Data Visualization of Section Identification

Set up enviornment and import notebook modules

import itertools as it
from harness import AppNotebookHarness, create_subplots

text_cells = False
harness = AppNotebookHarness()
analyzer = harness.get_analyzer()

Plot Section 3D

24 hour events, addendum, history hospital: this is what patient's time in hospital--their story asses and plan, and hosptial course are more overview of treatment plan

sec_comb = 0
analyzer = harness.get_analyzer(
    plot_components=3,
    outlier_proportion=0.001,
    calc_per_section=False,
    tfidf_threshold = 0.000001,
    tfidf_sections = set({0: 'past-medical-history past-surgical-history',
                          1: 'past-surgical-history',
                          2: 'past-medical-history',
                          6: 'code-status assessment-and-plan',
                          7: 'wet-read indication',
                          8: 'medication-history past-medical-history',
                          9: 'disposition labs-imaging',
                          10: 'impression image-type',
                          11: 'findings image-type',
                          12: 'chief-complaint discharge-condition',
                          13: 'patient-test-information procedure',
                          14: 'flowsheet-data-vitals indication',
                          15: '24-hour-events patient-test-information',
                          16: 'clinical-implications code-status',
                          17: 'infusions last-dose-of-antibiotics other-medications',
                          18: 'flowsheet-data-vitals infusions last-dose-of-antibiotics',
                          19: 'clinical-implications infusions last-dose-of-antibiotics',
                          20: 'clinical-implications contrast last-dose-of-antibiotics',
                          21: 'clinical-implications disposition patient-test-information',
                          22: 'facility image-type indication',
                          23: 'clinical-implications image-type indication',
                          24: 'flowsheet-data-vitals indication infusions',
                         }[sec_comb].split()))
#analyzer.set_tfidf_section_by_index(-1)
#analyzer.dim_reduction_meth = 'svd'
#analyzer.height = 800
if 1:
#    analyzer.normalize = 'unit'
#    analyzer.normalize = 'standardize'
    analyzer.axis_range = 2.7
else:
    analyzer.plot_type = 'rand'
    analyzer.axis_range = 200
    analyzer.tfidf_sections = {'past-medical-history'}
analyzer.section_plot_3d()
for dr in analyzer.dim_reducers:
    dr.write()

dim: 3
reduction_meth: pca
normalize: None
model_args:
description:
    components: 3
    explained_varainces:
        0.15547815062157622
        0.0437924327973983
        0.034508676123682136
    noise: 0.03109838247584281
    total_variance: 0.23377925954265666
n_points: 3713

Plot by Age Type

analyzer = harness.get_analyzer(
    plot_type='age'
    #normalize = True
)
analyzer.section_plot_3d()

TSNE on Sections

import itertools as it
from harness import AppNotebookHarness, create_subplots

text_cells = False
harness = AppNotebookHarness()
analyzer = harness.get_analyzer(
    dim_reduction_meth = 'tsne',
    tfidf_threshold = 0.1,
    outlier_significance=0.3,
#    normalize='unit',
)
#analyzer.tfidf_sections = {'labs-imaging', 'past-surgical-history', 'physical-examination'}
#analyzer.tfidf_sections = {'past-medical-history', 'past-surgical-history'}
analyzer.section_plot_3d()

Plot discharge summary against radiology

analyzer = harness.get_analyzer(
    plot_type='shared',
    tfidf_threshold = 0.1,
    outlier_significance=0.1,
)
analyzer.section_plot_3d()