Applies the selected modality's feature function to each native-grid segment and preserves its observation keys. A row is one cycle, not an independent participant. This function does not fit statistical models or enforce grouped train/test splits. Retain participant identifiers when building such analyses.
Usage
summarizeCycleFeatures(x, FUN, ..., incomplete = c("error", "exclude"))Arguments
- x
Result of
segmentMultimodalCycles()orsegmentCohortCycles().- FUN
Function, or a uniquely named list of functions matching modalities. Each function accepts a time-by-channel matrix and returns a nonempty, uniquely named, finite real numeric vector. Feature names must match across cycles within each modality; their order is reconciled by name. The function must handle the native grid appropriately; sampling rates are not equalized and samples are not weighted by duration automatically. Bind rate-dependent settings in per-modality functions using the recording metadata; metadata are not passed as implicit arguments to
FUN.- ...
Additional arguments passed to
FUN.- incomplete
"error"(default) or explicit"exclude". Exclusion removes a cycle from all blocks if any modality was incomplete.
Value
List with numeric matrix blocks, matching per-modality key tables
keys, retained cycle observations, all original status rows,
cycle_rows and excluded_cycles (original cycle row indices), and
recordings metadata.
Blocks and keys can be passed to PhysioCrossModal's keyed block adapter.
No invalid cycle is silently imputed or treated as a complete observation.