Early ICU Mortality Predicition

Predicit hospital mortality for patient admit#ted to ICU, using only data collected from the first 6 hours of ICU admission.

Results

• The Neural Net produced the best accuracy at an accuracy of 77%

Most Important Features

• The most important features from the Random forest are:
• Mean respiration rate across the 6 hours
• Maximum pCO2 across the 6 hours
• Minmum glucose measurement across the 6 hours
• Admission age

For more results and discussion see detailed report