complete response parser
Browse files- responseparser.py +222 -18
responseparser.py
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
import json
|
| 2 |
from datetime import datetime
|
| 3 |
-
from typing import List, Dict, Optional
|
| 4 |
|
| 5 |
class PatientDataExtractor:
|
| 6 |
-
"""Class to extract
|
| 7 |
|
| 8 |
def __init__(self, patient_data: str):
|
| 9 |
"""Initialize with patient data in JSON string format."""
|
|
@@ -36,13 +36,35 @@ class PatientDataExtractor:
|
|
| 36 |
"""Get the currently selected patient resource."""
|
| 37 |
return self.patients[self.current_patient_idx]
|
| 38 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 39 |
def get_first_name(self) -> str:
|
| 40 |
"""Extract patient's first name."""
|
| 41 |
patient = self._get_current_patient()
|
| 42 |
names = patient.get("name", [])
|
| 43 |
for name in names:
|
| 44 |
if name.get("use") == "official" and "given" in name:
|
| 45 |
-
return name["given"][0]
|
| 46 |
return ""
|
| 47 |
|
| 48 |
def get_last_name(self) -> str:
|
|
@@ -60,13 +82,31 @@ class PatientDataExtractor:
|
|
| 60 |
names = patient.get("name", [])
|
| 61 |
for name in names:
|
| 62 |
if name.get("use") == "official" and "given" in name and len(name["given"]) > 1:
|
| 63 |
-
return name["given"][1][0]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
return ""
|
| 65 |
|
|
|
|
| 66 |
def get_dob(self) -> str:
|
| 67 |
"""Extract patient's date of birth."""
|
| 68 |
-
|
| 69 |
-
return patient.get("birthDate", "")
|
| 70 |
|
| 71 |
def get_age(self) -> str:
|
| 72 |
"""Calculate patient's age based on birth date."""
|
|
@@ -78,12 +118,24 @@ class PatientDataExtractor:
|
|
| 78 |
age = today.year - birth_date.year - ((today.month, today.day) < (birth_date.month, birth_date.day))
|
| 79 |
return str(age)
|
| 80 |
|
| 81 |
-
def
|
| 82 |
-
"""Extract patient's
|
|
|
|
|
|
|
|
|
|
|
|
|
| 83 |
patient = self._get_current_patient()
|
| 84 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 85 |
|
| 86 |
-
|
|
|
|
| 87 |
"""Extract patient's street address."""
|
| 88 |
patient = self._get_current_patient()
|
| 89 |
addresses = patient.get("address", [])
|
|
@@ -107,20 +159,173 @@ class PatientDataExtractor:
|
|
| 107 |
addresses = patient.get("address", [])
|
| 108 |
return addresses[0]["postalCode"] if addresses and "postalCode" in addresses[0] else ""
|
| 109 |
|
| 110 |
-
def
|
| 111 |
-
"""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 112 |
return {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 113 |
"first_name": self.get_first_name(),
|
| 114 |
"last_name": self.get_last_name(),
|
| 115 |
"middle_initial": self.get_middle_initial(),
|
|
|
|
|
|
|
| 116 |
"dob": self.get_dob(),
|
| 117 |
"age": self.get_age(),
|
| 118 |
-
"
|
| 119 |
-
"
|
|
|
|
|
|
|
| 120 |
"city": self.get_city(),
|
| 121 |
"state": self.get_state(),
|
| 122 |
"zip_code": self.get_zip_code(),
|
| 123 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 124 |
"doctor_first_name": "",
|
| 125 |
"doctor_last_name": "",
|
| 126 |
"doctor_middle_initial": "",
|
|
@@ -143,20 +348,19 @@ class PatientDataExtractor:
|
|
| 143 |
}
|
| 144 |
|
| 145 |
def get_all_patients(self) -> List[Dict[str, str]]:
|
| 146 |
-
"""Return a list of dictionaries for all patients."""
|
| 147 |
original_idx = self.current_patient_idx
|
| 148 |
all_patients = []
|
| 149 |
for i in range(len(self.patients)):
|
| 150 |
self.set_patient_by_index(i)
|
| 151 |
all_patients.append(self.get_patient_dict())
|
| 152 |
-
self.set_patient_by_index(original_idx)
|
| 153 |
return all_patients
|
| 154 |
|
| 155 |
def get_patient_ids(self) -> List[str]:
|
| 156 |
"""Return a list of all patient IDs in the Bundle."""
