Page 272 - Artificial Intelligence in the Age of Neural Networks and Brain Computing

CHAPTER

Multiview Learning in

Biomedical Applications 13

Angela Serra, Paola Galdi, Roberto Tagliaferri
NeuRoNe Lab, DISA-MIS, University of Salerno, Fisciano, Italy

CHAPTER OUTLINE

1. Introduction .......................................................................................................265
2. Multiview Learning.............................................................................................266
2.1 Integration Stage................................................................................. 267
2.2 Type of Data ....................................................................................... 268
2.3 Types of Analysis ................................................................................ 269
3. Multiview Learning in Bioinformatics...................................................................269
3.1 Patient Subtyping ............................................................................... 269
3.2 Drug Repositioning.............................................................................. 272
4. Multiview Learning in Neuroinformatics...............................................................273
4.1 Automated Diagnosis Support Tools for Neurodegenerative Disorders....... 273
4.2 Multimodal Brain Parcellation .............................................................. 275
5. Deep Multimodal Feature Learning ......................................................................275
5.1 Deep Learning Application to Predict Patient’s Survival.......................... 276
5.2 Multimodal Neuroimaging Feature Learning With Deep Learning ............. 276
6. Conclusions.......................................................................................................277
References .............................................................................................................278

1. INTRODUCTION
In the past decades there has been a growing interest in applying data integration
methods in the field of biomedical research, with the consequent proliferation of sci-
entific literature devoted to this topic.
Multiview learning is the branch of machine learning concerned with the analysis
of multimodal data, that is, patterns represented by different sets of features extracted
from multiple data sources. Classical examples of multiview medical data are the
different clinical tests to which a patient can be subjected (see Fig. 13.1). These
data are generally not homogeneous (for example, they can consist of images, signals,
or text) but when analyzed together they help to better understand the patient’s
medical condition.
The reasons for the fast spreading of this learning approach lie in the constant
increase of real-world problems where heterogeneous data are available for
265
Artificial Intelligence in the Age of Neural Networks and Brain Computing. https://doi.org/10.1016/B978-0-12-815480-9.00013-X
Copyright © 2019 Elsevier Inc. All rights reserved.