https://github.com/coganlab/cross_patient_speech_decoding
Cross-patient speech decoding on neural data aligned to a shared latent space
Science Score: 67.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
○CITATION.cff file
-
✓codemeta.json file
Found codemeta.json file -
✓.zenodo.json file
Found .zenodo.json file -
✓DOI references
Found 1 DOI reference(s) in README -
✓Academic publication links
Links to: biorxiv.org -
✓Committers with academic emails
1 of 2 committers (50.0%) from academic institutions -
✓Institutional organization owner
Organization coganlab has institutional domain (coganlab.pages.oit.duke.edu) -
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (9.8%) to scientific vocabulary
Repository
Cross-patient speech decoding on neural data aligned to a shared latent space
Basic Info
Statistics
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 0
- Releases: 0
Metadata Files
README.md
Shared latent representations of speech production for cross-patient speech decoding
Overview
This repository contains code used in analyses and creation of figures for the paper "Shared latent representations of speech production for cross-patient speech decoding".
We use an approach based on canonical correlation analysis (CCA) to learn an alignment between latent neural representations of speech production recorded with micro-electrocorticography (μECoG) arrays from multiple patients. We show that patient-specific neural data can be aligned to a shared cross-patient latent space, enabling the training of cross-patient speech decoding models that outperform patient-specific models.
For more details, please check out our preprint! Spalding et al. 2025, bioRxiv
Requirements
All analyses were performed in Python $\geq$ 3.10. Packages used can be found in the environment.yml and requirements.txt files.
Usage
Analyses and code for all main figures in the paper (excluding figure 1, which is primarily illustrative) can be found in aligned_decoding/figure_analyses/fig_X.ipynb as notebooks stepping through anaylses performed in each figure. Analyses and code for relevant supplementary figures is also included in aligned_decoding/figure_analyses/supp/supp_fig_X.ipynb.
Aditional directories within aligned_decoding/ contain .py files with functionality relevant to various analyses:
- aligned_decoding/alignment/: Classes and utility funcitons for various alignment methods, including CCA, multiview CCA (MCCA), and joint PCA.
- aligned_decoding/decoders/: Wrapper classes to enable easy cross-patient decoding with scikit-learn-style decoders.
- aligned_decoding/decomposition/: Dimensionality reduction methods, including a wrapper to perform dimensionality reduction while properly reshaping data with more than two dimensions.
- aligned_decoding/nn_models/: Classes and utility functions defining PyTorch Lightning modules for training sequence-to-sequence recurrent neural networks with both patient-specific and cross-patient inputs.
- aligned_decoding/processing_utils/: Utility functions for processing neural data, including data saving and subsampling.
- aligned_decoding/scripts/: Scripts for running various decoding analyses (e.g. SVM-based, RNN-based, subsampled w.r.t. spatial characteristics, etc.) structured to be called by upstream compute-cluster job scripts.
Owner
- Name: coganlab
- Login: coganlab
- Kind: organization
- Email: coganlab_github@googlegroups.com
- Location: United States of America
- Website: https://coganlab.pages.oit.duke.edu/wiki/
- Twitter: CoganLab
- Repositories: 6
- Profile: https://github.com/coganlab
GitHub Events
Total
- Issues event: 6
- Watch event: 1
- Push event: 47
- Pull request event: 19
- Create event: 9
Last Year
- Issues event: 6
- Watch event: 1
- Push event: 47
- Pull request event: 19
- Create event: 9
Committers
Last synced: about 1 year ago
Top Committers
| Name | Commits | |
|---|---|---|
| Zac Spalding | s****c@g****m | 659 |
| Zachary Spalding | z****4@d****u | 1 |
Committer Domains (Top 20 + Academic)
Issues and Pull Requests
Last synced: 6 months ago
All Time
- Total issues: 49
- Total pull requests: 70
- Average time to close issues: 3 months
- Average time to close pull requests: less than a minute
- Total issue authors: 1
- Total pull request authors: 1
- Average comments per issue: 0.08
- Average comments per pull request: 0.03
- Merged pull requests: 69
- Bot issues: 0
- Bot pull requests: 0
Past Year
- Issues: 0
- Pull requests: 12
- Average time to close issues: N/A
- Average time to close pull requests: less than a minute
- Issue authors: 0
- Pull request authors: 1
- Average comments per issue: 0
- Average comments per pull request: 0.0
- Merged pull requests: 11
- Bot issues: 0
- Bot pull requests: 0
Top Authors
Issue Authors
- zspald (45)
Pull Request Authors
- zspald (73)