Removed the total_iv_time < 24 Removed the last iv, put back. Otherwise, the dataset is imbalanced.
Data collection window 6 hours. Finish the incompation I should also treat the gap smaller than 1 hour as continuous ventilation.
Positive sample: the patient is on an invasive ventilation and will have a failed extubation Negative sample: the patient is on an invasive ventilation and will have a successful extubation. (no re-intubation within 72 hours)
Without last iv: Train: Extubation Failed(positive): 15556 Extubation Succeed(negative): 3807
Test: Extubation Failed(positive): 3786 Extubation Succeed(negative): 455
Total: Extubation Failed(positive): 19342 Extubation Succeed(negative): 4262
With last iv: Train: Extubation Failed(positive): 15556 Extubation Succeed(negative): 18578
Test: Extubation Failed(positive): 3786 Extubation Succeed(negative): 3929
Total: Extubation Failed(positive): 19342 Extubation Succeed(negative): 22507