teamwork is a Python library designed to aid in the study of medical care team collaboration.
Teamwork is a Python library designed to aid in the study of medical care team collaboration.
Specifically, the library takes electronic health record encounter-level data (both discharges and notes) as inputs and does the following:
1) Identifies the members of medical care teams for unique patient visits based on medical note meta-data
2) Builds a network graph for each care team based on their common patients in the previous 90 days. The nodes of the graph are note authors, and the edge weights are the number of days where the authors have both written a note on common patient
The resulting care team network graphs can be used to compute the cumulative collaborative experience of care teams and other metrics.
Use the package manager pip to install teamwork.
pip install teamwork
import pandas as pdfrom teamwork import teamwork as tw# read medical notes (3 columns) into pandas DataFrametest_df = pd.read_csv('../data/sample_notes.csv')# enforce date columnstest_df['date'] = pd.to_datetime(test_df['date'])test_df['arrive_date'] = pd.to_datetime(test_df['arrive_date'])# set a 90 day window to find collaboration among care teamsWINDOW = 90# identify care teams within 2 day incrementsSTEP = 2corpus = tw.TeamworkCorpus(test_df)# now you can do something with the care team network graphfor visit_id, item in corpus.team_experience_dict.items():print(item['graph'])
| Contributor | Role |
|---|---|
| Henry Philofsky | Concept, SME |
| Grant DeLong | Developer, Architect |