DAYBREAKNEURAL

A CLEARER START IN NEURAL DECODING

Make sense
of neural data.
Build from there.

A guided workspace for emerging neurotech teams. Try a real EEG example, compare decoding methods, and understand the evidence behind your next step.

Try the working example See how it works

No account needed. EEG processing stays in your browser.

DECODER STUDIOYOUR FIRST WORKSPACE
Included example · Motor imagery

Left hand
or right hand?

Explore whether a model can distinguish two imagined movements from EEG.

  1. 01
    Get to know the dataTrials, channels, labels, quality
  2. 02
    Run a fair comparisonDecoders and a simple baseline
  3. 03
    Understand your resultMeasured performance, with context
Open the interactive workspace

Real examples, clear boundaries

EEG stays on your device

Results you can take with you

FROM FIRST LOOK TO NEXT EXPERIMENT

A path you can follow.
At your own depth.

Start with the essentials. Open the methods and preparation details when you need them. Every result should lead to a better question.

  1. 01

    Start with a question

    See what the example can decode and what it cannot. No prior machine learning experience needed.

  2. 02

    Try real EEG

    EEG measures electrical activity at the scalp. Open the included example of imagined hand movements; no installation or data preparation needed.

    Try it
  3. 03

    Prepare your data

    Follow the format and preprocessing instructions to create a compatible experiment bundle.

    Prepare
  4. 04

    Check the inputs

    Inspect channels, trial counts, labels, and quality checks before interpreting a prediction.

  5. 05

    Compare methods

    Run available decoders locally and compare them against a simple reference method, called a baseline, on the same trials.

  6. 06

    Understand the evidence

    Read accuracy alongside sample counts, errors, and uncertainty. Ask the assistant what a result means.

  7. 07

    Take the next step

    Export your results and use the reproducible preparation and training scripts to continue your work.

HELP, WHERE THE QUESTIONS HAPPEN

An assistant for
understanding.

Ask what a score means, why a baseline matters, or what to check next. The assistant explains the results produced by the workspace; it does not run EEG analyses.

Explore Decoder Studio
EXAMPLE QUESTIONS

“Does this result beat the baseline?”

“What does balanced accuracy mean?”

“What should our team test next?”

Measured results stay distinct from advice.Hosted AI receives the question, recent chat context, and a numerical result summary when you send a message. Raw EEG and imported model files remain local. Guided help is available when AI is unavailable.

WHEN YOU ARE READY FOR YOUR OWN DATA

Bring a compatible
experiment.

Start by reviewing the example. Your own recordings need preparation and a model suited to their channels, task, and preprocessing. The workspace is not a universal EEG reader.

Open data preparation

Before you import

  1. Define the task. Decide which labels you want to distinguish and what a held-out test should represent.
  2. Prepare in Python. Follow the provided script to align channels, sample rate, trial length, and preprocessing with your model.
  3. Keep evaluation separate. Fit preprocessing and choose models using training and validation data. Reserve final test data.
  4. Import the bundle. Load the compatible experiment JSON in Studio, inspect its inputs, and run inference locally.
Preparation guide Python training script

Raw EDF and arbitrary CSV uploads are not supported. Imported models need their own validation; an example result does not validate your data.

BUILT FOR EXPLORATION, WITH THE LIMITS IN VIEW

A useful beginning.
An honest scope.

For teams learning and prototyping

Work through an offline decoding example, discuss results together, and export a record of what you observed. Training custom models happens in Python.

Evidence before claims

A successful demo is not proof of performance for a new person, device, or session. This workspace supports exploration and evaluation, not clinical decisions or live device control.

RESEARCH AT DAYBREAK

Our study archive and open-source Neural Dynamics Explorer live in one place. Read the methods, inspect unfavorable results, and explore the mathematics.

Visit the research section

YOUR NEXT STEP IS A SMALL ONE

Start with an example.
Leave with understanding.

Launch Decoder Studio Talk to our team