What it does
@elata-biosciences/biosignal-analytics turns biosignal data into features and
scores in the browser, with no server or network connection:
- EEG window features (WASM): band powers, spectral entropy, dominant frequency, alpha peak, Hjorth parameters, and quality flags
- HRV and robust statistics: time-domain HRV, medians and MADs, rolling personal baselines
- Headline scores: Measurement Quality, Activation, Recovery, Focus, Readiness, and Resilience, each with its contributors exposed
- A metric registry: every metric has a registered definition, unit, evidence tier, and algorithm version
biosignal-session,
but doesn’t require it.
Analyze an EEG window
Headline scores
Scores are interpretations, not measurements, and they say so:- Measurement Quality describes the recording, not the person, and is always available.
- Activation, Recovery, Focus, Readiness, and Resilience each list every contributor with its weight, z-score, and quality.
- A score withholds itself (returns
nullwith a reason) rather than guessing when inputs are missing, quality is poor, or there’s no personal baseline yet.
insufficient_history includes counts, so you can tell the user exactly how
many more days are needed. Readiness needs 14 qualified days; Resilience needs
21 plus 6 recovered activation episodes.
Focus is deliberately not based on the theta/beta ratio, which isn’t a valid
attention measure.
Run analysis off the main thread
Entry points
Feature code is verified against Python (NumPy/SciPy) reference fixtures, so the
Rust, WASM, and TypeScript implementations agree.