Update model card: 5 models + 5 embedding repos

README.md

@@ -17,58 +17,35 @@ All models follow the AASM 5-class standard: **W, N1, N2, N3, REM**.
 
 ## Available Models
 
-| Model | Architecture | Training Dataset | Channels | Pipeline | Params | Reference |
-|---|---|---|---|---|---|---|
-| `seqsleepnet-phan` | BiLSTM + Attention | Sleep-EDF | EEG | seqsleepnet (STFT) | 659K | Phan et al. 2019 |
-| `tinysleepnet-supratak` | CNN + LSTM | Sleep-EDF | EEG | raw | 83K | Supratak & Guo 2020 |
-| `sleeptransformer-phan` | Transformer | SHHS | EEG | seqsleepnet (STFT) | 2.1M | Phan et al. 2022 |
-| `lseqsleepnet-phan` | Long-sequence BiLSTM | SHHS | EEG | seqsleepnet (STFT) | 1.8M | Phan et al. 2023 |
-| `chambon2018` | Braindecode CNN | MASS SS3 | EEG | raw (128 Hz) | 220K | Chambon et al. 2018 |
-| `tsinalis-2016` | 2-layer CNN | Sleep-EDF | EEG | identity | 145K | Tsinalis et al. 2016 |
-
-*Models are added as training completes. Check back for updates.*
+| Model | Architecture | Training Dataset | ACC | F1 | κ | Params | Reference |
+|---|---|---|---|---|---|---|---|
+| `sleeptransformer-phan` | Transformer | SHHS | **0.866** | **0.796** | **0.810** | 2.1M | Phan et al. 2022 |
+| `tinysleepnet-supratak` | CNN + LSTM | Sleep-EDF | 0.842 | 0.790 | 0.784 | 83K | Supratak & Guo 2020 |
+| `seqsleepnet-phan` | BiLSTM + Attention | Sleep-EDF | 0.810 | 0.758 | 0.742 | 659K | Phan et al. 2019 |
+| `chambon2018` | Braindecode CNN | MASS SS3 | 0.826 | 0.731 | 0.736 | 29K | Chambon et al. 2018 |
+| `tsinalis-2016` | 2-layer CNN | Sleep-EDF | 0.773 | 0.661 | 0.683 | 145K | Tsinalis et al. 2016 |
 
 ## Usage
 
 ```python
 from physioex.models import load_from_pretrained
 
-# Load a pretrained model
-model = load_from_pretrained("seqsleepnet-phan", verbose=True)
-
-# Use for inference or embedding extraction
-from physioex.models import extract_embeddings, load_embeddings
-
-# Extract embeddings on a new dataset
-path = extract_embeddings(model, dataset, model_name="seqsleepnet-phan", ...)
+model = load_from_pretrained("sleeptransformer-phan", verbose=True)
 
-# Or download pre-extracted embeddings from HuggingFace
-path = load_embeddings("seqsleepnet-phan", "hmc", verbose=True)
+# Extract or download embeddings
+from physioex.models import load_embeddings
+path = load_embeddings("sleeptransformer-phan", "hmc", verbose=True)
 ```
 
-## Cross-Dataset Evaluation (SeqSleepNet-Phan)
-
-Zero-shot transfer from Sleep-EDF to other datasets (voting evaluation):
-
-| Dataset | ACC | Macro F1 | κ |
-|---|---|---|---|
-| **Sleep-EDF** (train) | 0.8101 | 0.7575 | 0.7421 |
-| DCSM | 0.6649 | 0.5839 | 0.5370 |
-| MESA | 0.6330 | 0.5546 | 0.4921 |
-| MASS | 0.6033 | 0.5422 | 0.4655 |
-| HMC | 0.5851 | 0.5574 | 0.4464 |
-| WSC | 0.4928 | 0.4577 | 0.3255 |
-| Parkinsons | 0.4428 | 0.3902 | 0.2737 |
-| SHHS | 0.3989 | 0.3597 | 0.2434 |
-| Alzheimers | 0.2582 | 0.1449 | 0.0239 |
-
 ## Pre-extracted Embeddings
 
