ECG

Terms from Artificial Intelligence: humans at the heart of algorithms

An ECG (electo-cardiogram) is a device to measure electrical activity in the heart. The output is the heart beat trace that you will have seen in medical dramas. Analysing an ECG is challenging for time-series methods as it consist of bursts of actuvity (the heart beat) where the shape of each burst and the spacing between them are each important. Furthernore the detailed shape of each burst and the length also vary. Traditional time-series analysis methods prefer data that is more uniform in structure (even if stochastic), as do some neural metods. Some form of multi-level analysis is typically required either explicitly by applying different methods to separate the trace into components and then analyse each, or implicitly, for example in the layers of a deep neural network.

Used on page 155