Data import, batches and processing logs

First select a valid study folder, input format and analysis. Check the working settings before opening data or starting a batch.

Open one recording

Use Open datafile to start the selected analysis on one file. Depending on the input and analysis, EMEGS asks which signals to analyse, display or ignore. Check channel names, units, sample rates and sensor configuration rather than assuming an automatic selection is correct.

For supported single-channel analyses, configured labels can express a reference as ECG1---ECG2. This is not the EEG rereferencing workflow: for ERP/ERF, use the sensor configuration and continuous preparation.

A format appearing in a menu does not imply that every vendor/version or every analysis supports it. Unsupported inputs should be converted through a supported path, not renamed to an EMEGS extension.

Build a batch

The Batch Manager separates Search, Batch and Reports controls. Choose the search folder, whether to search recursively, and whether to exclude ANOVA/STATS directories. Review the actual result rows before processing.

• General files searches filename or relative-path patterns. Add/remove filter rows as needed; there is no fixed twelve-filter limit.
• SCADS condition files replaces *** in the pattern with each condition from the ordered condition table. For example, *at***.ar creates one block per condition. Known configuration/text/JSON sidecars are not datasets.
• EWH5 condition files searches containers and expands their internal conditions into logical dataset references.

With recursive search, a relative-path pattern such as **/branches/short-epochs/*.SES.APP.SES can select one processing branch. Patterns are relative to the search folder. Use forward slashes in these examples on both Linux and Windows. Narrow broad patterns when several processing variants of the same recording exist.

The factorial condition generator supports one to four factors, one digit per factor, with the last factor varying fastest. A 3 × 3 table is 11, 12, 13, 21, 22, 23, 31, 32, 33. Verify that this order matches the design used by the later calculation.

Build matched condition batches… prepares subject-matched comparison pairs or condition blocks for averaging. It checks subject identity rather than assuming that filesystem order matches subjects. Missing or duplicate matches prevent batch creation. Generated batches are preparation, not proof that a statistical calculation or its scientific design is valid.

Run a batch or selected rows

Use Run batch processing from the Open-datafile action menu to process the current batch. Run batch on selected files processes only highlighted rows, in table order, without changing the full list. Save your workspace to preserve the batch and relevant selection state.

Check the file count and first filename in the log. For a new setup, complete one representative recording before launching a long batch.

ERPERF has independent controls for the processing range, interaction mode and reuse of compatible outcomes. Unattended is not the same as “skip existing work”, and selecting a decision step does not itself request a review dialog. See ERPERF flow control. Other analyses have their own prompt and reuse controls.

Progress, Stop and continuation

Watch the processing log for the current file, stage, warnings and output paths. Use Stop to request cancellation and allow the worker to reach a safe stopping point. Do not start another run until it has stopped. Where an analysis explicitly pauses for inspection, use its Resume action after completing the decision.

Restarting the application does not resume a worker. Select a supported processing range and let that analysis validate its persisted prerequisites. Different analyses have different resume/checkpoint contracts.

Logs and output safety

The Log dock shows processing messages; the study’s qemegslog.html retains timestamped analysis logs. An ending batch summary is not proof that every file succeeded: inspect individual errors and outcome reports.

Output locations depend on the analysis. ERPERF uses per-recording bundles and optional branches. Calculate distinguishes per-dataset outputs, aggregate datasets and value exports. Read the selected destination and overwrite/reuse policy before running. Keep raw recordings and a separate study backup; neither a log nor a recovery workspace is a data backup.