Prediction of Mechanical Ventilator Weaning Outcome - A Deep Learning Approach

Kuei-Hung Shen, Yun-Ju Yu, Szu-Yin Chen, En-Ming Chang, Hsiu-Li Wu, Cheng-Kuan Lin, Yu-Chee Tseng

Abstract: Mechanical Ventilation (MV) is a type of therapy that helps patients breathe or breathes for patients when they can’t breathe on their own. Doctors and therapists determine whether a patient is ready for weaning of mechanical ventilation based on vitals and various medical test results. This study explores deep learning methods applied to prediction of mechanical ventilator weaning outcome using multiple physiological parameters. Our experiments showed a validation accuracy of 0.682 with limited samples and less features compared to similar studies. We expect to see improved performance with more data and features collected.