Standards and roadmap

Each row says what exists today and names the next concrete step. "Planned" means not built.

In place

Planned

Lab Streaming Layer (LSL) and XDF

LSL is the standard way to time-align several devices. One example is this ECG beside a phone magnetometer.

• Next step: a pylsl outlet in the bridge that publishes the sample stream with the host timestamp it already takes for timebase.py. Recording would be done with LabRecorder to XDF.
• Validate the way the LSL project does: send one hardware pulse into both streams and measure the offset. Until that pulse test is run, any claim that two streams are aligned is untested.

NeuroKit2 benchmark for ECG

NeuroKit2 publishes benchmarks for R-peak detection. This repo does not detect R peaks yet.

• Next step: if an ECG readout gains heart-rate analysis, use nk.ecg_peaks and report its published sensitivity and positive predictivity beside results from this board. Do not write a new detector.

MNE data objects and per-session QC reports

BIDS exports already open in MNE.

• Next step: an mne.Report per session containing:
  • the raw trace;
  • the PSD, showing the analog band and mains;
  • the contact verdict from /quality;
  • the measured sample rate.

A device-validation write-up (after Krigolson et al., 2017)

Krigolson et al. validated a consumer EEG headset (the Muse) by recovering known ERP components against published values. The same shape of study for this board:

1. Record a positive control whose answer is known, on this board and on a reference amplifier at the same time:
   • a 40 Hz auditory steady-state response;
   • an N1/P2 to tones;
   • eyes-closed alpha.
2. Report the agreement, the noise floor (apps/server/validation/VALIDATION.md already has it: median 157 nV at 40 Hz over 60 s; corrected 2026-09-13 from 238) and the measured clock error.
3. Pre-register the analysis as an experiment (see versioning.md).

Nothing here has yet recorded a response to a known stimulus. The noise is characterised; detection is not shown.

JOSS / pyOpenSci review

Both review checklists ask for:

• an installable package;
• documented functionality with examples;
• automated tests;
• a statement of need;
• community guidelines.

The last three exist. The package and the statement of need do not.

Not applicable here

Motion-BIDS (including its MAGN magnetometer channel type) is for motion and phone sensor recordings. It applies to a phone magnetometer recorder, not to this board.