Skip to content

custom_text_normalization

Transformations for normalizing MEDS datasets, across both categorical and continuous dimensions.

main(cfg)

Calls the normalize function on a polars dataframe.

Source code in meds_torch/utils/custom_text_normalization.py 
221
222
223
224
225
226
@hydra.main(
    version_base=None, config_path=str(PREPROCESS_CONFIG_YAML.parent), config_name=PREPROCESS_CONFIG_YAML.stem
)
def main(cfg: DictConfig):
    """Calls the normalize function on a polars dataframe."""
    map_over(cfg, compute_fn=normalize)

normalize(df, code_metadata, code_modifiers=None)

Normalize a MEDS dataset across both categorical and continuous dimensions.

This function expects a MEDS dataset in flattened form, with columns for: - subject_id - time - code - numeric_value - text_value

In addition, the code_metadata dataset should contain information about the codes in the MEDS dataset, including the mandatory columns: - code (categorical) - code/vocab_index (int) - Any code_modifiers columns, if specified

Additionally, it must either have: - Pre-computed means and standard deviations for the numeric values of the codes in the MEDS dataset, via: - values/mean (float) - values/std (float) - Or the necessary statistics to compute the per-occurrence mean and standard deviation of the numeric values of the codes in the MEDS dataset, via: - values/n_occurrences (int) - values/sum (float) - values/sum_sqd (float)

The values/* functions will be used to normalize the code numeric values to have a mean of 0 and a standard deviation of 1. The output dataframe will further be filtered to only contain rows where the code in the MEDS dataset appears in the code_metadata dataset, and the output code column will be converted to the code/vocab_index integral ID from the code_metadata dataset.

This function can further be customized by specifying additional columns to join on, via the code_modifiers parameter, which must appear in both the MEDS dataset and the code metadata. These columns will be discarded from the output dataframe, which will only contain the four expected input columns, though normalized.

Parameters:

Name Type Description Default
df LazyFrame

The MEDS dataset to normalize. See above for the expected schema.

 required
code_metadata DataFrame

Metadata about the codes in the MEDS dataset. See above for the expected schema.

 required
code_modifiers list[str] | None

Additional columns to join on, which will be discarded from the output dataframe.

 None

Returns:

Type Description
LazyFrame

The normalized MEDS dataset, with the schema described above.

Examples:

