Trend detection of pseudo-random variables using a exponentially mapped past statistical approach: an adjunct to computer assisted monitoring.

The early detection of deterioration in a patient depends on recognising trends: a capability poorly developed in monitor systems. Using exponentially mapped past (EMP) statistical variables, it has been proved that the difference between two EMP means is a function of only one variable, trend of input. Some of the signal variance appears on the difference signal, and so interpretation depends on a statistical approach based on knowledge of the normal variability of the monitored variable. The programs developed and built in a dedicated form utilise parallel hybrid computation of continuous data, but the principles are equally applicable to digital techniques with discrete data.