231023 sleep staging jathurshan

.gitignore

@@ -127,4 +127,7 @@ leaderboard/credentials.json
 leaderboard/rtd_token.txt
 
 # locally pre-trained models
-pyhealth/medcode/pretrained_embeddings/kg_emb/examples/pretrained_model
+pyhealth/medcode/pretrained_embeddings/kg_emb/examples/pretrained_model
+
+# local cache path
+pyhealth/__init__.py

examples/sleep_staging_manytomany_sleepedf_cross_modal_transformer.py

@@ -0,0 +1,75 @@
+import os
+os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
+os.environ["CUDA_VISIBLE_DEVICES"] = "1"
+import torch
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+from pyhealth.datasets import SleepEDFDataset
+from pyhealth.tasks.sleep_staging import multi_epoch_multi_modal_sleep_staging_sleepedf_fn
+from pyhealth.datasets.splitter import split_by_patient
+from pyhealth.datasets import split_by_patient, get_dataloader
+from pyhealth.models import Seq_Cross_Modal_Transformer_PyHealth
+from pyhealth.trainer import Trainer
+from pyhealth.metrics.multiclass import multiclass_metrics_fn
+from pyhealth.models.cross_modal_transformer import interpret_cmt, interactive_plot_cmt
+
+'''
+Example Colab tutorial can be found at:
+https://colab.research.google.com/drive/1jwD5NX8cR47MRtnOW_drKIS6lP1tbWFc?usp=sharing
+'''
+
+
+# step 1: load signal data
+sleepedf_ds = SleepEDFDataset(
+    root="/srv/local/data/SLEEPEDF/sleep-edf-database-expanded-1.0.0/sleep-cassette",#'/home/jp65/pyhealth_hackathon/storage.googleapis.com/pyhealth/sleepedf-sample',#"/srv/local/data/SLEEPEDF/sleep-edf-database-expanded-1.0.0/sleep-cassette",
+    dev=True,
+    refresh_cache=False,
+)
+
+
+# step 2: set task
+modality = ['EEG Fpz-Cz','EOG horizontal']
+num_epoch_seq = 15 # Defalut value is 15
+sleepedf_task_ds = sleepedf_ds.set_task(lambda x: multi_epoch_multi_modal_sleep_staging_sleepedf_fn(x,modality = modality,num_epoch_seq = num_epoch_seq))
+
+
+# split dataset
+# data split
+train_dataset, val_dataset, test_dataset = split_by_patient(sleepedf_task_ds, [0.8, 0.1, 0.1])
+
+# create dataloaders (they are <torch.data.DataLoader> object)
+batch_size = 64
+train_loader = get_dataloader(train_dataset, batch_size=batch_size, shuffle=True)
+val_loader = get_dataloader(val_dataset, batch_size=batch_size, shuffle=False)
+test_loader = get_dataloader(test_dataset, batch_size=batch_size, shuffle=False)
+
+# STEP 3: define model
+model = Seq_Cross_Modal_Transformer_PyHealth( dataset= sleepedf_task_ds, 
+                                            feature_keys= ['signal'], 
+                                            label_key= ['label'], 
+                                            mode= 'multiclass', 
+                                            d_model = 128,
+                                            num_epoch_seq = num_epoch_seq,
+                                            dim_feedforward=512,
+                                            window_size = 50,
+                                            num_classes = 6,).to(device)#("cuda:1")
+
+# STEP 4: define trainer
+trainer = Trainer(model=model,metrics=["cohen_kappa", "accuracy"])
+trainer.train(
+    train_dataloader=train_loader,
+    val_dataloader=val_loader,
+    epochs=100,
+    monitor="cohen_kappa",
+    monitor_criterion="max"
+)
+
+
+# STEP 5: evaluate
+score = trainer.evaluate(test_loader)
+print (score)
+
+
+# STEP 6: interpret
+test_data = interpret_cmt(trainer,test_loader,num_epoch_seq,num_classes=6)
+interactive_plot_cmt(test_data)

pyhealth/models/__init__.py

@@ -23,4 +23,5 @@
 from .stagenet import StageNet, StageNetLayer
 from .tcn import TCN, TCNLayer
 from .molerec import MoleRec, MoleRecLayer
-from .medlink import MedLink
+from .medlink import MedLink
+from .cross_modal_transformer import Seq_Cross_Modal_Transformer_PyHealth