cehrgpt 0.1.4__tar.gz → 0.1.6__tar.gz

{cehrgpt-0.1.4 → cehrgpt-0.1.6}/.github/workflows/tests.yaml

@@ -27,6 +27,7 @@ jobs:
       run: |
         python -m pip install --upgrade pip
         pip install flake8 pytest
+        pip install tensorflow==2.15.0 tensorflow-datasets==4.5.2
         pip install -e .
     - name: Lint with flake8
       run: |

{cehrgpt-0.1.4 → cehrgpt-0.1.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cehrgpt
-Version: 0.1.4
+Version: 0.1.6
 Summary: CEHR-GPT: Generating Electronic Health Records with Chronological Patient Timelines
 Author-email: Chao Pang <chaopang229@gmail.com>, Xinzhuo Jiang <xj2193@cumc.columbia.edu>, Krishna Kalluri <kk3326@cumc.columbia.edu>, Elise Minto <em3697@cumc.columbia.edu>, Jason Patterson <jp3477@cumc.columbia.edu>, Nishanth Parameshwar Pavinkurve <np2689@cumc.columbia.edu>, Karthik Natarajan <kn2174@cumc.columbia.edu>
 License: MIT License
@@ -12,8 +12,8 @@ Classifier: Programming Language :: Python :: 3
 Requires-Python: >=3.10.0
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: cehrbert>=1.4.8
-Requires-Dist: cehrbert_data>=0.1.1
+Requires-Dist: cehrbert==1.4.8
+Requires-Dist: cehrbert_data==0.1.1
 Requires-Dist: openai==1.54.3
 Requires-Dist: optuna==4.0.0
 Requires-Dist: transformers==4.44.1
@@ -104,7 +104,9 @@ python -u -m cehrgpt.runners.hf_cehrgpt_pretrain_runner \
   --sample_packing --max_tokens_per_batch 16384 \
   --warmup_ratio 0.01 --weight_decay 0.01 \
   --num_train_epochs 50 --learning_rate 0.0002 \
-  --use_early_stopping --early_stopping_threshold 0.001
+  --use_early_stopping \
+  --load_best_model_at_end true \
+  --early_stopping_threshold 0.001
 ```
 
 > **Tip**: Increase `max_position_embeddings` for longer context windows based on your use case.

{cehrgpt-0.1.4 → cehrgpt-0.1.6}/README.md

@@ -69,7 +69,9 @@ python -u -m cehrgpt.runners.hf_cehrgpt_pretrain_runner \
   --sample_packing --max_tokens_per_batch 16384 \
   --warmup_ratio 0.01 --weight_decay 0.01 \
   --num_train_epochs 50 --learning_rate 0.0002 \
-  --use_early_stopping --early_stopping_threshold 0.001
+  --use_early_stopping \
+  --load_best_model_at_end true \
+  --early_stopping_threshold 0.001
 ```
 
 > **Tip**: Increase `max_position_embeddings` for longer context windows based on your use case.

{cehrgpt-0.1.4 → cehrgpt-0.1.6}/pyproject.toml

@@ -28,8 +28,8 @@ classifiers = [
 ]
 
 dependencies = [
-    "cehrbert>=1.4.8",
-    "cehrbert_data>=0.1.1",
+    "cehrbert==1.4.8",
+    "cehrbert_data==0.1.1",
     "openai==1.54.3",
     "optuna==4.0.0",
     "transformers==4.44.1",

{cehrgpt-0.1.4 → cehrgpt-0.1.6}/src/cehrgpt/data/cehrgpt_data_processor.py

@@ -275,11 +275,7 @@ class CehrGptDataProcessor(DatasetMapping):
                         if demographic_tokens is not None
                         else self.empty_array
                     ),
-                    np.asarray(
-                        self._convert_time_to_event(
-                            record["concept_ids"][start_index:end_index]
-                        )
-                    ),
+                    np.asarray(record["time_to_visits"][start_index:end_index]),
                     np.asarray([-100.0]) if add_last_token else self.empty_array,
                 ]
             ).astype(np.float32)
@@ -303,6 +299,11 @@ class CehrGptDataProcessor(DatasetMapping):
             record["concept_ids"], record["epoch_times"]
         )
 
