Predictors for extubation failure in COVID-19 patients using a machine learning approach

Table 1 Included parameters

Predictor	Aggregation
Age
Apache-II score	Mean last 24 h
Body mass index (BMI)
C-reactive protein	Mean last 24 h
Creatinine	Mean last 24 h
FiO₂	Mean last 24 h, Last value
Fluid balance	Sum
Glasgow coma scale	Mean last 24 h
Glucose	Mean last 24 h
Cardiac comorbidity
Diabetes comorbidity
Respiratory comorbidity
Renal comorbidity
Heart rate	Mean last 24 h
Hematocrit	Mean last 24 h
Gender
Duration of mechanical ventilation
Leukocyte count	Mean last 24 h
Benzodiazepine dose	Given in last 24 h
Clonidine dose	Given in last 24 h, Total dose last 24 h
Dexmedetomidine dose	Given in last 24 h, Total dose last 24 h
Fentanyl dose	Given in last 24 h, Total dose last 24 h
Haloperidol dose	Given in last 24 h, Total dose last 24 h
Midazolam dose	Given in last 24 h, Total dose last 24 h
Propofol dose	Given in last 24 h, Total dose last 24 h
Quetiapine dose	Given in last 24 h, Total dose last 24 h
p0.1	Mean last 24 h, Last value
P/F ratio	Mean last 24 h
PCO₂ arterial	Last value
PEEP	Last value
pH	Last value
Hours since last proning session
Pressure above PEEP	Mean last 24 h, Last value
Respiratory rate	Mean last 24 h
RASS score	Mean last 24 h, Last value
Thrombocyte count	Mean last 24 h
Tidal volume per kg ideal body weight	Mean last 24 h, Last value
Duration of controlled mode
Hours since last controlled mode
Ventilatory ratio	Mean last 24 h

Overview of included parameters and their aggregation
FiO₂: Fraction of inspired oxygen, PEEP: positive end expiratory pressure, P/F ratio: PaO₂/FiO₂ ratio, PCO₂: partial pressure of carbon dioxide, RASS score: Richmond Agitation and Sedation Scale

ISSN: 1364-8535