TBI Patient Endotype Clustering - using LCA to demonstrate the existence of patient endotypes in a TBI population that differ in their comorbidity profiles and clinical outcomes.
- Requires MIMIC III v1.4 database to be built in SQLite format locally
tbi2_admit_icd.xlsx
- contains all TBI patient demographic and comorbidity information (e.g., DOB, sex, comorbidity ICDs, discharge status)- Query found in
MIMIC3 Linked TBI Desktop v2.accdb
MS Access file undertbi_admit_icd
query - Note that the
.accdb
file must be manually linked to the locally built SQLite MIMIC III database in order for the query to be run
- Query found in
tbi2_admit_proced.xlsx
- contains neurosurgical intervention events for TBI patients- Full query found in
sql_queries.py
- Full query found in
tbi2_admit_chevents_gcs.xlsx
- contains recorded GCS scores for TBI patients- Full query found in
sql_queries.py
- Full query found in
- The core data sources are combined and processed by
processing.py
to generatetbi2_admit_icd_dates_nsx_gcs_elix.xlsx
containing all key clinical information of interest - Using ICD-9 data from
tbi2_admit_icd_dates_nsx_gcs_elix.xlsx
, comorbidity endotypes are identified usingcluster.r
and appended for each patient onto the original Excel file to createtbi2_admit_icd_dates_nsx_gcs_elix_annotated.xlsx
- Statistical analyses and visulizations are generated using data from
tbi2_admit_icd_dates_nsx_gcs_elix_annotated.xlsx
using the following files:graph_builder.py
andgraph_renderer.py
for comorbidity network graphs and relative risk between comorbiditiesvisulization_stats.py
for conventional descriptive statistics and stratified analyses (by GCS and age)cluster.r
for visulization of comorbidity distributionsvisulization_stats.r
for jitter plot and ANCOVA and regression analysis for endotypes