Heart monitoring apparatus and method

A heart monitoring method comprising the steps of:

• obtaining an electrocardiograph signal from a patient during a monitoring phase,
• preprocessing the electrocardiograph signal to enhance the salient features of the electrocardiograph signal and suppress the noise, and to generate a plurality n of values representative of the features of the electrocardiograph signal,
• forming an n dimensional vector from said plurality n of values,
• comparing the n dimensional vector with a stored plurality m of n dimensional reference vectors defining an n dimensional Kohonen feature map to determine the proximity of the n dimensional vector to said reference vectors, and
• outputting a signal if it is determined that the n dimensional vector is within or beyond a threshold range of said reference vectors.

Heart monitoring apparatus comprising:

• input means for receiving an electrocardiograph signal from a patient;
• preprocessing means for processing said electrocardiograph signal to suppress the noise and extract important features of the electrocardiograph signal to obtain an n dimensional vector comprising a plurality n of values representative of the important features of said electrocardiograph signal;
• storage means for storing a plurality m of n dimensional reference vectors;
• neural network means for receiving said n dimensional vector, for comparing said n dimensional vector with said stored plurality m of n dimensional reference vectors defining an n dimensional volume to determine the proximity of said n dimensional vector to said n dimensional volume, and for outputting an indication of whether said n dimensional vector lies within or beyond a threshold range of said n dimensional volume.

Heart monitoring apparatus comprising:

• measuring means for measuring the activity of a patient's heart and outputting an electrocardiograph signal during a learning phase and a monitoring phase;
• preprocessing means for processing said electrocardiograph signal to extract important features therefrom to generate an n dimensional vector comprising a plurality n of values representative of the features of the electrocardiograph signal;
• and
• Kohonen neural network means for, during the learning phase, using said n dimensional vector to generate an n dimensional Kohonen feature map with a plurality m of reference vectors defining a normal range of electrocardiograph features for said patient, and for during the monitoring phase, comparing said n dimensional vector with said plurality of m dimensional reference vectors defining the n dimensional Kohonen feature map to determine the proximity of said n dimensional vector to said reference vectors, and to output a signal if it is determined that said n dimensional vector is within or beyond a threshold range of said reference vectors.