TimeRangePhenotype

Bases: Phenotype

As the name implies, TimeRangePhenotype is designed for working with time ranges. If the input data has a start and an end date, use TimeRangePhenotype to identify other events (or patients) that occur within this time range. The most common use case of this is working with 'health insurance coverage' data i.e. on 'OBSERVATION_PERIOD' table. These tables have one or many rows per patient with the start of coverage and end of coverage i.e. domains compatible with TimeRangePhenotype require a START_DATE and an END_DATE column. At it's simplest, TimeRangePhenotype identifies patients who have their INDEX_DATE (or other anchor date of interest) within this time range. Additionally, a minimum or maximum number of days from the anchor date to the beginning/end of the time range can be defined. The returned Phenotype has the following interpretation:

DATE: If relative_time_range.when='before', then DATE is the beginning of the coverage period containing the anchor phenotype. If relative_time_range.when='after', then DATE is the end of the coverage period containing the anchor date. VALUE: Coverage (in days) relative to the anchor date. By convention, always non-negative.

There are two primary use cases for TimeRangePhenotype

1. Identify patients with some minimum duration of coverage prior to anchor_phenotype date e.g. "identify patients with 1 year of continuous coverage prior to index date"
2. Determine the date of loss to followup (right censoring) i.e. the duration of coverage after the anchor_phenotype event

Data for TimeRangePhenotype

This phenotype requires a table with PersonID and a coverage start date and end date. Depending on the datasource used, this information is a separate ObservationPeriod table or found in the PersonTable. Use an PhenexObservationPeriodTable to map required coverage start and end date columns. For tables with overlapping time ranges, use the CombineOverlappingPeriods derived table to combine time ranges into a single time range.

PersonID	startDate	endDate
1	2009-01-01	2010-01-01
2	2008-01-01	2010-01-02

One assumption that is made by TimeRangePhenotype is that there are NO overlapping coverage periods.

Parameters:

Name	Type	Description	Default
name	Optional[str]	The name of the phenotype.	'TIME_RANGE'
domain	Optional[str]	The domain of the phenotype. Default is 'observation_period'.	'OBSERVATION_PERIOD'
relative_time_range	Optional[RelativeTimeRangeFilter]	Filter returned persons based on the duration of coverage in days. The relative_time_range.anchor_phenotype defines the reference date with respect to calculate coverage. In typical applications, the anchor phenotype will be the entry criterion. The relative_time_range.when 'before', 'after'. If before, the return date is the start of the coverage period containing the anchor_phenotype. If after, the return date is the end of the coverage period containing the anchor_phenotype.	None
allow_null_end_date	bool	TimeRangePhenotype checks that anchor date is within the time range of interest. This requires that the start date is not null, and the end date is either null or after the anchor date. If you want to require that the end date is not null, set allow_null_end_date to False.	True

Example:

# make sure to create an entry phenotype, for example 'atrial fibrillation diagnosis'
entry_phenotype = CodelistPhenotype(...)
# one year continuous coverage prior to index
one_year_coverage = TimeRangePhenotype(
    relative_time_range = RelativeTimeRangeFilter(
        min_days=GreaterThanOrEqualTo(365),
        anchor_phenotype = entry_phenotype,
        when = 'before',
    ),
)
# determine the date of loss to followup
loss_to_followup = TimeRangePhenotype(
    relative_time_range = RelativeTimeRangeFilter(
        anchor_phenotype = entry_phenotype
        when = 'after',
    )
)

# determine the date when a drug was discontinued
drug_discontinuation = TimeRangePhenotype(
    relative_time_range = RelativeTimeRangeFilter(
        anchor_phenotype = entry_phenotype
        when = 'after',
    )
)

Source code in phenex/phenotypes/time_range_phenotype.py

class TimeRangePhenotype(Phenotype):
    """
    As the name implies, TimeRangePhenotype is designed for working with time ranges. If the input data has a start and an end date, use TimeRangePhenotype to identify other events (or patients) that occur within this time range. The most common use case of this is working with 'health insurance coverage' data i.e. on 'OBSERVATION_PERIOD' table. These tables have one or many rows per patient with the start of coverage and end of coverage i.e. domains compatible with TimeRangePhenotype require a START_DATE and an END_DATE column. At it's simplest, TimeRangePhenotype identifies patients who have their INDEX_DATE (or other anchor date of interest) within this time range. Additionally, a minimum or maximum number of days from the anchor date to the beginning/end of the time range can be defined. The returned Phenotype has the following interpretation:

    DATE: If relative_time_range.when='before', then DATE is the beginning of the coverage period containing the anchor phenotype. If relative_time_range.when='after', then DATE is the end of the coverage period containing the anchor date.
    VALUE: Coverage (in days) relative to the anchor date. By convention, always non-negative.

