Hybrid SVM classification technique to detect mental stress in human beings using ECG signals

In recent years, stress has become ingrained part of our life, being stressed by our financial worries, our job, etc. Stress causes physical illnesses, such as heart attacks, arthritis, and chronic headaches or psychological diseases like mental illness, anger, anxiety, and depression. There are several research works coming up to resolve the limitations on measuring, analyzing and identifying the human stress levels Amongst the many stress monitoring methods the more reliable method to determine the human stress level is to use physiological signals. In this work, Heart Rate Variability (HRV) determined from ECG signal, an efficient parameter to detect the stress level is used. The features extracted from HRV are given as input, to the two stage classifier, to classify the stress into one of the four levels as no stress, low stress, medium stress and high stress. In the first stage of the classifier, Self Organizing Map is used to classify the stress into two classes as `stress level 1´(no stress & low stress) and `stress level 2´ (medium stress & high stress). In the second stage Support Vector Machine is used with RBF kernel to subdivide the `stress level 1´ into two classes `No Stress´ and `Low Stress´. The stress level 2 is subdivided into twoclasses `Medium Stress´ and `High Stress´. The performance of this hybrid structure is better and the efficiency of classification is 91%.