API Fetcher¶
Fetches data from healthcare-focused APIs with optional presets for common FHIR and medical endpoints. Supports single or parallel requests, simple Bearer token authentication, automatic response format detection (JSON/XML/text), and basic retry handling. Returns response data, status information, headers, and timing metrics.
Usage¶
Use this node to retrieve healthcare data (e.g., FHIR resources or clinical trial data) either as a one-off request or as multiple requests executed in parallel. Select a preset endpoint or provide a custom URL, optionally add an auth token and JSON-formatted query parameters, and set max items to limit large JSON arrays. In parallel mode, supply a JSON array of request configs to fan out multiple calls and aggregate results.
Inputs¶
| Field | Required | Type | Description | Example |
|---|---|---|---|---|
| preset_endpoint | True | CHOICE | Select a preconfigured healthcare API or choose 'custom' to use your own URL. Presets include: fhir_patients, fhir_observations, fhir_medications, fhir_conditions, fhir_diagnostic_reports, covid_tracker, drug_interaction_checker, medical_codes_icd10, clinical_trials. | fhir_patients |
| api_url | True | STRING | The API URL to fetch data from when 'preset_endpoint' is set to 'custom'. Ignored for non-custom presets. | https://hapi.fhir.org/baseR4/Patient |
| execution_mode | True | CHOICE | Choose 'single' for a single request or 'parallel' to execute multiple requests concurrently using 'batch_requests' and 'max_workers'. | single |
| max_workers | False | INT | Number of concurrent workers for parallel execution. Used only when execution_mode is 'parallel'. | 4 |
| batch_requests | False | STRING | JSON array describing multiple requests for parallel mode. Each item can include 'url' and optional 'params' (JSON object string). If a preset is selected (not 'custom'), 'url' can be omitted and the preset URL is used. | [{"url": "https://hapi.fhir.org/baseR4/Patient", "params": "{\"_count\": 5}"}, {"url": "https://clinicaltrials.gov/api/v2/studies", "params": "{\"query\": \"diabetes\"}"}] |
| auth_token | False | STRING | Authentication token sent as a Bearer token in the Authorization header. | |
| query_params | False | STRING | JSON object string of query parameters merged into the request URL. Applies to single request mode and as default params for batch items that omit 'params'. | {"_count": 10, "name": "John"} |
| max_items | False | INT | Maximum number of items to return when the response body is a JSON array (single request mode only). | 10 |
Outputs¶
| Field | Type | Description | Example |
|---|---|---|---|
| response_data | STRING | The fetched data as a string. Auto-formatted for JSON (pretty-printed), or raw text/XML depending on response. | { "resourceType": "Bundle", "entry": [ ... ] } |
| status_message | STRING | Human-readable status of the fetch, including item counts or errors. | Successfully fetched 10 items from API |
| success | BOOLEAN | Indicates whether the request(s) completed successfully. | True |
| status_code | INT | HTTP status code for the request. In parallel mode, returns 200 if at least one request succeeded. | 200 |
| response_headers | STRING | Response headers as a JSON string. In parallel mode, contains an execution log JSON object. | {"content-type": "application/fhir+json;charset=utf-8"} |
| response_time | FLOAT | Total elapsed time in seconds for the operation. | 0.84 |
Important Notes¶
- Preset handling: If 'preset_endpoint' is not 'custom', the preset URL is used and 'api_url' is ignored.
- Query parameters: 'query_params' must be a valid JSON object string (e.g., "{\"_count\": 10}"). Invalid JSON is ignored with a warning.
- Authentication: 'auth_token' is included as 'Authorization: Bearer
'. No other auth types are supported in this simplified interface. - Parallel mode inputs: 'batch_requests' must be a JSON array. Each item supports 'url' and 'params' (JSON object string).
- Timeouts and retries: Each request uses a ~30s timeout and basic retry logic (up to 3 attempts) on transient errors.
- Response processing: The node auto-detects response format. JSON arrays may be truncated to 'max_items' in single mode.
- Headers output: For single requests, headers are returned as a JSON string. For parallel execution, 'response_headers' contains a JSON object with an 'execution_log' array.
- Success in parallel: 'success' is true if at least one request succeeds; review the execution log for per-request outcomes.
- HIPAA considerations: When fetching patient-related data, ensure appropriate authorization and compliant handling of sensitive information.
Troubleshooting¶
- Invalid query_params: If query parameters are not valid JSON, they will be ignored. Provide a proper JSON object string (e.g., "{\"_count\": 10}").
- Invalid batch_requests: Ensure 'batch_requests' is a JSON array string. Each element should include at least 'url' (for custom) or rely on a preset plus optional 'params'.
- Unauthorized or 401/403: Verify 'auth_token' is correct and has necessary scopes. Place the literal token value in the field (e.g., "
"). - Timeouts: Large or slow endpoints may exceed the ~30s timeout. Reduce payload via query params, lower 'max_items', or split requests in parallel.
- Mixed outcomes in parallel: If some requests fail, 'success' may still be true. Inspect 'response_headers' (execution_log) and individual result entries in 'response_data'.
- Wrong endpoint used: When using a preset, 'api_url' is ignored. Set 'preset_endpoint' to 'custom' to use your custom URL.
- Unexpected text/XML: If JSON is expected but the API returns text/XML, verify endpoint and headers. The node does not force content-type.