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
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211 | def get_i2b2_table(
spark_session: SparkSession, db_name: str, db_source: str, table: str
) -> SparkDataFrame:
"""
Convert a Spark table from i2b2 to OMOP format.
Parameters
----------
db_name: str
Name of the database where the data is stored.
table: str
Name of the table to extract.
Returns
-------
df: Spark DataFrame
Spark DataFrame extracted from the i2b2 database given and converted to OMOP standard.
"""
i2b2_table_name = i2b2_tables[db_source][table]
# Dictionary of omop_col -> i2b2_col
columns = i2b2_renaming.get(table)
# Can be None if creating a table from scratch (e.g. concept_relationship
if columns is not None:
query = f"describe {db_name}.{i2b2_table_name}"
available_columns = set(spark_session.sql(query).toPandas().col_name.tolist())
if db_source == "cse":
columns.pop("i2b2_action", None)
cols = ", ".join(
[
f"{i2b2} AS {omop}"
for omop, i2b2 in columns.items()
if i2b2 in available_columns
]
)
query = f"SELECT {cols} FROM {db_name}.{i2b2_table_name}"
df = spark_session.sql(query)
# Special mapping for i2b2 :
# CIM10
if table == "condition_occurrence":
df = df.withColumn(
"condition_source_value",
F.substring(F.col("condition_source_value"), 7, 20),
)
# CCAM
elif table == "procedure_occurrence":
df = df.withColumn(
"procedure_source_value",
F.substring(F.col("procedure_source_value"), 6, 20),
)
# Visits
elif table == "visit_occurrence":
df = df.withColumn(
"visit_source_value",
mapping_dict(visit_type_mapping, "Non Renseigné")(
F.col("visit_source_value")
),
)
if db_source == "cse":
df = df.withColumn("row_status_source_value", F.lit("Actif"))
df = df.withColumn(
"visit_occurrence_source_value", df["visit_occurrence_id"]
)
else:
df = df.withColumn(
"row_status_source_value",
F.when(
F.col("row_status_source_value").isin([-1, -2]), "supprimé"
).otherwise("Actif"),
)
# Retrieve Hospital trigram
ufr = spark_session.sql(
f"SELECT * FROM {db_name}.{i2b2_tables[db_source]['visit_detail']}"
)
ufr = ufr.withColumn(
"care_site_id",
F.substring(F.split(F.col("concept_cd"), ":").getItem(1), 1, 3),
)
ufr = ufr.withColumnRenamed("encounter_num", "visit_occurrence_id")
ufr = ufr.drop_duplicates(subset=["visit_occurrence_id"])
ufr = ufr.select(["visit_occurrence_id", "care_site_id"])
df = df.join(ufr, how="inner", on=["visit_occurrence_id"])
# Patients
elif table == "person":
df = df.withColumn(
"gender_source_value",
mapping_dict(sex_cd_mapping, "Non Renseigné")(F.col("gender_source_value")),
)
# Documents
elif table.startswith("note"):
df = df.withColumn(
"note_class_source_value",
F.substring(F.col("note_class_source_value"), 4, 100),
)
if db_source == "cse":
df = df.withColumn("row_status_source_value", F.lit("Actif"))
else:
df = df.withColumn(
"row_status_source_value",
F.when(F.col("row_status_source_value") < 0, "SUPP").otherwise("Actif"),
)
# Hospital trigrams
elif table == "care_site":
df = df.withColumn("care_site_type_source_value", F.lit("Hôpital"))
df = df.withColumn(
"care_site_source_value",
F.split(F.col("care_site_source_value"), ":").getItem(1),
)
df = df.withColumn(
"care_site_id", F.substring(F.col("care_site_source_value"), 1, 3)
)
df = df.drop_duplicates(subset=["care_site_id"])
df = df.withColumn(
"care_site_short_name",
mapping_dict(dict_code_UFR, "Non Renseigné")(F.col("care_site_id")),
)
# UFR
elif table == "visit_detail":
df = df.withColumn(
"care_site_id", F.split(F.col("care_site_id"), ":").getItem(1)
)
df = df.withColumn("visit_detail_type_source_value", F.lit("PASS"))
df = df.withColumn("row_status_source_value", F.lit("Actif"))
# measurement
elif table == "measurement":
df = df.withColumn(
"measurement_source_concept_id",
F.substring(F.col("measurement_source_concept_id"), 5, 20),
).withColumn("row_status_source_value", F.lit("Validé"))
# concept
elif table == "concept":
df = (
df.withColumn(
"concept_source_value",
F.substring(
F.col("concept_source_value"), 5, 20
), # TODO: use regexp_extract to take substring after ':'
)
.withColumn("concept_id", F.col("concept_source_value"))
.withColumn("concept_code", F.col("concept_id"))
.withColumn("vocabulary_id", F.lit("ANABIO"))
)
# Adding LOINC
if "get_additional_i2b2_concept" in registry.data.get_all():
loinc_pd = registry.get("data", "get_additional_i2b2_concept")()
assert len(loinc_pd.columns) == len(df.columns)
loinc_pd = loinc_pd[df.columns] # for columns ordering
df = df.union(
spark_session.createDataFrame(loinc_pd, df.schema, verifySchema=False)
).cache()
# fact_relationship
elif table == "fact_relationship":
# Retrieve UF information
df = df.withColumn(
"fact_id_1",
F.split(F.col("care_site_source_value"), ":").getItem(1),
)
df = df.withColumn("domain_concept_id_1", F.lit(57)) # Care_site domain
# Retrieve hospital information
df = df.withColumn("fact_id_2", F.substring(F.col("fact_id_1"), 1, 3))
df = df.withColumn("domain_concept_id_2", F.lit(57)) # Care_site domain
df = df.drop_duplicates(subset=["fact_id_1", "fact_id_2"])
# Only UF-Hospital relationships in i2b2
df = df.withColumn("relationship_concept_id", F.lit(46233688)) # Included in
elif table == "concept_relationship":
data = []
schema = T.StructType(
[
T.StructField("concept_id_1", T.StringType(), True),
T.StructField("concept_id_2", T.StringType(), True),
T.StructField("relationship_id", T.StringType(), True),
]
)
if "get_additional_i2b2_concept_relationship" in registry.data.get_all():
data = registry.get("data", "get_additional_i2b2_concept_relationship")()
df = spark_session.createDataFrame(data, schema).cache()
return df
|