|
| 157 |
return [patient["id"] for patient in self.patients]
|
| 158 |
|
| 159 |
-
|
| 160 |
# # Example usage with integration into app.py
|
| 161 |
# def integrate_with_app(patient_data: str):
|
| 162 |
# """Integrate PatientDataExtractor with the Gradio app."""
|
|
|
|
| 1 |
import json
|
| 2 |
from datetime import datetime
|
| 3 |
+
from typing import List, Dict, Optional, Union
|
| 4 |
|
| 5 |
class PatientDataExtractor:
|
| 6 |
+
"""Class to extract all fields from a FHIR Patient resource in a Bundle response."""
|
| 7 |
|
| 8 |
def __init__(self, patient_data: str):
|
| 9 |
"""Initialize with patient data in JSON string format."""
|
|
|
|
| 36 |
"""Get the currently selected patient resource."""
|
| 37 |
return self.patients[self.current_patient_idx]
|
| 38 |
|
| 39 |
+
# Basic Identification Fields
|
| 40 |
+
def get_id(self) -> str:
|
| 41 |
+
"""Extract FHIR Patient ID."""
|
| 42 |
+
return self._get_current_patient().get("id", "")
|
| 43 |
+
|
| 44 |
+
def get_resource_type(self) -> str:
|
| 45 |
+
"""Extract resource type (should always be 'Patient')."""
|
| 46 |
+
return self._get_current_patient().get("resourceType", "")
|
| 47 |
+
|
| 48 |
+
def get_meta_last_updated(self) -> str:
|
| 49 |
+
"""Extract last updated timestamp from meta."""
|
| 50 |
+
return self._get_current_patient().get("meta", {}).get("lastUpdated", "")
|
| 51 |
+
|
| 52 |
+
def get_meta_profile(self) -> List[str]:
|
| 53 |
+
"""Extract profile URIs from meta."""
|
| 54 |
+
return self._get_current_patient().get("meta", {}).get("profile", [])
|
| 55 |
+
|
| 56 |
+
def get_text_div(self) -> str:
|
| 57 |
+
"""Extract generated text narrative (div content)."""
|
| 58 |
+
return self._get_current_patient().get("text", {}).get("div", "")
|
| 59 |
+
|
| 60 |
+
# Name Fields
|
| 61 |
def get_first_name(self) -> str:
|
| 62 |
"""Extract patient's first name."""
|
| 63 |
patient = self._get_current_patient()
|
| 64 |
names = patient.get("name", [])
|
| 65 |
for name in names:
|
| 66 |
if name.get("use") == "official" and "given" in name:
|
| 67 |
+
return name["given"][0]
|
| 68 |
return ""
|
| 69 |
|
| 70 |
def get_last_name(self) -> str:
|
|
|
|
| 82 |
names = patient.get("name", [])
|
| 83 |
for name in names:
|
| 84 |
if name.get("use") == "official" and "given" in name and len(name["given"]) > 1:
|
| 85 |
+
return name["given"][1][0]
|
| 86 |
+
return ""
|
| 87 |
+
|
| 88 |
+
def get_name_prefix(self) -> str:
|
| 89 |
+
"""Extract patient's name prefix (e.g., Mr., Mrs.)."""
|
| 90 |
+
patient = self._get_current_patient()
|
| 91 |
+
names = patient.get("name", [])
|
| 92 |
+
for name in names:
|
| 93 |
+
if name.get("use") == "official" and "prefix" in name:
|
| 94 |
+
return name["prefix"][0]
|
| 95 |
+
return ""
|
| 96 |
+
|
| 97 |
+
def get_maiden_name(self) -> str:
|
| 98 |
+
"""Extract patient's maiden name if available."""
|
| 99 |
+
patient = self._get_current_patient()
|
| 100 |
+
names = patient.get("name", [])
|
| 101 |
+
for name in names:
|
| 102 |
+
if name.get("use") == "maiden" and "family" in name:
|
| 103 |
+
return name["family"]
|
| 104 |
return ""
|
| 105 |
|
| 106 |
+
# Demographic Fields
|
| 107 |
def get_dob(self) -> str:
|
| 108 |
"""Extract patient's date of birth."""
|
| 109 |
+
return self._get_current_patient().get("birthDate", "")
|
|
|
|
| 110 |
|
| 111 |
def get_age(self) -> str:
|
| 112 |
"""Calculate patient's age based on birth date."""
