Project to analyze long term covid-19 outcomes
This GitHub directory has the following:
- Code to create charts relevant to longhauler analysis
- SQL code that can make files form an i2b2 database
- R code that analyses and renders the data files into graphs
- Simulated data from two different databases
- A simulated database from the Italian sites (Sim)
- An i2b2 project simulated database (i2b2syn)
- Tables that map icd9 and icd10 codes into PheCodes
- Publications relevant to longhauler analysis
To modify this document, use markdown
The code is able to create graphs of what is afflicting patients who have had COXID-19 infections and now are being followed in the health care system.
-
The current analysis shows just the codes (grouped into PheNorm codes) over time. some issues are:
- It does not account for patients who are lost to follow-up. An analysis of only patients who have codes after 90 days (for example) might be better. In the "Versions" folder under the "subtraction-calculation" folder is a plan for developing this.
- Instead of just showing days post admission, it may be better to group the codes into pre-covid, covid admission, and number of days after discharge.
- Could add stratification displays (perhaps in SHINEY) for severity, age, race, ethnicity.
-
The i2b2syn SQL code could be enhanced to :
- Produce age/race/severity/ethnicity in the PatientSummary table.
- Produce a file (and code) that does not always set in-hospital="1". This was done to adapt to the Sim analysis which only shows in-hospital patients.
In the syn(thetic) data there are not many readmissions, usually only admission (encounter #1) and outpatient (encounter #2).