Skip to content

Visualizing event sequences

When studying sequences of events (e.g care trajectories, drug sequences, ...), it might be useful to visualize individual sequences. To that end, we provide the plot_event_sequences function to plot individual sequences given an events dataframe.

Load a synthetic dataset

An events dataset has been created to illustrate the visualization function. It can be load as follows :

from eds_scikit.datasets.synthetic.event_sequences import load_event_sequences

df_events = load_event_sequences()

The df_events dataset contains occurrences of 12 events, derived from 7 events' families ("A", "B", "C", "D", "E", "F", "G).

Events can be both one-time and continuous.

An index_date is also provided and refers to the inclusion date of each patient in the cohort.

The first raws of the dataframe are as follows :

df_events.head()
# Out:  person_id event_family event event_start_datetime event_end_datetime index_date
0          1           A        a1        2020-01-01         2020-01-02      2020-01-01
1          1           A        a2        2020-01-03         2020-01-04      2020-01-01
2          1           B        b1        2020-01-03         2020-01-06      2020-01-01
3          1           C        c1        2020-01-05         NaT             2020-01-01
4          1           C        c2        2020-01-06         2020-01-08      2020-01-01

Visualize individual sequences

Basic usage

Individual sequences of events can be plotted using the plot_event_sequences function :

from eds_scikit.plot.event_sequences import plot_event_sequences

chart = plot_event_sequences(df_events)
chart

Image title

Advanced parameters

Further configuration can be provided, including : - dim_mapping : dictionary to set colors and labels for each event type. - family_col: column name of events' families. - list_person_ids: List of specific person_id - same_x_axis_scale: boolean to set all individual charts to the same scale

Here we provide an exemple of dim_mapping, and we plot sequences aggregated following the event_family classification.

dim_mapping = {
    "a1":{"color":(255, 200, 150), "label":"eventA1"},
    "a2":{"color":(255, 150, 150), "label":"eventA2"},
    "a3":{"color":(255, 100, 150), "label":"eventA3"},
    "b1":{"color":(100, 200, 150), "label":"eventB1"},
    "c1": {"color":(50, 255, 255), "label":"eventC1"},
    "c2": {"color":(50, 200, 255), "label":"eventC2"},
    "c3": {"color":(50, 100, 255), "label":"eventC3"},
    "d1": {"color":(180, 200, 100), "label":"eventD1"},
    "d2": {"color":(180, 150, 100), "label":"eventD2"},
    "e1": {"color":(130, 60, 10), "label":"eventE1"},
    "f1": {"color":(255, 0, 0), "label":"eventF1"},
    "g1": {"color":(100, 0, 200), "label":"eventG1"},
}

plot_event_sequences(
    df_events,
    family_col = 'event_family',
    dim_mapping = dim_mapping,
    same_x_axis_scale=True,
    title="Event sequences",
)

Image title

Please check the documentation for further details on the function's parameters.