Big Data and Deep Learning in Healthcare

Abstract

The huge amount of digital data in the field of healthcare has given rise to research in machine learning based on the data-driven approach. Deep learning is gaining primary importance in machine learning as a strong technique for big data as it has the capability to represent features and pattern recognition. This chapter gives an inclusive view about deep learning with big data in health care. Various types of datasets from clinical nature are discussed in detail that includes medical images, clinical notes, lab results, demographic informatics, and vital signs. The big data and significant considerations for use of machine learning in health care is discussed. Besides some models of deep learning that are commonly used along with their characteristics are discussed. Various deep learning with big data applications for medical and clinical data are illustrated. It is still a big challenge to apply deep learning techniques to data collected from medical organizations but at the same time it is beneficial to anticipate to a promising future for deep learning with big data applications in healthcare in the direction of precise medication. Commonly used data mining techniques like Naive Bayes, SVM, Random Forest, KNN, Decision Tree, ANN and CNNare applied on PIMA dataset without using any preprocessing techniques for prediction of diabetes mellitus at early stage. Finally, the effect of pre-processing techniques namely Z-Score and Min-Max Scalar on dataset is shown by running ANN algorithm on same dataset but after application of pre-processing techniques. The results obtained shows significant improvement in early diagnosis of diabetes mellitus after the application of pre-processing techniques on dataset.