Patients on weaning trials from mechanical ventilation classified with neural networks and feature selection.

摘要: One of the challenges in intensive care is process weaning from mechanical ventilation. We studied differences respiratory pattern variability between patients capable maintaining spontaneous breathing during trials and that fail to maintain breathing. In this work, neural networks were applied study these differences. 64 ventilation are studied: Group S with 32 Successful F Failed reconnected. A performance 64.56% well classified was obtained using a network trained whole set 35 features. After application feature selection procedure (backward selection) 84.56% only 8