Synthesize

The Synthesize API returns a summary of guidelines and research for any clinical query

How it works

1

Craft query with or without patient context

2

Retrieve response with references & scores

API specifications

Initiate a synthesis

Initiate synthesis

post

https://cg-api-prod.system.comhttps://cg-api-stage.system.comhttps://cg-api-dev.system.comhttp://localhost:80

/v1/synthesis

Initiate synthesis.

Authorizations

APIKeyHeader

x-api-keystringRequired

Body

application/json

Payload to kickoff synthesis generation.

user_querystringRequired

User query.

population_filterany ofOptional

or

nullOptional

publication_filterany ofOptional

or

nullOptional

fast_modeany ofOptional

Fast mode flag for fast synthesis generation.

Default: false

booleanOptional

or

nullOptional

assembled_patientany ofOptional

or

nullOptional

Responses

202

Job ID

application/json

204

No content

422

Validation Error

application/json

503

Service Unavailable

post

/v1/synthesis

HTTPcURLJavaScriptPython

POST /v1/synthesis HTTP/1.1
Host: cg-api-prod.system.com
x-api-key: YOUR_API_KEY
Content-Type: application/json
Accept: */*
Content-Length: 1050

{
  "user_query": "text",
  "population_filter": {
    "age": 1,
    "sex": "female",
    "race": "american indian or alaskan native"
  },
  "publication_filter": {
    "publication_date": {
      "start_year": 1,
      "end_year": 1
    },
    "publication_type": [
      "experimental trial"
    ],
    "publication_quality": "high"
  },
  "fast_mode": false,
  "assembled_patient": {
    "nodes": [
      {
        "node_type": "text",
        "fhir_id": "text",
        "notable": true,
        "system_id": "text",
        "system_preferred_name": "text",
        "other_system_ids": [
          "text"
        ],
        "coding": [
          {
            "code": "text",
            "system": "text",
            "display": "text"
          }
        ],
        "metadata": {
          "condition_metadata": {
            "clinicalStatus": "active",
            "family_history": true
          },
          "observation_metadata": {
            "interpretation": "carrier",
            "risk": "optimal",
            "risk_direction": "high",
            "observation_time": "2026-03-13T09:16:48.197Z"
          },
          "intervention_metadata": {
            "intervention_type": "medication",
            "status": "active",
            "intervention_time": "2026-03-13T09:16:48.197Z"
          },
          "allergy_metadata": {
            "clinicalStatus": "active",
            "verificationStatus": "unconfirmed",
            "criticality": "text"
          }
        }
      }
    ],
    "relationships": [
      {
        "source": "text",
        "target": "text"
      }
    ],
    "demographics": {
      "age": 1,
      "gender": "text",
      "race": "text"
    }
  }
}

text

Retrieve response from job_id

Get synthesis by ID

get

https://cg-api-prod.system.comhttps://cg-api-stage.system.comhttps://cg-api-dev.system.comhttp://localhost:80

/v1/synthesis/{job_id}

Get synthesis.

Authorizations

APIKeyHeader

x-api-keystringRequired

Path parameters

job_idstringRequired

Responses

200

Successful Response

application/json

Synthesis fetching output schema.

statusstring · enumRequired

Polling status enum.

Possible values:

synthesisany ofRequired

or

nullOptional

422

Validation Error

application/json

get

/v1/synthesis/{job_id}

HTTPcURLJavaScriptPython

GET /v1/synthesis/{job_id} HTTP/1.1
Host: cg-api-prod.system.com
x-api-key: YOUR_API_KEY
Accept: */*

{
  "status": "not_exists",
  "synthesis": {
    "synthesis": {
      "text": "text",
      "references": [
        {
          "index": 1,
          "reference_text": "text",
          "document_title": "text",
          "authoring_societies_or_journal": [
            "text"
          ],
          "publish_date": "text",
          "link_or_doi": "text",
          "cited_by": 1,
          "source_type": "text",
          "authors": [
            "text"
          ],
          "journal_quality": "text"
        }
      ]
    },
    "response_flags": {
      "contains_guideline": true,
      "contains_trial": true,
      "quality": "low",
      "quantity": "limited",
      "relevance": "unanswered",
      "confidence_score": "low"
    },
    "synthesis_logs": {
      "total_time_seconds": 1,
      "guideline_search_time_seconds": 1,
      "guideline_lookup_time_seconds": 1,
      "evidence_retrieval_time_seconds": 1,
      "blending_time_seconds": 1
    }
  }
}

Using Assembled Patient as Context

The assembled_patient field in the synthesis request body allows you to pass a structured patient graph as clinical context. When provided, the synthesis engine uses the patient's conditions, observations, interventions, allergies, and demographics to personalize and focus the evidence synthesis to be directly relevant to that specific patient's clinical profile.

How It Works

The assembled_patient field accepts an AssembledPatient object — the same output structure returned by the POST /v1/assemble endpoint. This creates a natural integration pattern:

CopyPOST /v1/assemble (FHIR bundle) → AssembledPatient ↓ POST /v1/synthesis (user_query + assembled_patient) → Personalized Synthesis

The Assemble endpoint converts a FHIR bundle into a structured patient graph of typed nodes and edges. Passing this output directly into the synthesis request contextualizes the evidence retrieval and blending pipeline around the patient's specific clinical picture.

AssembledPatient Structure

Field	Type	Required	Description
nodes	AssembleNode[]	Yes	Patient clinical nodes (conditions, observations, interventions, allergies)
relationships	AssembleEdge[]	Yes	Edges connecting nodes in the patient graph
demographics	PatientDemographics	Yes	Patient age, gender, and race

AssembleNode Fields

Field	Type	Required	Description
node_type	string	Yes	Discriminator for the node subtype (e.g., condition, observation)
fhir_id	string	Yes	Original FHIR resource identifier
notable	boolean	Yes	Whether this node is clinically notable/important
system_id	string	Yes	System-assigned identifier for the concept
system_preferred_name	string	Yes	Preferred clinical name of the concept
other_system_ids	string[]	No	Additional system identifiers
coding	ExternalCoding[]	No	External coding references (e.g., SNOMED, ICD-10)
metadata	AssembleNodeMetadata	No	Type-specific clinical metadata

AssembleNodeMetadata

Each node can carry rich clinical metadata depending on its type:

Metadata Type	Key Fields	Description
condition_metadata	clinicalStatus, family_history, ccsr_categories, onsetDateTime	For diagnosis/condition nodes
observation_metadata	risk, risk_direction, interpretation, observation_time, valueQuantity	For lab results and clinical observations
intervention_metadata	intervention_type, status, intervention_time	For medications, procedures, and immunizations
allergy_metadata	clinicalStatus, criticality, verificationStatus	For allergy and intolerance nodes

AssembleEdge Fields

Field	Type	Required	Description
source	string	Yes	system_id of the source node
target	string	Yes	system_id of the target node
relationship_id	string	Yes	Identifier of the relationship type between nodes

Data security

Synthesize can be integrated into electronic health records (EHRs) to receive de-identified patient data as context for clinical queries. No requests include patient identifiers.