>>> from datetime import datetime
>>> MEDS_df = pl.DataFrame(
...     {
...         "subject_id": [1, 1, 1, 2, 2, 2, 3],
...         "time": [
...             datetime(2021, 1, 1),
...             datetime(2021, 1, 1),
...             datetime(2021, 1, 2),
...             datetime(2022, 10, 2),
...             datetime(2022, 10, 2),
...             datetime(2022, 10, 2),
...             datetime(2022, 10, 2),
...         ],
...         "code": ["lab//A", "lab//A", "dx//B", "lab//A", "dx//D", "lab//C", "lab//F"],
...         "text_value": [None, "fever", None, None, "cough", None, None],
...         "numeric_value": [1, 3, None, 3, None, None, None],
...         "unit": ["mg/dL", "g/dL", None, "mg/dL", None, None, None],
...     },
...     schema = {
...         "subject_id": pl.UInt32,
...         "time": pl.Datetime,
...         "code": pl.Utf8,
...         "text_value": pl.String,
...         "numeric_value": pl.Float64,
...         "unit": pl.Utf8,
...    },
... )
>>> code_metadata = pl.DataFrame(
...     {
...         "code": ["lab//A", "lab//C", "dx//B", "dx//E", "lab//F"],
...         "code/vocab_index": [0, 2, 3, 4, 5],
...         "values/mean": [2.0, None, None, None, 3],
...         "values/std": [0.5, None, None, None, 0.2],
...     },
...     schema = {
...         "code": pl.Utf8,
...         "code/vocab_index": pl.UInt32,
...         "values/mean": pl.Float64,
...         "values/std": pl.Float64,
...     },
... )
>>> normalize(MEDS_df.lazy(), code_metadata).collect()
shape: (6, 5)
┌────────────┬─────────────────────┬──────┬───────────────┬────────────┐
│ subject_id ┆ time                ┆ code ┆ numeric_value ┆ text_value │
│ ---        ┆ ---                 ┆ ---  ┆ ---           ┆ ---        │
│ u32        ┆ datetime[μs]        ┆ u32  ┆ f64           ┆ str        │
╞════════════╪═════════════════════╪══════╪═══════════════╪════════════╡
│ 1          ┆ 2021-01-01 00:00:00 ┆ 0    ┆ -2.0          ┆ null       │
│ 1          ┆ 2021-01-01 00:00:00 ┆ 0    ┆ 2.0           ┆ fever      │
│ 1          ┆ 2021-01-02 00:00:00 ┆ 3    ┆ null          ┆ null       │
│ 2          ┆ 2022-10-02 00:00:00 ┆ 0    ┆ 2.0           ┆ null       │
│ 2          ┆ 2022-10-02 00:00:00 ┆ 2    ┆ null          ┆ null       │
│ 3          ┆ 2022-10-02 00:00:00 ┆ 5    ┆ null          ┆ null       │
└────────────┴─────────────────────┴──────┴───────────────┴────────────┘
>>> MEDS_df = pl.DataFrame(
...     {
...         "subject_id": [1, 1, 1, 2, 2, 2, 3],
...         "time": [
...             datetime(2021, 1, 1),
...             datetime(2021, 1, 1),
...             datetime(2021, 1, 2),
...             datetime(2022, 10, 2),
...             datetime(2022, 10, 2),
...             datetime(2022, 10, 2),
...             datetime(2022, 10, 2),
...         ],
...         "code": ["lab//A", "lab//A", "dx//B", "lab//A", "dx//D", "lab//C", "lab//F"],
...         "text_value": [None, "fever", None, None, "cough", None, None],
...         "numeric_value": [1, 3, None, 3, None, None, None],
...         "unit": ["mg/dL", "g/dL", None, "mg/dL", None, None, None],
...     },
...     schema = {
...         "subject_id": pl.UInt32,
...         "time": pl.Datetime,
...         "code": pl.Utf8,
...         "text_value": pl.String,
...         "numeric_value": pl.Float64,
...         "unit": pl.Utf8,
...    },
... )
>>> code_metadata = pl.DataFrame(
...     {
...         "code": ["lab//A", "lab//A", "lab//C", "dx//B", "dx//E", "lab//F"],
...         "unit": ["mg/dL", "g/dL", None, None, None, None],
...         "code/vocab_index": [0, 1, 2, 3, 4, 5],
...         "values/mean": [2.0, 3.0, None, None, None, 3],
...         "values/std": [0.5, 2.0, None, None, None, 0.2],
...     },
...     schema = {
...         "code": pl.Utf8,
...         "unit": pl.Utf8,
...         "code/vocab_index": pl.UInt32,
...         "values/mean": pl.Float64,
...         "values/std": pl.Float64,
...     },
... )
>>> normalize(MEDS_df.lazy(), code_metadata, ["unit"]).collect()
shape: (6, 5)
┌────────────┬─────────────────────┬──────┬───────────────┬────────────┐
│ subject_id ┆ time                ┆ code ┆ numeric_value ┆ text_value │
│ ---        ┆ ---                 ┆ ---  ┆ ---           ┆ ---        │
│ u32        ┆ datetime[μs]        ┆ u32  ┆ f64           ┆ str        │
╞════════════╪═════════════════════╪══════╪═══════════════╪════════════╡
│ 1          ┆ 2021-01-01 00:00:00 ┆ 0    ┆ -2.0          ┆ null       │
│ 1          ┆ 2021-01-01 00:00:00 ┆ 1    ┆ 0.0           ┆ fever      │
│ 1          ┆ 2021-01-02 00:00:00 ┆ 3    ┆ null          ┆ null       │
│ 2          ┆ 2022-10-02 00:00:00 ┆ 0    ┆ 2.0           ┆ null       │
│ 2          ┆ 2022-10-02 00:00:00 ┆ 2    ┆ null          ┆ null       │
│ 3          ┆ 2022-10-02 00:00:00 ┆ 5    ┆ null          ┆ null       │
└────────────┴─────────────────────┴──────┴───────────────┴────────────┘
Source code in meds_torch/utils/custom_text_normalization.py 
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
def normalize(
    df: pl.LazyFrame, code_metadata: pl.DataFrame, code_modifiers: list[str] | None = None
) -> pl.LazyFrame:
    """Normalize a MEDS dataset across both categorical and continuous dimensions.

    This function expects a MEDS dataset in flattened form, with columns for:
      - `subject_id`
      - `time`
      - `code`
      - `numeric_value`
      - `text_value`

    In addition, the `code_metadata` dataset should contain information about the codes in the MEDS dataset,
    including the mandatory columns:
      - `code` (`categorical`)
      - `code/vocab_index` (`int`)
      - Any `code_modifiers` columns, if specified

    Additionally, it must either have:
      - Pre-computed means and standard deviations for the numeric values of the codes in the MEDS dataset,
        via:
        - `values/mean` (`float`)
        - `values/std` (`float`)
      - Or the necessary statistics to compute the per-occurrence mean and standard deviation of the numeric
        values of the codes in the MEDS dataset, via:
        - `values/n_occurrences` (`int`)
        - `values/sum` (`float`)
        - `values/sum_sqd` (`float`)


    The `values/*` functions will be used to normalize the code numeric values to have a mean of 0 and a
    standard deviation of 1. The output dataframe will further be filtered to only contain rows where the
    `code` in the MEDS dataset appears in the `code_metadata` dataset, and the output `code` column will be
    converted to the `code/vocab_index` integral ID from the `code_metadata` dataset.

