Exploring and sampling the raw data - quintelligence-health/medline-dashboard GitHub Wiki

Use the search box and the filters to search the MEDLINE dataset with exactly what you are looking for. The filters permit the user to search for all the abstracts that are hand annotated with a particular MeSH Heading. To do this use the drop down filter (in the image) and select “MeshHeadingList.desc” + “is” + “mental health”, knowing that “mental health” is the exact description of the related MeSH HEading (see in MeSH tree browser: https://meshb.nlm.nih.gov/record/ui?ui=D008603) This can also be done by using directly the Lucene language in the searchbox above with the input: “MeshHeadingList.desc:mental health”. The difference between using the keywords and using the filter, much as in PubMed, is that keywords will subset MEDLINE by all the articles that mention them, sometimes not exactly matching the articles discussing in full the topic assign to those keywords used. We can also determine the time window for the chosen set, which is sometimes relevant when, e.g., we want to exclude information from a time period where the knowledge of a certain disease was confusing or even incorrect.

Use the search box and the filters to search the MEDLINE dataset with exactly what you are looking for. The filters permit the user to search for all the abstracts that are hand annotated with a particular MeSH Heading. To do this use the drop down filter (in the image) and select “MeshHeadingList.desc” + “is” + “mental health”, knowing that “mental health” is the exact description of the related MeSH HEading (see in MeSH tree browser: https://meshb.nlm.nih.gov/record/ui?ui=D008603) This can also be done by using directly the Lucene Query Syntax in the searchbox above, allowing for powerful queries when searching for the topic of interest, including a range of operators. Examples of these are:

MeshHeadingList.desc:"public health"

to sample all the articles that include the MeSH heading descriptor "public health"

"public+health" AND NOT disease

to sample all the articles with the key phrase "public+health" not including the key word "disease". The difference between using the keywords and using the filter, much as in PubMed, is that keywords will subset MEDLINE by all the articles that mention them, sometimes not exactly matching the articles discussing in full the topic assign to those keywords used. We can also determine the time window for the chosen set, which is sometimes relevant when, e.g., we want to exclude information from a time period where the knowledge of a certain disease was confusing or even incorrect.

The subset then contains only the articles annotated with the ”mental health” MeSH Heading. This is reflected in the bar chart of articles per year, and in the articles described in the list below the chart. Opening each one of them in the “+” sign one can see in detail all the meta information collected on each of the articles.

After the user being satisfied with the search results, (s)he can save the search using the option above in order to build visualisation modules over that search in particular.

Back to the MEDLINE exploratory dashboard