|
|
|
|
| 118 |
age = today.year - birth_date.year - ((today.month, today.day) < (birth_date.month, birth_date.day))
|
| 119 |
return str(age)
|
| 120 |
|
| 121 |
+
def get_gender(self) -> str:
|
| 122 |
+
"""Extract patient's gender."""
|
| 123 |
+
return self._get_current_patient().get("gender", "").capitalize()
|
| 124 |
+
|
| 125 |
+
def get_birth_sex(self) -> str:
|
| 126 |
+
"""Extract patient's birth sex from extensions."""
|
| 127 |
patient = self._get_current_patient()
|
| 128 |
+
for ext in patient.get("extension", []):
|
| 129 |
+
if ext.get("url") == "http://hl7.org/fhir/us/core/StructureDefinition/us-core-birthsex":
|
| 130 |
+
return ext.get("valueCode", "")
|
| 131 |
+
return ""
|
| 132 |
+
|
| 133 |
+
def get_multiple_birth(self) -> Union[bool, None]:
|
| 134 |
+
"""Extract multiple birth status."""
|
| 135 |
+
return self._get_current_patient().get("multipleBirthBoolean", None)
|
| 136 |
|
| 137 |
+
# Address Fields
|
| 138 |
+
def get_address_line(self) -> str:
|
| 139 |
"""Extract patient's street address."""
|
| 140 |
patient = self._get_current_patient()
|
| 141 |
addresses = patient.get("address", [])
|
|
|
|
| 159 |
addresses = patient.get("address", [])
|
| 160 |
return addresses[0]["postalCode"] if addresses and "postalCode" in addresses[0] else ""
|
| 161 |
|
| 162 |
+
def get_country(self) -> str:
|
| 163 |
+
"""Extract patient's country."""
|
| 164 |
+
patient = self._get_current_patient()
|
| 165 |
+
addresses = patient.get("address", [])
|
| 166 |
+
return addresses[0]["country"] if addresses and "country" in addresses[0] else ""
|
| 167 |
+
|
| 168 |
+
def get_geolocation(self) -> Dict[str, float]:
|
| 169 |
+
"""Extract geolocation (latitude and longitude) from address extension."""
|
| 170 |
+
patient = self._get_current_patient()
|
| 171 |
+
addresses = patient.get("address", [])
|
| 172 |
+
if not addresses:
|
| 173 |
+
return {"latitude": None, "longitude": None}
|
| 174 |
+
for ext in addresses[0].get("extension", []):
|
| 175 |
+
if ext.get("url") == "http://hl7.org/fhir/StructureDefinition/geolocation":
|
| 176 |
+
geo = {}
|
| 177 |
+
for sub_ext in ext.get("extension", []):
|
| 178 |
+
if sub_ext.get("url") == "latitude":
|
| 179 |
+
geo["latitude"] = sub_ext.get("valueDecimal")
|
| 180 |
+
elif sub_ext.get("url") == "longitude":
|
| 181 |
+
geo["longitude"] = sub_ext.get("valueDecimal")
|
| 182 |
+
return geo
|
| 183 |
+
return {"latitude": None, "longitude": None}
|
| 184 |
+
|
| 185 |
+
# Contact Fields
|
| 186 |
+
def get_phone(self) -> str:
|
| 187 |
+
"""Extract patient's phone number."""
|
| 188 |
+
patient = self._get_current_patient()
|
| 189 |
+
telecoms = patient.get("telecom", [])
|
| 190 |
+
for telecom in telecoms:
|
| 191 |
+
if telecom.get("system") == "phone" and telecom.get("use") == "home":
|
| 192 |
+
return telecom.get("value", "")
|
| 193 |
+
return ""
|
| 194 |
+
|
| 195 |
+
# Identifiers
|
| 196 |
+
def get_identifiers(self) -> Dict[str, str]:
|
| 197 |
+
"""Extract all identifiers (e.g., SSN, MRN, Driver's License)."""
|
| 198 |
+
patient = self._get_current_patient()
|
| 199 |
+
identifiers = patient.get("identifier", [])
|
| 200 |
+
id_dict = {}
|
| 201 |
+
for id_entry in identifiers:
|
| 202 |
+
id_type = id_entry.get("type", {}).get("text", "Unknown")
|
| 203 |
+
id_dict[id_type] = id_entry.get("value", "")
|
| 204 |
+
return id_dict
|
| 205 |
+
|
| 206 |
+
# Extensions
|
| 207 |
+
def get_race(self) -> str:
|
| 208 |
+
"""Extract patient's race from extensions."""
