# Windows

# calculate_days

windows.calculate_days(df, name, date_ref)

Helper function to calculate the number of days between a reference event

# Arguments

df: Pandas DataFrame
name: The name of the result column (will be appended with "Days").
date_ref:
- String: The name of the column in df to reference against.
- Pandas DataFrame: Contains two columns: PatientID and a date column as the reference date.

# Returns

The same Pandas DataFrame, df, with an extra calculated column.

# Example

String Example

from ehr_functions.features import windows
import pandas as pd 

df = pd.DataFrame({
    'PatientID': [1, 1, 1, 1],
    'EncounterDate': ['01/01/2020', '01/05/2020', '01/18/2020', '01/30/2020'],
    'InjuryDate': ['01/07/2020', '01/07/2020', '01/07/2020', '01/07/2020'],
    'Diagnosis': ['A', 'F', 'D', 'C'],
})

features = windows.calculate_days(df, 'Injury', 'InjuryDate')
print(features.head())

PatientID	EncounterDate	InjuryDate	Diagnosis	InjuryDays
1	2020-01-01	2020-01-07	A	-6
1	2020-01-05	2020-01-07	F	-2
1	2020-01-18	2020-01-07	D	11
1	2020-01-30	2020-01-07	C	23

DataFrame Example

from ehr_functions.features import windows
import pandas as pd 

df = pd.DataFrame({
    'PatientID': [1, 1, 1, 1],
    'EncounterDate': ['01/01/2020', '01/05/2020', '01/18/2020', '01/30/2020'],
    'Diagnosis': ['A', 'F', 'D', 'C'],
})

lookup = pd.DataFrame({
    'PatientID': [1],
    'InjuryDate': ['01/07/2020'],
})

features = windows.calculate_days(df, 'Injury', lookup)
print(features.head())

PatientID	EncounterDate	Diagnosis	InjuryDays
1	2020-01-01	A	-6
1	2020-01-05	F	-2
1	2020-01-18	D	11
1	2020-01-30	C	23

# build_windows

windows.build_windows(df, windows, days_ref, flip_negative=False)

A common technique to count the number of encounters that occur within each window/timeframe. This method expects for a column that exist that contains the number of days since an event, where 0 is the event of interest to relate against.

# Arguments

df: Pandas DataFrame
windows: List of windows with the begin and end day values
days_ref: Name of the column containing days to reference.
flip_negative: A helper variable to multiply all windows by -1 rather than manually having to do it.

# Returns

Pandas DataFrame with PatientID and a column for each window.

# Example

from ehr_functions.features import windows
import pandas as pd 

df = pd.DataFrame({
    'PatientID': [1, 1, 1, 1],
    'EventDays': [0, 5, 9, 15],
    'Diagnosis1': ['A', 'F', 'D', 'C'],
    'Diagnosis2': [None, 'B', 'B', 'A'],
    'Diagnosis3': [None, None, 'C', 'A'],
})

features = windows.build_windows(df, [[0, 10], [10, 30]], 'EventDays')
print(features.head())

PatientID	Window_0	Window_1
1	3	1

← Demographics