OMOP Use Case: Features for extubation modelling

[myyong]

This forms part of a series of issues to document how SqlSynthGen is used to create a synthetic version of CCHIC OMOP data. This specific issue will document how we selected and created synthetic data containing features used for extubation modelling.

These suggestions of features were taken from Slack messages:

Initial set of suggestions

• [ ] ethnicity
• [ ] hospital admission/discharge dates (for LoS)
• [ ] hospital mortality
• [ ] simple vitals - heart rate, blood pressure etc, RR
• [ ] simple labs - Hb, WCC, U&E
• [ ] presence of NG feeding tubes
• [ ] presence of ETT tubes
• [ ] mandatory resp rate
• [ ] airway pressures (e.g. PEEP, peak pressure)
• [ ] feeding rates

Questions about what we can do

• [ ] Congruency/internal logic eg Systolic Diastolic
• [ ] Preengineered variables
• [ ] Outlier events
• [ ] Temporal variation

First set of suggestions

• [ ] Pulse rate
• [ ] Arterial oxygen saturation
• [ ] Respiratory rate
• [ ] Systolic/Diastolic blood pressure
• [ ] Urine output per hour
• [ ] Body temperature
• [ ] Inspired oxygen concentration
• [ ] SOFA (Sequential Organ Failure Assessment) score
• [ ] Oral Fluid input
• [ ] Tidal volume
• [ ] Ventilator delivered minute volume
• [ ] End tidal carbon dioxide concentration
• [ ] Total breath rate
• [ ] ph of blood
• [ ] Carbon dioxide in blood
• [ ] Oxygen in blood
• [ ] Base excess in blood by calculation
• [ ] Chloride [Moles/volume] in blood

[myyong]

In order to get the statistical distributions of ethnicity reflected in synthetic data, we perform the following steps:

1. Add the queries to real data into config.yaml example here

2. Write row generator function to reference the source stats created above example here

3. Name the row generator function in config.yaml and assign function outputs to table columns example here