Applications of Deep Learning in Healthcare and Biomedicine

Abstract

The increasing advancements and improvements in medicine and healthcare in the past few decades have ushered us into a data-driven era where a huge amount of data is collected and stored. With this change, there is a need for analytical and technological upgradation of existing systems and processes. Data collected is in the form of Electronic Health Data taken from individuals or patients which can be in the form of readings, texts, speeches or images. A means to Artificial Intelligence—‘Machine Learning’ is the study of models that computer systems use to self-learn instructions based on the weight of parameters without being provided explicit instructions. Parallelly with biomedical advancements in the past decade, it has been observed that there has been an increasing refinement of algorithms and tools of machine learning. Deep Learning is one of the more promising of these algorithms. It is an Artificial Neural Network that designs models computationally that are composed of many processing layers, in order to learn data representations with numerous levels of abstraction. Research suggests that deep learning might have benefits over previous algorithms of machine learning and its’ suggestive better predictive performance is, hence garnering significant attention. With their multiple levels of representation and results that surpass human accuracy, deep learning has particularly found widespread applications in health informatics and biomedicine. These are in the field of molecular diagnostics comprising pharmacogenomics and identification of pathogenic variants, in experimental data interpretation comprising DNA sequencing and gene splicing, in protein structure classification and prediction, in biomedical imaging, drug discovery, medical informatics and more. The aim of this chapter is to discuss these applications and to elaborate on how they are being instrumental in improving healthcare and medicine in the modern context. Algorithms of deep learning show an improved potential in learning patterns and extracting attributes from a complex dataset. We would first introduce deep learning and developments in artificial neural network and then go on to discuss its applications in healthcare and finally talk about its’ relevance in biomedical informatics and computational biology research in the public health domain. In the end future scope of deep learning algorithms would be discussed from a modern healthcare perspective.