    There are two primary use cases for TimeRangePhenotype:
        1. Identify patients with some minimum duration of coverage prior to anchor_phenotype date e.g. "identify patients with 1 year of continuous coverage prior to index date"
        2. Determine the date of loss to followup (right censoring) i.e. the duration of coverage after the anchor_phenotype event

    ## Data for TimeRangePhenotype
    This phenotype requires a table with PersonID and a coverage start date and end date. Depending on the datasource used, this information is a separate ObservationPeriod table or found in the PersonTable. Use an PhenexObservationPeriodTable to map required coverage start and end date columns. For tables with overlapping time ranges, use the CombineOverlappingPeriods derived table to combine time ranges into a single time range.

    | PersonID    |   startDate          |   endDate          |
    |-------------|----------------------|--------------------|
    | 1           |   2009-01-01         |   2010-01-01       |
    | 2           |   2008-01-01         |   2010-01-02       |

    One assumption that is made by TimeRangePhenotype is that there are **NO overlapping coverage periods**.

    Parameters:
        name: The name of the phenotype.
        domain: The domain of the phenotype. Default is 'observation_period'.
        relative_time_range: Filter returned persons based on the duration of coverage in days. The relative_time_range.anchor_phenotype defines the reference date with respect to calculate coverage. In typical applications, the anchor phenotype will be the entry criterion. The relative_time_range.when 'before', 'after'. If before, the return date is the start of the coverage period containing the anchor_phenotype. If after, the return date is the end of the coverage period containing the anchor_phenotype.
        allow_null_end_date: TimeRangePhenotype checks that anchor date is within the time range of interest. This requires that the start date is not null, and the end date is either null or after the anchor date. If you want to require that the end date is not null, set allow_null_end_date to False.

    Example:
    ```python
    # make sure to create an entry phenotype, for example 'atrial fibrillation diagnosis'
    entry_phenotype = CodelistPhenotype(...)
    # one year continuous coverage prior to index
    one_year_coverage = TimeRangePhenotype(
        relative_time_range = RelativeTimeRangeFilter(
            min_days=GreaterThanOrEqualTo(365),
            anchor_phenotype = entry_phenotype,
            when = 'before',
        ),
    )
    # determine the date of loss to followup
    loss_to_followup = TimeRangePhenotype(
        relative_time_range = RelativeTimeRangeFilter(
            anchor_phenotype = entry_phenotype
            when = 'after',
        )
    )

    # determine the date when a drug was discontinued
    drug_discontinuation = TimeRangePhenotype(
        relative_time_range = RelativeTimeRangeFilter(
            anchor_phenotype = entry_phenotype
            when = 'after',
        )
    )
    ```
    """

    def __init__(
        self,
        name: Optional[str] = "TIME_RANGE",
        domain: Optional[str] = "OBSERVATION_PERIOD",
        relative_time_range: Optional["RelativeTimeRangeFilter"] = None,
        allow_null_end_date: bool = True,
        **kwargs
    ):
        super(TimeRangePhenotype, self).__init__(name=name, **kwargs)
        self.domain = domain
        self.relative_time_range = relative_time_range
        self.allow_null_end_date = allow_null_end_date
        if self.relative_time_range is not None:
            if self.relative_time_range.anchor_phenotype is not None:
                self.add_children(self.relative_time_range.anchor_phenotype)

    def _execute(self, tables: Dict[str, Table]) -> PhenotypeTable:
        table = tables[self.domain]
        table, reference_column = attach_anchor_and_get_reference_date(
            table, self.relative_time_range.anchor_phenotype
        )

        # Ensure that the observation period includes anchor date
        # Allow END_DATE to be null (ongoing periods) if allow_null_end_date is True
        if self.allow_null_end_date:
            table = table.filter(
                (table.START_DATE <= reference_column)
                & ((reference_column <= table.END_DATE) | (table.END_DATE.isnull()))
            )
        else:
            table = table.filter(
                (table.START_DATE <= reference_column)
                & (reference_column <= table.END_DATE)
            )

        if (
            self.relative_time_range is None
            or self.relative_time_range.when == "before"
        ):
            VALUE = reference_column.delta(table.START_DATE, "day")
            EVENT_DATE = table.START_DATE
        else:
            VALUE = table.END_DATE.delta(reference_column, "day")
            EVENT_DATE = table.END_DATE

        table = table.mutate(VALUE=VALUE, EVENT_DATE=EVENT_DATE)

        if self.relative_time_range is not None:
            value_filter = ValueFilter(
                min_value=self.relative_time_range.min_days,
                max_value=self.relative_time_range.max_days,
                column_name="VALUE",
            )
            ibis.options.interactive = True
            table = value_filter.filter(table)

        return self._perform_final_processing(table)

dependencies property

Recursively collect all dependencies of a node (including dependencies of dependencies).

