Slide 09 · Benchmark

Revised Software Implementation

Midpoint work is building the common pipeline before the first comparison runs.
Current plan

Initial Run Set

These are the implementations we are preparing to run once the shared preprocessing and scheduling pipeline is ready.

metric: gross AUPRC
Common dataset
Common label mapping
Common record splits
Shared job pipeline
Challenge DRCNN pipeline
setup in progress

Howe-Patterson

Published multichannel DRCNN codebase being adapted to the common apnea labels and split protocol.
Published feature pipeline
setup in progress

Bragi-Marinosson

Published engineered-feature plus BiLSTM ensemble being adapted to the same evaluation pipeline.
Classical ML baselines
planned next

Logistic regression + random forest

Planned once the shared preprocessing and scheduling pipeline is in place.
Other paper candidates are still in literature screening and are not part of the first scheduled runs.