---
language:
- en
license: mit
library_name: scikit-learn
pipeline_tag: tabular-regression
tags:
- healthcare
- physiology
- wearable
- engagement
---

# MindPulse Engagement Model

This model predicts child engagement score (`0..1`) from wearable physiological features:

- `heart_rate`
- `hrv_rmssd`
- `motion_level`
- `hr_baseline`
- `rmssd_baseline`

## Training Target

- `engagement_score`

## Data Processing Logic

- `HR_norm = (heart_rate - hr_baseline) / hr_baseline`
- `RMSSD_norm = hrv_rmssd / rmssd_baseline`
- `arousal = normalize(HR_norm)`
- `valence = normalize(RMSSD_norm)`
- `engagement_score = arousal * valence` (bounded to `0..1`)

## Usage

Load `mindpulse_rf.joblib` and run prediction with all 5 input features.

Repository: https://huggingface.co/AtharvaXX/mindpulse