Documentation

Welcome to the osl-dynamics documentation!

New users can start with the Getting Started guide for a quick introduction and then work through the tutorials below. The FAQ covers common questions about data preparation, model training, and post-hoc analysis.

API Reference

The API reference provides documentation for all classes, methods, and functions in osl-dynamics.

Models

Model	State type	Temporal model	Best for
HMM	Discrete (mutually exclusive)	Markovian (transition probability matrix)	Resting-state analysis; interpretable summary stats
DyNeMo	Continuous (linear mixture of modes)	Non-Markovian (RNN)	Task data; overlapping network activity
DyNeSte	Discrete (mutually exclusive)	Non-Markovian (RNN)	Discrete states with long-range dynamics
M-DyNeMo	Continuous (linear mixture; separate for power and FC)	Non-Markovian (RNN)	Separate power and connectivity dynamics
HIVE	Discrete (mutually exclusive)	Markovian (transition probability matrix)	Modelling inter-session variability (e.g. subjects, scanners, sites)

Also see the FAQ for guidance on choosing a model and hyperparameters.

Parcellations

For information regarding the parcellations available in osl-dynamics, see here.

Tutorials

The following tutorials illustrate basic usage and analysis that can be done with osl-dynamics.

M/EEG processing tutorials:

• MEG: Preprocessing, Source Reconstruction, Parcellation.

• Cam-CAN MEG Batch Processing.

• LEMON EEG Batch Processing.

Also see Canonical-HMM-Networks for start-to-end tutorials with Elekta, CTF, OPM and EEG data.

Data tutorials:

• Loading Data.

• Preparing M/EEG Data.

• Preparing fMRI Data.

• Time-Delay Embedding.

Static (time-averaged) modelling tutorials for MEG:

• Static: Spectral Analysis.

• Static: Power Analysis.

• Static: Amplitude Envelope Correlation (AEC) Analysis.

Dynamic modelling tutorials:

• Sliding Window Analysis.

• HMM: Training.

• DyNeMo: Training.

• HMM/DyNeMo: Get Inferred Parameters.

HMM post-hoc analysis tutorials:

• HMM: Multitaper Spectra.

• HMM: Plotting MEG Networks.

• HMM: Summary Statistics.

• HMM: fMRI Dual Estimation.

• HMM: Plotting fMRI Networks.

DyNeMo post-hoc analysis tutorials:

• DyNeMo: Regression Spectra.

• DyNeMo: Plotting Networks.

• DyNeMo: Mixing Coefficients.

Task analysis tutorials:

• Epoching: Network Response.

• Dynamic Network Analysis of Electrophysiological Task Data.

Group-level analysis tutorials:

• Group Contrast.

• Network Response.

Examples Directory

More examples scripts can be found in the examples directory of the repository.

Workshops

The OHBA Methods Group organises teaching workshops for analysing M/EEG data using osl-ephys and osl-dynamics.

Links to past workshops:

• 2023 OSL workshop.

• 2025 OSL workshop.