
        FilterStep('crea', lambda x: x.isnull())
    ])


    print('   Select cohort\n')
    cohort = Cohort(patients)
    cohort.add_criterion([excl1, excl2, excl3, excl4, excl5, excl6, excl7])
    print(cohort.criteria)
    patients, attrition = cohort.select()
    print('\n')

    print('   Load and format input data')
    outc_formatting = Pipeline("Prepare kidney function")
    outc_formatting.add_step([
        InputStep(kf),
        CustomStep(make_patient_mapper(patients)),
        RenameStep(outc_var, 'label')
    ])
    outc = outc_formatting.apply()

    dyn_formatting = Pipeline("Prepare dynamic variables")
    dyn_formatting.add_step([
        load_dynamic,
        CustomStep(make_grid_mapper(patients, step_size=1))
    ])
    dyn = dyn_formatting.apply()

    sta_formatting = Pipeline("Prepare static variables")
    sta_formatting.add_step([
        load_static,
        CustomStep(make_patient_mapper(patients))
    ])
    sta = sta_formatting.apply()

    return (outc, dyn, sta), attrition


if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument('--src', default='mimic_demo', help='name of datasource',
                        choices=['aumc', 'eicu', 'eicu_demo', 'hirid', 'mimic', 'mimic_demo', 'miiv'])
    parser.add_argument('--out_dir', default='../data/kidney_function', help='path where to store extracted data')
    args = parser.parse_known_args()[0]

    (outc, dyn, sta), attrition = create_kf_task(args)

    save_dir = os.path.join(args.out_dir, args.src)
    os.makedirs(save_dir, exist_ok=True)
    pq.write_table(pa.Table.from_pandas(outc), os.path.join(save_dir, 'outc.parquet'))
    pq.write_table(pa.Table.from_pandas(dyn), os.path.join(save_dir, 'dyn.parquet'))
    pq.write_table(pa.Table.from_pandas(sta), os.path.join(save_dir, 'sta.parquet'))

    attrition.to_csv(os.path.join(save_dir, 'attrition.csv'))