POST /api/multimodal-therapy/analyze-session
``` Proxies to AI Services with auth + biomarker persistence + metered credit accounting.
### Request (multipart/form-data)
| Field | Type | Notes |
| :--- | :--- | :--- |
| `session_id` | string | Existing session ID |
| `transcript` | string | Optional |
| `audio_file` | file | WAV / Opus, 16 kHz mono recommended |
| `video_file` | file | MP4 / MOV |
| `duration_seconds` | int | Session duration |
| `patient_id` | string | Patient ID |
### Response
Canonical multimodal payload (see [Multimodal Architecture](/multimodal/architecture)).
Includes `facial`, `vocal`, `rdoc`, and `clinical_context` envelopes.
### Quality gate
20+ words of transcript + 30 s of session signal required. Below threshold the analyzer falls back to Medera reasoning transcript-inferred output with `requires_human_review: true`.
Analyze Video
POST /api/multimodal-therapy/analyze-session
Full multimodal analysis
Backend: