About MLTrends
MLTrends graphs word and phrase usage trends in the Medline database of
biomedical abstracts.
Searching
Commas separate multiple search terms. For example: breast, cancer.
Quotes delimit phrases. For example: "breast cancer".
AND, OR, NOT and brackets allow Boolean search term grouping, For
example:breast AND (cancer OR tumour).
Adjacent words or phrases without explicit operators are implicitly
OR-ed together. For example:breast AND (cancer tumour).
Note that searches are case insensitive, though the Boolean operators
must be upper case to distinguish them from search terms.
Data normalization
MLTrends offers three normalization options:
- None: this graphs the count of documents matching the query per year. As the number of publications is increasing yearly, this tends to lead to an apparent upward trend for most publications.
- Publications: this graphs the documents matching the query, divided by the total number of publications per year and corrects for the increasing number of publications.
- Word count: this graphs the documents matching the query, divided by the total number of words in the searched fields of publications that year. This can be useful, as abstracts tend to be longer in recent years, and records before 1980 generally contain only titles. This normalization method avoids an artifactual jump in counts in the years where abstracts begin to appear in the Medline records.
Developed by Gareth Palidwor
and Miguel Andrade
Hosted at the Ottawa Health Research Institute
Powered by Lucene and Google Charts