Custom index

score="custom" scores your own category definitions. Pass them as a polars (or pandas) DataFrame in definition_data, or as a CSV in definition_file_path.

column	content
`category`	category name
`icd9_codes`, `icd10_codes`	code prefixes separated by `\\|`; only the column of the chosen `icd_version` is needed
weight column	name it with `weight_col_name` (default `"weights"`)
`and_group` (optional)	rows of one `category` with the same non-empty `and_group` count only if all are present
`year` (optional)	a row applies from this year on. Needs a year column in the data: `year_col` (default `"year"`) is used automatically if that column exists

A category is present if any of its rows matches (OR). Codes are matched by prefix.

import polars as pl
from pycomorb import comorbidity

definitions = pl.DataFrame({
    "category": ["Heart", "Cancer", "Advanced cancer"],
    "icd10_codes": ["I21|I22", "C", "C77|C78"],
    "weights": [2, 1, 3],
})

comorbidity(
    "custom", df, id_col="id", code_col="code",
    definition_data=definitions, weight_col_name="weights",
    mutual_exclusion_rules=[("Advanced cancer", "Cancer")],
    score_col_name="My score", return_categories=True,
)

mutual_exclusion_rules is a list of (keep, suppress) pairs: if keep is present, suppress is set to 0 and is not counted. Both categories must exist in the definitions (ValueError otherwise).

To publish or reuse a definition, see Reporting.