|
| 209 |
+
patient = self._get_current_patient()
|
| 210 |
+
for ext in patient.get("extension", []):
|
| 211 |
+
if ext.get("url") == "http://hl7.org/fhir/us/core/StructureDefinition/us-core-race":
|
| 212 |
+
for sub_ext in ext.get("extension", []):
|
| 213 |
+
if sub_ext.get("url") == "text":
|
| 214 |
+
return sub_ext.get("valueString", "")
|
| 215 |
+
return ""
|
| 216 |
+
|
| 217 |
+
def get_ethnicity(self) -> str:
|
| 218 |
+
"""Extract patient's ethnicity from extensions."""
|
| 219 |
+
patient = self._get_current_patient()
|
| 220 |
+
for ext in patient.get("extension", []):
|
| 221 |
+
if ext.get("url") == "http://hl7.org/fhir/us/core/StructureDefinition/us-core-ethnicity":
|
| 222 |
+
for sub_ext in ext.get("extension", []):
|
| 223 |
+
if sub_ext.get("url") == "text":
|
| 224 |
+
return sub_ext.get("valueString", "")
|
| 225 |
+
return ""
|
| 226 |
+
|
| 227 |
+
def get_mothers_maiden_name(self) -> str:
|
| 228 |
+
"""Extract patient's mother's maiden name from extensions."""
|
| 229 |
+
patient = self._get_current_patient()
|
| 230 |
+
for ext in patient.get("extension", []):
|
| 231 |
+
if ext.get("url") == "http://hl7.org/fhir/StructureDefinition/patient-mothersMaidenName":
|
| 232 |
+
return ext.get("valueString", "")
|
| 233 |
+
return ""
|
| 234 |
+
|
| 235 |
+
def get_birth_place(self) -> Dict[str, str]:
|
| 236 |
+
"""Extract patient's birth place from extensions."""
|
| 237 |
+
patient = self._get_current_patient()
|
| 238 |
+
for ext in patient.get("extension", []):
|
| 239 |
+
if ext.get("url") == "http://hl7.org/fhir/StructureDefinition/patient-birthPlace":
|
| 240 |
+
addr = ext.get("valueAddress", {})
|
| 241 |
+
return {
|
| 242 |
+
"city": addr.get("city", ""),
|
| 243 |
+
"state": addr.get("state", ""),
|
| 244 |
+
"country": addr.get("country", "")
|
| 245 |
+
}
|
| 246 |
+
return {"city": "", "state": "", "country": ""}
|
| 247 |
+
|
| 248 |
+
def get_disability_adjusted_life_years(self) -> Optional[float]:
|
| 249 |
+
"""Extract disability-adjusted life years from extensions."""
|
| 250 |
+
patient = self._get_current_patient()
|
| 251 |
+
for ext in patient.get("extension", []):
|
| 252 |
+
if ext.get("url") == "http://synthetichealth.github.io/synthea/disability-adjusted-life-years":
|
| 253 |
+
return ext.get("valueDecimal")
|
| 254 |
+
return None
|
| 255 |
+
|
| 256 |
+
def get_quality_adjusted_life_years(self) -> Optional[float]:
|
| 257 |
+
"""Extract quality-adjusted life years from extensions."""
|
| 258 |
+
patient = self._get_current_patient()
|
| 259 |
+
for ext in patient.get("extension", []):
|
| 260 |
+
if ext.get("url") == "http://synthetichealth.github.io/synthea/quality-adjusted-life-years":
|
| 261 |
+
return ext.get("valueDecimal")
|
| 262 |
+
return None
|
| 263 |
+
|
| 264 |
+
# Marital Status
|
| 265 |
+
def get_marital_status(self) -> str:
|
| 266 |
+
"""Extract patient's marital status."""
|
| 267 |
+
patient = self._get_current_patient()
|
| 268 |
+
status = patient.get("maritalStatus", {}).get("text", "")
|
| 269 |
+
return status if status else patient.get("maritalStatus", {}).get("coding", [{}])[0].get("display", "")
|
| 270 |
+
|
| 271 |
+
# Communication
|
| 272 |
+
def get_language(self) -> str:
|
| 273 |
+
"""Extract patient's preferred language."""
|
| 274 |
+
patient = self._get_current_patient()
|
| 275 |
+
comms = patient.get("communication", [])
|
| 276 |
+
return comms[0]["language"]["text"] if comms and "language" in comms[0] else ""
|
| 277 |
+
|
| 278 |
+
# Comprehensive Extraction
|
| 279 |
+
def get_all_patient_data(self) -> Dict[str, Union[str, Dict, List, float, bool, None]]:
|
| 280 |
+
"""Extract all available data for the current patient."""