-Contextualized per-epoch embeddings are available as separate dataset repos:
-
 | Model | HuggingFace Repo | Datasets |
 |---|---|---|
-| SeqSleepNet-Phan | [`4rooms/seqsleepnet-phan-embeddings`](https://huggingface.co/datasets/4rooms/seqsleepnet-phan-embeddings) | 20 datasets, 12.5K subjects |
+| SeqSleepNet-Phan | [`4rooms/seqsleepnet-phan-embeddings`](https://huggingface.co/datasets/4rooms/seqsleepnet-phan-embeddings) | 25 datasets |
+| TinySleepNet-Supratak | [`4rooms/tinysleepnet-supratak-embeddings`](https://huggingface.co/datasets/4rooms/tinysleepnet-supratak-embeddings) | 25 datasets |
+| SleepTransformer-Phan | [`4rooms/sleeptransformer-phan-embeddings`](https://huggingface.co/datasets/4rooms/sleeptransformer-phan-embeddings) | 8 datasets |
+| Chambon2018 | [`4rooms/chambon2018-embeddings`](https://huggingface.co/datasets/4rooms/chambon2018-embeddings) | 8 datasets |
+| Tsinalis-2016 | [`4rooms/tsinalis-2016-embeddings`](https://huggingface.co/datasets/4rooms/tsinalis-2016-embeddings) | 8 datasets |
 
 ## Datasets
 
@@ -95,46 +72,4 @@ Contextualized per-epoch embeddings are available as separate dataset repos:
     url={http://iopscience.iop.org/article/10.1088/1361-6579/adaf73},
     year={2025},
 }
-
-@article{phan2019seqsleepnet,
-    title={SeqSleepNet: End-to-End Hierarchical Recurrent Neural Network for Sequence-to-Sequence Automatic Sleep Staging},
-    author={Phan, Huy and Andreotti, Fernando and Cooray, Navin and Chen, Oliver Y and De Vos, Maarten},
-    journal={IEEE Transactions on Neural Systems and Rehabilitation Engineering},
-    year={2019},
-}
-
-@inproceedings{supratak2020tinysleepnet,
-    title={TinySleepNet: An Efficient Deep Learning Model for Sleep Stage Scoring based on Raw Single-Channel EEG},
-    author={Supratak, Akara and Guo, Yike},
-    booktitle={IEEE EMBC},
-    year={2020},
-}
-
-@article{phan2022sleeptransformer,
-    title={SleepTransformer: Automatic Sleep Staging with Interpretability and Uncertainty Quantification},
-    author={Phan, Huy and Mikkelsen, Kaare and Ch\'en, Oliver Y and Koch, Philipp and Mertins, Alfred and De Vos, Maarten},
-    journal={IEEE Transactions on Biomedical Engineering},
-    year={2022},
-}
-
-@article{phan2023lseqsleepnet,
-    title={L-SeqSleepNet: Whole-cycle Long Sequence Modelling for Automatic Sleep Staging},
-    author={Phan, Huy and Ch\'en, Oliver Y and Koch, Philipp and Lu, Zhongxiang and McLoughlin, Ian and Mertins, Alfred and De Vos, Maarten},
-    journal={IEEE Transactions on Neural Systems and Rehabilitation Engineering},
-    year={2023},
-}
-
-@article{chambon2018deep,
-    title={A deep learning architecture for temporal sleep stage classification using multivariate and multimodal time series},
-    author={Chambon, Stanislas and Galtier, Mathieu and Arnal, Pierrick and Wainrib, Gilles and Gramfort, Alexandre},
-    journal={IEEE Transactions on Neural Systems and Rehabilitation Engineering},
-    year={2018},
-}
-
-@article{tsinalis2016automatic,
-    title={Automatic Sleep Stage Scoring with Single-Channel EEG Using Convolutional Neural Networks},
-    author={Tsinalis, Orestis and Matthews, Paul M and Guo, Yike and Zafeiriou, Stefanos},
-    journal={arXiv:1610.01683},
-    year={2016},
-}
 ```