Early Detection of Paralysis in Human Subjects from ECG Signal of Human Subjects using AI

In this study, a neural network with more than 80 deep learning layers has been designed and tested for early detection and validation of paralytic symptoms in human subjects. The Electrocardiogram (ECG) and other anthropogenic data of 500 health and 500 abnormal human subjects has been used to train the neural network for conclusive results. As per absolute error value of 0.22 and relative AUC value of 87.47%, it has been observed that the T wave of affected patients is low than the health subjects and other segments such as QT interval and ST segments affected by low potassium level in affect human subjects. The results clearly indicated the slow intervals in both time and frequency domain features of measured ECG signals from affect subjects. As per statistical analysis, there is a significant relation found between the domain features and level of body parameters for paralysis detection. Our study supports the research community for making strategies for early diagnosis of causes such as paralysis.