BCINexus · Start here
What BCINexus is
BCINexus is the publishing and peer-review layer for biosignal research. BCILattice is the desktop application where the research actually happens. This page explains the split, because almost every other question depends on it.
Two halves of one workflow
BCILattice runs on your machine. It imports recordings, cleans and epochs them, extracts features, builds pipelines, and trains models — all locally, against files on your own disk. BCINexus is the account you sign into from the app or the website. It holds nothing of your raw data until you deliberately publish or share something.
| BCILattice (desktop) | BCINexus (platform) | |
|---|---|---|
| Runs | On your computer | In the browser and behind the app |
| Holds | Your recordings, studies, models, reports | What you chose to publish or share |
| Used for | Analysis, feature extraction, ML training | Publishing, peer review, citation, collaboration |
| Needs an account | Yes — always | Yes — always |
| Needs internet | Yes — there is no offline mode | Yes |
| Docs | BCILattice track | This track |
BCILattice is not an offline application
The processing is local, but the connection is not optional. An active account is required to run it, and paid features are unlocked by a licence check against the server rather than by anything stored on your disk. Lose the connection mid-session and the app keeps working, with paid features surviving about ten minutes on the current licence token before dropping to Free until the check succeeds again.
Recorded data stays local by default
Nothing you import or record in BCILattice is uploaded because you happen to be signed in. Data leaves your machine only through an explicit action: publishing a study, publishing a dataset, sharing to a team, or creating a share link. If you never take one of those actions, the platform never sees the files.
The lifecycle of a piece of work
- 1
Build it locally
Create a study in BCILattice, import your recordings, run the Analysis Suite, and train in the ML Suite. Everything at this stage is a local file.
- 2
Submit it
When you publish, the platform first runs a readiness preflight — the same rule set the desktop dialog shows you — and refuses anything missing the minimum bar. What passes enters review.
- 3
Get it reviewed
Eligible reviewers pick the submission up from a pool, comment on it, and record a stage decision. You see the comments they choose to share, and you can resubmit a new version in response.
- 4
Have it published and cited
On final approval the study becomes publicly visible, joins the community record and the knowledge graph, and can be found, reproduced, benchmarked, and cited by other people.
The things you can publish
- Study
- A piece of research: metadata, pipeline definitions, experiments, metrics, and reports. The main unit of publication and the thing peer review acts on.
- Dataset
- Recordings packaged for reuse, with versions and a required paradigm that describes the task and labels. Datasets go through the same publication state machine as studies.
- Model
- Trained weights plus the graph that produced them, published to the registry so someone else can load and evaluate them.
- Benchmark record
- A scored result on a shared task, so approaches can be compared on the same footing rather than on prose claims.
Who does what
- Researchers build and submit work, respond to review, and answer questions on their published studies.
- Reviewers are researchers who applied and were approved. They pick up submissions, leave structured comments, and record decisions.
- Team and lab members share studies and data inside a private workspace before anything is public.
- Admins run the moderation queue, manage reviewer pools and review pipelines, and handle integrity reports.
Where to go next
Something missing or out of date? Email support or request a doc page. Include the page name and what you expected to find.