IEEE Computational Intelligence Magazine - November 2020 - 57

distinguished from the most general to the
It is used to annotate scientific articles, news articles
most specific in up to 13 hierarchical depth
levels. Although the recent introduction of
and reports relating to the COVID-19, allowing for the
the supplementary concept "COVID-19" on
utilization of the MeSH headings as search topics ...
January 13, 2020, the MeSH heading "Coronavirus" was introduced in 1994 referring to
The MIDAS platform also includes an exploratory dashthe group of related viruses and prior known strains. These
board that provides access to all MEDLINE records and
are located in the MeSH tree under the Coronaviridae family,
enables users to explore these directly, and save samples based
introduced in 1999. MEDLINE includes 5976 research artion queries. These are stored as JSON files in an elasticsearch
cles on coronavirus, relating to 4097 other health topics and
based database, utilizing robust and well-established technology
1706 substances. The articles that are hand-annotated with
[11]. The external MIDAS MEDLINE explorer does not
the MeSH class "Coronavirus" can help researchers better
require the user to have expert technical knowledge and allows
understand the new strain from the available scientific literathe average biomedical researcher to explore MEDLINE with
ture. With this in mind, the MIDAS platform offers an
further insight but little technical skills required. It also enables
exploratory tool (see Figure 5) that allows the user to explore
the user to rapidly build several data visualization modules
the published research through: (i) a query, based on keywords
(based on the Kibana open-source data visualization technoloand operators, or an advanced query based on the syntax of
gy) that are easily configurable and based on templates, over the
the Lucene language; and (ii) a target pointer, which the user
queried data (saved as a subset of records). It includes a variety
interacts by dragging it over the tag cloud to explore the
of charts, tag clouds, heat maps and lists, as well as dashboards
results on the subtopics it relates to and to reprioritize the
that integrate the created dynamic data visualization modules
search results obtained. An example of such a precise syntax
(see Figure 6). These enable MIDAS users to build and share
query (including two types of search categories-MeSH
dashboards that analyze published biomedical research on
headings as health topics, or Chemicals)-MeshHeadingList.
COVID-19 in relation to relevant topics for this study (similardesc:" Coronavirus" NOT ChemicalList.NameOfSubstance: "Viral
ly to diabetes, mental health etc.), and recover the biomedical
Proteins"-provides the user with, e.g. a subset of MEDLINE
articles relating to it. Moreover, this dashboard is served with a
articles that are hand-annotated with the MeSH heading
powerful API that allows the user to access and query the data
"Coronavirus" but are not labelled with the substance "Viral
from other systems. An extended instance of it enables easy
Proteins." The user can further explore the subset restricting it
to labelled publications with the "Biomarkers" MeSH heading
ingestion of new articles, reports or news that can be annotat--
to explore new treatments in this specific context.
ed using a classifier, loaded, visualized and explored in the

FIGURE 5 Exploring the biomarkers related to the Coronavirus in the published research using the MIDAS MEDLINE Explorer.

NOVEMBER 2020 | IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE

https://www.ncbi.nlm.nih.gov/pubmed/14623149 https://www.ncbi.nlm.nih.gov/pubmed/25819115 https://www.ncbi.nlm.nih.gov/pubmed/11578135 https://www.ncbi.nlm.nih.gov/pubmed/20942081

IEEE Computational Intelligence Magazine - November 2020

Table of Contents for the Digital Edition of IEEE Computational Intelligence Magazine - November 2020