|
| 281 |
return {
|
| 282 |
+
"id": self.get_id(),
|
| 283 |
+
"resource_type": self.get_resource_type(),
|
| 284 |
+
"meta_last_updated": self.get_meta_last_updated(),
|
| 285 |
+
"meta_profile": self.get_meta_profile(),
|
| 286 |
+
"text_div": self.get_text_div(),
|
| 287 |
"first_name": self.get_first_name(),
|
| 288 |
"last_name": self.get_last_name(),
|
| 289 |
"middle_initial": self.get_middle_initial(),
|
| 290 |
+
"name_prefix": self.get_name_prefix(),
|
| 291 |
+
"maiden_name": self.get_maiden_name(),
|
| 292 |
"dob": self.get_dob(),
|
| 293 |
"age": self.get_age(),
|
| 294 |
+
"gender": self.get_gender(),
|
| 295 |
+
"birth_sex": self.get_birth_sex(),
|
| 296 |
+
"multiple_birth": self.get_multiple_birth(),
|
| 297 |
+
"address_line": self.get_address_line(),
|
| 298 |
"city": self.get_city(),
|
| 299 |
"state": self.get_state(),
|
| 300 |
"zip_code": self.get_zip_code(),
|
| 301 |
+
"country": self.get_country(),
|
| 302 |
+
"geolocation": self.get_geolocation(),
|
| 303 |
+
"phone": self.get_phone(),
|
| 304 |
+
"identifiers": self.get_identifiers(),
|
| 305 |
+
"race": self.get_race(),
|
| 306 |
+
"ethnicity": self.get_ethnicity(),
|
| 307 |
+
"mothers_maiden_name": self.get_mothers_maiden_name(),
|
| 308 |
+
"birth_place": self.get_birth_place(),
|
| 309 |
+
"disability_adjusted_life_years": self.get_disability_adjusted_life_years(),
|
| 310 |
+
"quality_adjusted_life_years": self.get_quality_adjusted_life_years(),
|
| 311 |
+
"marital_status": self.get_marital_status(),
|
| 312 |
+
"language": self.get_language()
|
| 313 |
+
}
|
| 314 |
+
|
| 315 |
+
def get_patient_dict(self) -> Dict[str, str]:
|
| 316 |
+
"""Return a dictionary of patient data mapped to discharge form fields (for app.py compatibility)."""
|
| 317 |
+
patient_data = self.get_all_patient_data()
|
| 318 |
+
return {
|
| 319 |
+
"first_name": patient_data["first_name"],
|
| 320 |
+
"last_name": patient_data["last_name"],
|
| 321 |
+
"middle_initial": patient_data["middle_initial"],
|
| 322 |
+
"dob": patient_data["dob"],
|
| 323 |
+
"age": patient_data["age"],
|
| 324 |
+
"sex": patient_data["gender"],
|
| 325 |
+
"address": patient_data["address_line"],
|
| 326 |
+
"city": patient_data["city"],
|
| 327 |
+
"state": patient_data["state"],
|
| 328 |
+
"zip_code": patient_data["zip_code"],
|
| 329 |
"doctor_first_name": "",
|
| 330 |
"doctor_last_name": "",
|
| 331 |
"doctor_middle_initial": "",
|
|
|
|
| 348 |
}
|
| 349 |
|
| 350 |
def get_all_patients(self) -> List[Dict[str, str]]:
|
| 351 |
+
"""Return a list of dictionaries for all patients (for app.py)."""
|
| 352 |
original_idx = self.current_patient_idx
|
| 353 |
all_patients = []
|
| 354 |
for i in range(len(self.patients)):
|
| 355 |
self.set_patient_by_index(i)
|
| 356 |
all_patients.append(self.get_patient_dict())
|
| 357 |
+
self.set_patient_by_index(original_idx)
|
| 358 |
return all_patients
|
| 359 |
|
| 360 |
def get_patient_ids(self) -> List[str]:
|
| 361 |
"""Return a list of all patient IDs in the Bundle."""
|
| 362 |
return [patient["id"] for patient in self.patients]
|
| 363 |
|
|
|
|
| 364 |
# # Example usage with integration into app.py
|
| 365 |
# def integrate_with_app(patient_data: str):
|
| 366 |
# """Integrate PatientDataExtractor with the Gradio app."""
|