4. Engagement & Focus
Track your Engagement Index (EI) in real time and calibrate a binary focus classifier.
Engagement Index Trend
EI = β / (α + θ) — higher EI = more engaged.
| Element | Meaning |
|---|---|
| Blue line | EMA-smoothed EI |
| Red horizontal line | Alert threshold (default 0.3) |
| Red curve segment | EI below threshold |
| Δ 30S / 30S AVG | Change and average over last 30s |
FFT window: 2s (500 samples), hop: 0.5s, EMA α = 0.1. First 30s excluded. Reference: Pope et al. (1995).
Focus Classification
A 4-state machine: IDLE → Waiting Warmup (30s) → Collecting Baseline (15s) → Active.
During baseline, choose a calibration video. The median EI is captured as the reference. Every decision window (default 15s), the current EI median is compared against the baseline → Focused or Not focused.
Reference EI is editable after baseline completes.
Tuning
| Parameter | Default | Env Variable |
|---|---|---|
| Warmup | 30s | VITE_FOCUS_WARMUP |
| Baseline window | 15s | VITE_FOCUS_BASELINE |
| Decision window | 15s | VITE_FOCUS_DECISION |
Next
→ Ask AI to interpret your EEG data → Configure tuning parameters