patients
Generate patient memory context for AI assistants
Changed onReturns live operational facts plus any durable PatientMemory narrative. Used by realtime-agent-service (generateMemory) for voice and SMS assistants.
Live facts are queried on demand. Durable sections are the latest memory-service rebuild, if one exists.
Scopes:
- full: Live referrals, appointments, calls, faxes, balance + all durable sections
- clinical: Live referrals and faxes + durable clinical
- scheduling: Live referrals needing scheduling + appointments
- other: Live calls and balance + durable preferences, interactions, administrative
Output:
- json: Structured live + durable payload
- markdown: Deterministic live block, then durable notes
Channel only affects date formatting in markdown:
- voice: August 20, 2026
- sms: 08-20-2026
get/patients/{patientId}/memory
Request
- Base URLhttps://api.lumahealth.io/api/v2
- AuthHTTP bearer
- TypeScriptA function with typed inputs. Its response is checked against the schema.
Path parameters
patientIdstring required
Patient ID
Query parameters
scope'full' | 'clinical' | 'scheduling' | 'other'
Data scope to retrieve
output'json' | 'markdown'
Output format
channel'voice' | 'sms'
Communication channel (affects markdown date formatting)
Response
Patient memory generated successfully
Example response
{
"memory": "Patient: John Smith\nDOB: 01-01-1990\n\n## Referrals\n- id 123: with provider Jane Doe (needsScheduling)\n\n## Notes\n### Preferences\nPrefers SMS.\n"
}Changes
- info
endpoint added
- info