    This function can further be customized by specifying additional columns to join on, via the
    `code_modifiers` parameter, which must appear in both the MEDS dataset and the code metadata. These
    columns will be discarded from the output dataframe, which will only contain the four expected input
    columns, though normalized.

    Args:
        df: The MEDS dataset to normalize. See above for the expected schema.
        code_metadata: Metadata about the codes in the MEDS dataset. See above for the expected schema.
        code_modifiers: Additional columns to join on, which will be discarded from the output dataframe.

    Returns:
        The normalized MEDS dataset, with the schema described above.

    Examples:
        >>> from datetime import datetime
        >>> MEDS_df = pl.DataFrame(
        ...     {
        ...         "subject_id": [1, 1, 1, 2, 2, 2, 3],
        ...         "time": [
        ...             datetime(2021, 1, 1),
        ...             datetime(2021, 1, 1),
        ...             datetime(2021, 1, 2),
        ...             datetime(2022, 10, 2),
        ...             datetime(2022, 10, 2),
        ...             datetime(2022, 10, 2),
        ...             datetime(2022, 10, 2),
        ...         ],
        ...         "code": ["lab//A", "lab//A", "dx//B", "lab//A", "dx//D", "lab//C", "lab//F"],
        ...         "text_value": [None, "fever", None, None, "cough", None, None],
        ...         "numeric_value": [1, 3, None, 3, None, None, None],
        ...         "unit": ["mg/dL", "g/dL", None, "mg/dL", None, None, None],
        ...     },
        ...     schema = {
        ...         "subject_id": pl.UInt32,
        ...         "time": pl.Datetime,
        ...         "code": pl.Utf8,
        ...         "text_value": pl.String,
        ...         "numeric_value": pl.Float64,
        ...         "unit": pl.Utf8,
        ...    },
        ... )
        >>> code_metadata = pl.DataFrame(
        ...     {
        ...         "code": ["lab//A", "lab//C", "dx//B", "dx//E", "lab//F"],
        ...         "code/vocab_index": [0, 2, 3, 4, 5],
        ...         "values/mean": [2.0, None, None, None, 3],
        ...         "values/std": [0.5, None, None, None, 0.2],
        ...     },
        ...     schema = {
        ...         "code": pl.Utf8,
        ...         "code/vocab_index": pl.UInt32,
        ...         "values/mean": pl.Float64,
        ...         "values/std": pl.Float64,
        ...     },
        ... )
        >>> normalize(MEDS_df.lazy(), code_metadata).collect()
        shape: (6, 5)
        ┌────────────┬─────────────────────┬──────┬───────────────┬────────────┐
        │ subject_id ┆ time                ┆ code ┆ numeric_value ┆ text_value │
        │ ---        ┆ ---                 ┆ ---  ┆ ---           ┆ ---        │
        │ u32        ┆ datetime[μs]        ┆ u32  ┆ f64           ┆ str        │
        ╞════════════╪═════════════════════╪══════╪═══════════════╪════════════╡
        │ 1          ┆ 2021-01-01 00:00:00 ┆ 0    ┆ -2.0          ┆ null       │
        │ 1          ┆ 2021-01-01 00:00:00 ┆ 0    ┆ 2.0           ┆ fever      │
        │ 1          ┆ 2021-01-02 00:00:00 ┆ 3    ┆ null          ┆ null       │
        │ 2          ┆ 2022-10-02 00:00:00 ┆ 0    ┆ 2.0           ┆ null       │
        │ 2          ┆ 2022-10-02 00:00:00 ┆ 2    ┆ null          ┆ null       │
        │ 3          ┆ 2022-10-02 00:00:00 ┆ 5    ┆ null          ┆ null       │
        └────────────┴─────────────────────┴──────┴───────────────┴────────────┘
        >>> MEDS_df = pl.DataFrame(
        ...     {
        ...         "subject_id": [1, 1, 1, 2, 2, 2, 3],
        ...         "time": [
        ...             datetime(2021, 1, 1),
        ...             datetime(2021, 1, 1),
        ...             datetime(2021, 1, 2),
        ...             datetime(2022, 10, 2),
        ...             datetime(2022, 10, 2),
        ...             datetime(2022, 10, 2),
        ...             datetime(2022, 10, 2),
        ...         ],
        ...         "code": ["lab//A", "lab//A", "dx//B", "lab//A", "dx//D", "lab//C", "lab//F"],
        ...         "text_value": [None, "fever", None, None, "cough", None, None],
        ...         "numeric_value": [1, 3, None, 3, None, None, None],
        ...         "unit": ["mg/dL", "g/dL", None, "mg/dL", None, None, None],
        ...     },
        ...     schema = {
        ...         "subject_id": pl.UInt32,
        ...         "time": pl.Datetime,
        ...         "code": pl.Utf8,
        ...         "text_value": pl.String,
        ...         "numeric_value": pl.Float64,
        ...         "unit": pl.Utf8,
        ...    },
        ... )
        >>> code_metadata = pl.DataFrame(
        ...     {
        ...         "code": ["lab//A", "lab//A", "lab//C", "dx//B", "dx//E", "lab//F"],
        ...         "unit": ["mg/dL", "g/dL", None, None, None, None],
        ...         "code/vocab_index": [0, 1, 2, 3, 4, 5],
        ...         "values/mean": [2.0, 3.0, None, None, None, 3],
        ...         "values/std": [0.5, 2.0, None, None, None, 0.2],
        ...     },
        ...     schema = {
        ...         "code": pl.Utf8,
        ...         "unit": pl.Utf8,
        ...         "code/vocab_index": pl.UInt32,
        ...         "values/mean": pl.Float64,
        ...         "values/std": pl.Float64,
        ...     },
        ... )
        >>> normalize(MEDS_df.lazy(), code_metadata, ["unit"]).collect()
        shape: (6, 5)
        ┌────────────┬─────────────────────┬──────┬───────────────┬────────────┐
        │ subject_id ┆ time                ┆ code ┆ numeric_value ┆ text_value │
        │ ---        ┆ ---                 ┆ ---  ┆ ---           ┆ ---        │
        │ u32        ┆ datetime[μs]        ┆ u32  ┆ f64           ┆ str        │
        ╞════════════╪═════════════════════╪══════╪═══════════════╪════════════╡
        │ 1          ┆ 2021-01-01 00:00:00 ┆ 0    ┆ -2.0          ┆ null       │
        │ 1          ┆ 2021-01-01 00:00:00 ┆ 1    ┆ 0.0           ┆ fever      │
        │ 1          ┆ 2021-01-02 00:00:00 ┆ 3    ┆ null          ┆ null       │
        │ 2          ┆ 2022-10-02 00:00:00 ┆ 0    ┆ 2.0           ┆ null       │
        │ 2          ┆ 2022-10-02 00:00:00 ┆ 2    ┆ null          ┆ null       │
        │ 3          ┆ 2022-10-02 00:00:00 ┆ 5    ┆ null          ┆ null       │
        └────────────┴─────────────────────┴──────┴───────────────┴────────────┘
    """

