Efficient Disease Risk Prediction based on Deep Learning Approach

Abstract

In recent years, the advancement of artificial intelligence (AI) and the progress of machine intelligence has allowed the people to perceive the great future of AI in the healthcare field. Deep learning technology has shown the promising results in early disease prediction. The performance of multi disease prediction has been improved dramatically due to progressive development from machine learning to deep learning technology. The most difficult task is accurate and early disease prediction. It aims to demonstrate the significant relationship between deep learning and healthcare industry mainly for early disease prediction. In this paper, deep learning based multi disease prediction such as diabetes, breast cancer and covid 19 detection are proposed and analysed. The selected deep learning models in this paper were ANN and CNN. These networks were chosen, as they contain only less number of layers than complex architectures like Densenet and Resnet model. Kaggle datasets are used for all three different diseases for efficient detection. The performance of deep learning classification algorithms is evaluated using a variety of evaluation metrics such as accuracy, precision, sensitivity and specificity. Our obtained results shows that ANN and deep CNN model achieves higher accuracy than existing machine learning models. Our proposed model has shown the greater accuracy of 73.37%, 96.49%, 96.66% in diabetes, breast cancer and covid-19 disease detection.