A Ventricular Assistive Device (VAD) is a mechanical pump used to assist the functioning of a weak heart. A catastrophic obstruction in the VAD system could cost the patient their life. This paper discusses a fault detection approach using the commercially available Jarvik 2000 Flowmaker® VAD in a closed loop circuit that incorporates the ability to alter common causes of VAD congestion. Principal Component Analysis, a data compression technique used to discover patterns in data of high dimension, is implemented using frequency analysis of the VAD’s acoustic signature. This is followed by a health classification based on Bayes theorem. The classification results indicate that this technique is accurate to a high degree in detecting three levels of obstruction in the VAD system.

This content is only available via PDF.
You do not currently have access to this content.