> ## Documentation Index
> Fetch the complete documentation index at: https://docs.medera.info/llms.txt
> Use this file to discover all available pages before exploring further.

# Multimodal Release Notes

> Vocal, facial, Neurobehavioral Construct engines

## Engine baseline

* **Facial Physiological Engine** — `src/multimodal/engines/facial_physiological_engine.py` — HR \< 5 bpm MAE, HRV (SDNN, RMSSD, LF/HF, SD1, SD2), BP, RR, stress (PNS / SNS)

* **Vocal Acoustic Engine** — `src/multimodal/engines/vocal_acoustic_engine.py` — F0 with gender-aware range, jitter (\< 1.04 % normal), shimmer (\< 3.81 % normal), HNR (> 20 dB normal), MFCC (13), prosody, clinical markers

* **Neurobehavioral Construct Computer** — `src/rdoc/rdoc_construct_computer.py` — 15 named constructs across 5 domains with zero overlap

## Confidence calibration

`ConfidenceLevel` enum: `VERY_HIGH` ≥ 0.95, `HIGH` ≥ 0.85, `MODERATE` ≥ 0.70, `LOW` ≥ 0.50, `VERY_LOW` \< 0.50. Below MODERATE the result is delivered with `requires_human_review: True`.