Detection of hemodynamic changes in clinical monitoring by time-delay neural networks

Small changes that occur in a patient's physiology over long periods of time are difficult to detect, yet they can lead to catastrophic outcomes. Detecting such changes is even more difficult in intensive care unit (ICU) environments where clinicians are bombarded by a barrage of complex monitoring signals from various devices. Early detection accompanied by appropriate intervention can lead to improvement in patient care. Neural networks can be used as the basis for an intelligent early warning system. We developed time-delay neural networks (TDNN) for classifying and detecting hemodynamic changes. A matrix of physiological parameters were extracted from raw signals collected during cardiovascular experiments in mongrel dogs. These matrices represented several episodes of stable, decreasing, and increasing cardiac filling in normal, exerted, and heart failure conditions. The TDNN were trained with these matrices and subsequently tested to predict unseen cases. The TDNN perform remarkably not only in identifying all hemodynamic conditions, but also in quickly detecting their changes. On average, the networks were able to detect the hemodynamic changes in less than 1 s after the onset. Based on the results of this pilot investigation, the use of this form of TDNN to successfully predict hemodynamic conditions appears to be promising.