Returns:

Type	Description
Set[Node]	List[Node]: A list of Node objects on which this Node depends.

dependency_graph property

Build a dependency graph where each node maps to its direct dependencies (children).

Returns:

Type	Description
Dict[Node, Set[Node]]	Dict[Node, Set[Node]: A mapping of Node's to their children Node's.

namespaced_table property

A PhenotypeTable has generic column names 'person_id', 'boolean', 'event_date', and 'value'. The namespaced_table prepends the phenotype name to all of these columns. This is useful when joining multiple phenotype tables together.

Returns:

Name	Type	Description
table	Table	The namespaced table for the current phenotype.

reverse_dependency_graph property

Build a reverse dependency graph where each node maps to nodes that depend on it (parents).

Returns:

Type	Description
Dict[Node, Set[Node]]	Dict[Node, List[Node]: A mapping of Node's to their parent Node's.

execute(tables=None, con=None, overwrite=False, lazy_execution=False, n_threads=1)

Executes the Node computation for the current node and its dependencies. Supports lazy execution using hash-based change detection to avoid recomputing Node's that have already executed.

Parameters:

Name	Type	Description	Default
tables	Dict[str, Table]	A dictionary mapping domains to Table objects.	None
con	Optional[object]	Connection to database for materializing outputs. If provided, outputs from the node and all children nodes will be materialized (written) to the database using the connector.	None
overwrite	bool	If True, will overwrite any existing tables found in the database while writing. If False, will throw an error when an existing table is found. Has no effect if con is not passed.	False
lazy_execution	bool	If True, only re-executes if the node's definition has changed. Defaults to False. You should pass overwrite=True with lazy_execution as lazy_execution is intended precisely for iterative updates to a node definition. You must pass a connector (to cache results) for lazy_execution to work.	False
n_threads	int	Max number of Node's to execute simultaneously when this node has multiple children.	1

Returns:

Name	Type	Description
Table	Table	The resulting table for this node. Also accessible through self.table after calling self.execute().

Source code in phenex/node.py

def execute(
    self,
    tables: Dict[str, Table] = None,
    con: Optional[object] = None,
    overwrite: bool = False,
    lazy_execution: bool = False,
    n_threads: int = 1,
) -> Table:
    """
    Executes the Node computation for the current node and its dependencies. Supports lazy execution using hash-based change detection to avoid recomputing Node's that have already executed.

    Parameters:
        tables: A dictionary mapping domains to Table objects.
        con: Connection to database for materializing outputs. If provided, outputs from the node and all children nodes will be materialized (written) to the database using the connector.
        overwrite: If True, will overwrite any existing tables found in the database while writing. If False, will throw an error when an existing table is found. Has no effect if con is not passed.
        lazy_execution: If True, only re-executes if the node's definition has changed. Defaults to False. You should pass overwrite=True with lazy_execution as lazy_execution is intended precisely for iterative updates to a node definition. You must pass a connector (to cache results) for lazy_execution to work.
        n_threads: Max number of Node's to execute simultaneously when this node has multiple children.

    Returns:
        Table: The resulting table for this node. Also accessible through self.table after calling self.execute().
    """
    # Handle None tables
    if tables is None:
        tables = {}

    # Use multithreaded execution if we have multiple children and n_threads > 1
    if len(self.children) > 1 and n_threads > 1:
        return self._execute_multithreaded(
            tables, con, overwrite, lazy_execution, n_threads
        )
    else:
        return self._execute_sequential(tables, con, overwrite, lazy_execution)

visualize_dependencies()

Create a text visualization of the dependency graph for this node and its dependencies.

Returns:

Name	Type	Description
str	str	A text representation of the dependency graph

Source code in phenex/node.py

def visualize_dependencies(self) -> str:
    """
    Create a text visualization of the dependency graph for this node and its dependencies.

    Returns:
        str: A text representation of the dependency graph
    """
    lines = [f"Dependencies for Node '{self.name}':"]

    # Get all dependencies
    all_deps = self.dependencies
    nodes = {node.name: node for node in all_deps}
    nodes[self.name] = self  # Add self to the nodes

    # Build dependency graph
    dependency_graph = self._build_dependency_graph(nodes)

    for node_name in sorted(nodes.keys()):
        dependencies = dependency_graph.get(node_name, set())
        if dependencies:
            deps_str = ", ".join(sorted(dependencies))
            lines.append(f"  {node_name} depends on: {deps_str}")
        else:
            lines.append(f"  {node_name} (no dependencies)")

    return "\n".join(lines)