The use of artificial neural networks featuring biological guidance, like the Restricted Boltzmann Machine (RBM), has generated a lot of attention because of its superior estimation capabilities for challenging tasks. They are useful for many applications, especially machine learning techniques. A powerful DNN design based on integral stochastic computation is provided in the current system. Quasi-synchronous execution of the suggested architecture results in a 33% decrease in the use of energy. The suggested method provides a block processing unit that utilizes energy-efficient stochastic computing. The data frame is handled block by block. These results are further examined using the correlation and decorrelation functions. The proposal is to classify atrial fibrillation data as normal or pathological. A formula for the Matthew's correlation constant (MCC) is developed.