Page 276 - Artificial Intelligence in the Age of Neural Networks and Brain Computing

3. Multiview Learning in Bioinformatics 269

2.3 TYPES OF ANALYSIS
The choice of the analysis to be performed is evidently determined by the type of
data involved in the experiments and by the kind of integration that needs to be
accomplished. Two broad categories of analyses can be identified: integrative
analysis and metaanalysis. Metaanalysis in based on previous results, and in this
sense it can be considered as a late integration approach. It consists in aggregating
summary statistics from several studies and therefore it requires data to be homoge-
neous [3,4]. On the other hand, integrative analysis is a more flexible methodology,
since it allows the fusion of different data sources to get more stable and reliable
results. Many methods have been developed that differ according to the type of
data and the chosen stage for integration and span a landscape of techniques
comprising graph theory, machine learning, and statistics.

3. MULTIVIEW LEARNING IN BIOINFORMATICS
3.1 PATIENT SUBTYPING
One of the main difficulties in the treatment of complex diseasesdsuch as cancer,
neuropsychiatric diseases, and autoimmune disordersdis the consistent variability
in manifestations among affected individuals [5]. Precision medicine (or personalized
medicine) is a new discipline that has emerged in recent years, which aims to solve
this problem [6]. Its goal is individualizing the practice of medicine by taking into
account individual variability in genes, lifestyle, and environment in order to predict
disease progression and transitions between disease stages, and target the most appro-
priate medical treatments [7]. Under these premises, the task of subtyping patients
assumes a key role: in fact, once subpopulations of patients with similar characteris-
tics are identified, more accurate diagnostic and treatment strategies can be developed
for each of such groups. Moreover, the ability to refine the prognosis for a category of
patients can reduce the uncertainty about the expected outcome of a clinical treatment
on the individual (Fig. 13.4).

FIGURE 13.4
Precision medicine.