User Guide

Task-oriented guides to using PhysioEx: loading physiological datasets, preprocessing signals, training and evaluating models, driving the command-line interface, and explaining trained models.

New to PhysioEx? Start with Data to see how recordings are loaded, then Preprocessing to shape the signals, then Training to fit a model.

For the complete symbol-level reference see the API Reference; for the internal design (class diagrams) see the architecture links in Contributing.

• Data
  • Resolving a dataset by name
  • Instantiating a dataset
  • Splitting
  • Merging datasets
  • Building a DataLoader
  • API
• Preprocessing
  • Presets
  • Composing a pipeline manually
  • API
• Training (programmatic API)
  • Trainer at a glance
  • Example
  • API
• Command-line interface of the train module
  • Unified dataset flags
  • train
  • finetune
  • test_model
• Model architectures
  • Classic architectures
  • Foundation encoders
  • Loading a model
  • Adding your own model
• Explain Module
  • Post-hoc attribution — physioex.explain.posthoc
  • Foundational explainability — physioex.explain.foundational
  • Prototypes — physioex.explain.prototypes