Extract multiple fields at different levels of nesting json using chatgpt and jmespath
- 4 minutes read - 656 wordsHow do you extract multiple fields at different levels of nesting json resources? jsonpath, jq, jmespath or chatgpt?
In FHIR, bundle resource in json format usually have multiple level of details nested in json. It’s quite different from other use cases.
My first thought is using jsonpath as it is quite similar to FHIRPath. There are several limitations of jsonpath which make it infeasible.
-
It doesn’t support to the use case, even though JsonPath supports deep scan and Bracket-notated child or children. It supports to access multiple fields at one level only.
-
No specification. Some path expressions work on one implemention, not others.
-
Not an optimal solution. I can loop throught expressions to extract multiple values in several rounds. In real projects, it will complicate implemetations and other things.
My second thought is jq. It is the most powerful one to my knowledge, However it might be hard to integrate with systems as jq is a command line tool.
The third one is jmespath. I worked out an jmespath expression to extract care plan id, first name and last name of patients from a json. In FHIR, bundle json might have different content. A generalized solution will be good. GenAI is a good fit for that, especially chatgpt.
The final one is chatgpt. Below are the three prompts I used. The third prompt will get an acceptable answer from chatgpt.
Prompts:
-
Extract care plan id and the name of patient’s name from the below json using jmespath
-
Extract care plan id and the name of patient’s name from the below json using jmespath, put them in an object
-
Extract care plan id, the first name and given name of patient from the below json using jmespath, put them in an object
The acceptable answer of the third prompt is as following:
//answer
{
"planId": "53333",
"patient_name": [
{
"family": "Smith",
"given": [
"James"
]
}
]
}You can check the answers and json resource at The jmespath example from chatgpt
Bundle json
//bundle json
{
"resourceType": "Bundle",
"id": "98d056c8-f5d7-4d1d-ae7d-d1bb30317747",
"meta": {
"lastUpdated": "2024-03-02T13:31:11.945+00:00"
},
"type": "searchset",
"total": 1,
"link": [
{
"relation": "self",
"url": "https://hapi.fhir.org/baseR4/CarePlan?_format=json&_id=53333&_include=CarePlan%3Acare-team&_include=CarePlan%3Apatient&_pretty=true"
}
],
"entry": [
{
"fullUrl": "https://hapi.fhir.org/baseR4/CarePlan/53333",
"resource": {
"resourceType": "CarePlan",
"id": "53333",
"meta": {
"versionId": "1",
"lastUpdated": "2019-11-01T21:37:05.589+00:00",
"source": "#Pzd7PO3A3sZn4z63"
},
"status": "active",
"intent": "proposal",
"subject": {
"reference": "Patient/53254",
"type": "Patient"
},
"activity": [
{
"outcomeCodeableConcept": [
{
"coding": [
{
"code": "6025007",
"display": "Laparoscopic appendectomy"
},
{
"system": "http://snomed.info/sct",
"code": "6025007",
"display": "Laparoscopic appendectomy"
}
]
}
],
"outcomeReference": [
{
"reference": "Observation/53330",
"type": "Observation"
},
{
"reference": "Procedure/53331",
"type": "Procedure"
},
{
"reference": "Medication/9352fd3a-d63e-4957-a57d-daf4878d2385",
"type": "Medication"
}
],
"detail": {
"code": {
"coding": [
{
"system": "http://snomed.info/sct",
"code": "6025007",
"display": "Laparoscopic appendectomy"
}
]
},
"status": "completed"
}
},
{
"detail": {
"code": {
"coding": [
{
"system": " http://www.nlm.nih.gov/research/umls/rxnorm",
"code": "258395",
"display": "Lisinopril"
}
]
},
"status": "in-progress"
}
},
{
"detail": {
"status": "unknown"
}
}
]
},
"search": {
"mode": "match"
}
},
{
"fullUrl": "https://hapi.fhir.org/baseR4/Patient/53254",
"resource": {
"resourceType": "Patient",
"id": "53254",
"meta": {
"versionId": "4",
"lastUpdated": "2023-09-29T00:45:23.054+00:00",
"source": "#yv8lN6mkhf7deTiS"
},
"text": {
"status": "generated",
"div": "<div xmlns=\"http://www.w3.org/1999/xhtml\"><div class=\"hapiHeaderText\">James <b>SMITH </b></div><table class=\"hapiPropertyTable\"><tbody><tr><td>Identifier</td><td>110107073916280</td></tr><tr><td>Address</td><td><span>1234 Six Forks </span><br/><span>Cary </span><span>NC </span><span>US </span></td></tr><tr><td>Date of birth</td><td><span>01 September 1965</span></td></tr></tbody></table></div>"
},
"identifier": [
{
"use": "usual",
"system": "1.3.6.1.4.1.22812.3.99930.3",
"value": "110107073916280"
},
{
"use": "official",
"system": "https://www.edifecs.com/fhir/patient/identifier",
"value": "74306e5e90359c6574c90a69b2e9325c1106c44a22d9733a67eaa53412bb3251"
}
],
"active": true,
"name": [
{
"family": "Smith",
"given": [
"James"
]
}
],
"telecom": [
{
"system": "phone",
"value": "tel:+1-(919)854-3333",
"use": "home"
},
{
"value": "tel:+1-(919)854-4444",
"use": "home"
}
],
"gender": "male",
"birthDate": "1965-09-01",
"deceasedBoolean": false,
"address": [
{
"line": [
"1234 Six Forks"
],
"city": "Cary",
"state": "NC",
"postalCode": "27513",
"country": "US"
}
],
"contact": [
{
"relationship": [
{
"coding": [
{
"display": "Wife"
}
]
}
],
"telecom": [
{
"value": "7894561235"
}
]
}
],
"communication": [
{
"preferred": true
}
],
"managingOrganization": {
"reference": "Organization/53256",
"type": "Organization"
}
},
"search": {
"mode": "include"
}
}
]
}