LAUSR

Availability of the information related to the medical tests are wide and thus the demand for advanced analysis is continuously growing. In the recent past, a number of automatic disease detection algorithms are generated. The studies focus on various life threatening diseases such as cardiovascular diseases or cervical diseases or cancerous growth in tissues or non-malignant tumours. Nevertheless, considering various reports from the health organizations and government survey reports, the researchers have highest focus on cardiovascular diseases. The major area of focus is to determine or predict the possibilities of heart diseases based on predictive analysis. It is been learnt from the parallel research outcomes that the use of neural network based analytical frameworks are highly accurate for these purposes. The majority of the research attempts have demonstrated the use of single activation function for determining the predictive value. Nonetheless, the activation functions have significantly different properties and the effective selection of the activation function can generate higher accuracy. In the other hand ineffective selection of the activation functions, may also lead to incorrect determination of the disease and disease severity. Thus for a clinical decision support system, it is very crucial to have effective activation function in the framework. Hence, this work attempts to identify the characteristics of these activation functions and provides a framework for dynamic selection of the activation function based on the nature of the dataset. Also, the training strategies for any neural model play a major role for disease detection and prediction, which is again subjected to the time complexity of the learning function. Nevertheless, this work proposes to identify the optimal deep learning function by deploying the proposed framework. The major outcome of this work demonstrates nearly 96% accuracy for disease prediction due to the dynamic selection of the activation function and establishes the novelty of the research by defining the severity of the diseases.