thbndi commited on
Commit
898a7ce
1 Parent(s): a72fda4

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +65 -0
README.md ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Dataset Usage
2
+
3
+ ## Description
4
+
5
+ The `load_dataset` function is a powerful tool to efficiently load and prepare the Mimic-IV dataset for various healthcare analysis tasks. It offers a wide range of options for encoding data and generating cohorts, allowing for seamless integration into your research or application.
6
+
7
+ ## Function Signature
8
+
9
+ ```python
10
+ load_dataset('thbndi/Mimic4Dataset', task, mimic_path=mimic_data, config_path=config_file, encoding=encod, generate_cohort=gen_cohort, val_size=size, cache_dir=cache)
11
+ ```
12
+
13
+ ## Arguments
14
+
15
+ 1. `task` (string) :
16
+ - Description: Specifies the task you want to perform with the dataset.
17
+ - Default: "Mortality"
18
+ - Note: Possible Values : 'Phenotype', 'Length of Stay', 'Readmission', 'Mortality'
19
+
20
+ 2. `mimic_path` (string) :
21
+ - Description: Complete path to the Mimic-IV raw data on user's machine.
22
+ - Note: You need to provide the appropriate path where the Mimic-IV data is stored.
23
+
24
+ 3. `config_path` (string) optionnal :
25
+ - Description: Path to the configuration file for the cohort generation choices (more infos in '/config/readme.md').
26
+ - Default: Configuration file provided in the 'config' folder.
27
+
28
+ 4. `encoding` (string) optionnal :
29
+ - Description: Data encoding option for the features.
30
+ - Options: "concat", "aggreg", "tensor", "raw", "text"
31
+ - Default: "concat"
32
+ - Note: Choose one of the following options for data encoding:
33
+ - "concat": Concatenates the one-hot encoded diagnoses, demographic data vector, and dynamic features at each measured time instant, resulting in a high-dimensional feature vector.
34
+ - "aggreg": Concatenates the one-hot encoded diagnoses, demographic data vector, and dynamic features, where each item_id is replaced by the average of the measured time instants, resulting in a reduced-dimensional feature vector.
35
+ - "tensor": Represents each feature as an 2D array. There are separate arrays for labels, demographic data ('DEMO'), diagnosis ('COND'), medications ('MEDS'), procedures ('PROC'), chart/lab events ('CHART/LAB'), and output events data ('OUT'). Dynamic features are represented as 2D arrays where each row contains values at a specific time instant.
36
+ - "raw": Provide cohort from the pipeline without any encoding for custom data processing.
37
+ - "text": Represents diagnoses as text suitable for BERT or other similar text-based models.
38
+
39
+ 5. `generate_cohort` (bool) optionnal :
40
+ - Description: Determines whether to generate a new cohort from Mimic-IV data.
41
+ - Default: True
42
+ - Note: Set it to True to generate a cohort, or False to skip cohort generation.
43
+
44
+ 6. `val_size`, 'test_size' (float) optionnal :
45
+ - Description: Proportion of the dataset used for validation during training.
46
+ - Default: 0.1 for validation size and 0.2 for testing size.
47
+ - Note: Can be set to 0.
48
+
49
+ 7. `cache_dir` (string) optionnal :
50
+ - Description: Directory where the processed dataset will be cached.
51
+ - Note: Providing a cache directory for each encoding type can avoid errors when changing the encoding type.
52
+
53
+ ## Example Usage
54
+
55
+ ```python
56
+ from your_module import load_dataset
57
+
58
+ # Example 1: Load dataset with default settings
59
+ dataset = load_dataset('thbndi/Mimic4Dataset', task="Mortality")
60
+
61
+ # Example 2: Load dataset with custom settings
62
+ dataset = load_dataset('thbndi/Mimic4Dataset', task="Phenotype", mimic_path="/path/to/mimic_data", config_path="/path/to/config_file", encoding="aggreg", generate_cohort=False, val_size=0.2, cache_dir="/path/to/cache_dir")
63
+ ```
64
+
65
+ Please note that the provided examples are for illustrative purposes only, and you should adjust the paths and settings based on your actual dataset and specific use case.