    if code_modifiers is None:
        code_modifiers = []

    cols_to_select = ["code", "code/vocab_index"] + code_modifiers

    mean_col = pl.col("values/sum") / pl.col("values/n_occurrences")
    stddev_col = (pl.col("values/sum_sqd") / pl.col("values/n_occurrences") - mean_col**2) ** 0.5

    code_metadata_columns = set(code_metadata.columns)
    if "values/mean" in code_metadata_columns:
        cols_to_select.append("values/mean")
    else:
        cols_to_select.append(mean_col.alias("values/mean"))

    if "values/std" in code_metadata_columns:
        cols_to_select.append("values/std")
    else:
        cols_to_select.append(stddev_col.alias("values/std"))

    idx_col = "_row_idx"
    df_cols = df.collect_schema().names()
    while idx_col in df_cols:
        idx_col = f"_{idx_col}"

    return (
        df.with_row_index(idx_col)
        .join(
            code_metadata.lazy().select(cols_to_select),
            on=["code"] + code_modifiers,
            how="inner",
            join_nulls=True,
        )
        .select(
            idx_col,  # Keep row index to ensure proper sorting later
            "subject_id",  # Include subject identifier
            "time",  # Include the time column
            pl.col("code/vocab_index").alias("code"),  # Normalized code column
            ((pl.col("numeric_value") - pl.col("values/mean")) / pl.col("values/std")).alias(
                "numeric_value"
            ),  # Normalized numeric_value
            "text_value",  # Keep the text_value column intact (this was previously dropped)
        )
        .sort(idx_col)  # Sort by the index column
        .drop(idx_col)  # Drop the temporary index column
    )