+        if self.include_ttv_prediction:
+            record["time_to_visits"] = np.asarray(
+                self._convert_time_to_event(record["concept_ids"])
+            )
+
         # Return the record directly if the actual sequence length is less than the max sequence
         if seq_length <= new_max_length:
             # We only add [END] to the end of the sequence in pre-training

{cehrgpt-0.1.4 → cehrgpt-0.1.6}/src/cehrgpt/data/hf_cehrgpt_dataset_collator.py

@@ -528,6 +528,7 @@ class SamplePackingCehrGptDataCollator(CehrGptDataCollator):
         current_epoch_times = []
         current_value_indicators = []
         current_values = []
+        current_time_to_visits = []
 
         # MOTOR inputs
         current_motor_censor_times = []
@@ -567,6 +568,16 @@ class SamplePackingCehrGptDataCollator(CehrGptDataCollator):
             )
             current_epoch_times.extend(epoch_times + [max(epoch_times)])
 
+            if self.include_ttv_prediction:
+                current_time_to_visits.extend(
+                    (
+                        example["time_to_visits"].tolist()
+                        if isinstance(example["time_to_visits"], torch.Tensor)
+                        else list(example["time_to_visits"])
+                    )
+                    + [-100]
+                )
+
             if self.include_values:
                 current_value_indicators.extend(
                     (
@@ -649,6 +660,9 @@ class SamplePackingCehrGptDataCollator(CehrGptDataCollator):
             "epoch_times": current_epoch_times,
         }
 
+        if self.include_ttv_prediction:
+            packed_example.update({"time_to_visits": current_time_to_visits})
+
         if self.include_values:
             packed_example.update(
                 {"value_indicators": current_value_indicators, "values": current_values}

{cehrgpt-0.1.4 → cehrgpt-0.1.6}/src/cehrgpt.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cehrgpt
-Version: 0.1.4
+Version: 0.1.6
 Summary: CEHR-GPT: Generating Electronic Health Records with Chronological Patient Timelines
 Author-email: Chao Pang <chaopang229@gmail.com>, Xinzhuo Jiang <xj2193@cumc.columbia.edu>, Krishna Kalluri <kk3326@cumc.columbia.edu>, Elise Minto <em3697@cumc.columbia.edu>, Jason Patterson <jp3477@cumc.columbia.edu>, Nishanth Parameshwar Pavinkurve <np2689@cumc.columbia.edu>, Karthik Natarajan <kn2174@cumc.columbia.edu>
 License: MIT License
@@ -12,8 +12,8 @@ Classifier: Programming Language :: Python :: 3
 Requires-Python: >=3.10.0
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: cehrbert>=1.4.8
-Requires-Dist: cehrbert_data>=0.1.1
+Requires-Dist: cehrbert==1.4.8
+Requires-Dist: cehrbert_data==0.1.1
 Requires-Dist: openai==1.54.3
 Requires-Dist: optuna==4.0.0
 Requires-Dist: transformers==4.44.1
@@ -104,7 +104,9 @@ python -u -m cehrgpt.runners.hf_cehrgpt_pretrain_runner \
   --sample_packing --max_tokens_per_batch 16384 \
   --warmup_ratio 0.01 --weight_decay 0.01 \
   --num_train_epochs 50 --learning_rate 0.0002 \
-  --use_early_stopping --early_stopping_threshold 0.001
+  --use_early_stopping \
+  --load_best_model_at_end true \
+  --early_stopping_threshold 0.001
 ```
 
 > **Tip**: Increase `max_position_embeddings` for longer context windows based on your use case.

{cehrgpt-0.1.4 → cehrgpt-0.1.6}/src/cehrgpt.egg-info/requires.txt

@@ -1,5 +1,5 @@
-cehrbert>=1.4.8
-cehrbert_data>=0.1.1
+cehrbert==1.4.8
+cehrbert_data==0.1.1
 openai==1.54.3
 optuna==4.0.0
 transformers==4.44.1