Saved study / v13 · September 2026
Follow the individual response.
15 participants · 3 bands · 5 latent factors
Participant-specific descriptive fits
Loading the saved study…
Standardized residual
Observed minus fitted, divided by the supplied fitted SD.
Move the slider, use its arrow keys, or select a point on a plot. Shaded gaps mark missing observations; the model can still supply a fitted value.
CSV contains the complete selected participant, band, and mode.
Inspect the displayed values
Values are rounded here for readability. The source CSV retains the exported precision. Missing observations are shown as —.
| Time (s) | Phase | Window | Observed | Fitted | Residual / SD |
|---|
How to read this study
Fit quality is one part of the picture.
The model and phase entropy corrections use the participant’s complete accepted trajectory. Both views are descriptive. A free run stops updating the latent state from later observations, but its parameters and entropy corrections still come from this fitted dataset.
The plotted ±2 SD bands use the exported uncertainty. The reported 95.313% entropy coverage is within the fitted data, not an accuracy score for future participants. A coordinate χ² / nominal NDF near one does not establish trajectory prediction.
What stays visible
Gaps, phases, and provenance.
Each band has a separate model. The two cycles belong to one recording; the duplicated second baseline is removed. Rejected windows remain explicit missing positions. p21 was excluded before fitting by the frozen signal-quality gate.
Entropy is measured in nats; its scale depends on the retained electrode dimension. The gap is derived as Sdiag − Sᵥₙ. Latent factors have no assigned biological labels.
Methods, audit note & source report ↗