Efficient Deep Learning Approaches for Health Informatics
暂无分享,去创建一个
Abstract The need of data analytics in health informatics for better decision making is a challenging domain for the past decade. This stimulates more interest of researchers for the design of data driven models on the basis of machine learning such as deep learning models for health informatics. Deep learning is an emerging machine learning technique with various applications for health care monitoring such as medical imaging, bioinformatics, pervasive sensing and public health care, etc. There exist various deep learning approaches with their own pros and cons for health informatics to address multiple challenges in medical data processing such as high dimensional, heterogeneous, incomplete, unstructured biomedical, temporally dependent and irregular data, and so on. Deep learning techniques have their own added characteristics suited for health informatics such as enhanced performance, end-to-end learning embedded with features learning, executing complex and multimodal data, etc. Deep learning has proven its performance in multiple domains and showed progressive improvements, which in turn provide more opportunities to design new models for health informatics. Although the existing deep learning approaches designed for health informatics addressed multiple challenges, processed and analyzed complex data, new medical issues and kinds of data are arising every day in the health care domain. Hence, there is a need for in-depth exploration of deep learning approaches to address various challenges related to the health care features. Thus this book chapter deals with detailed analysis of various deep learning approaches in health informatics. A comparative analysis shows the pros and cons of each model and their applications in the health care domain. Also the work explores the challenges and limitations existing in the deep learning models and highlights the new kinds of data which have to be identified to connect with health care information work flow and also for clinical decision making.