- General acceptance scores published for more than 5,000 uncertified bodily injury hypotheses
- Literature explorer feature now available on the chemical profile page
- Two additions to the chemical inventory, five newly certified bodily injury hypotheses involving chemical hazards, and three newly certified bodily injury hypotheses involving pharmaceutical hazards
General acceptance scores for uncertified hypotheses
We introduced the concept of hypothesis "certification" when we first released algorithmic literature scoring in 2018. ChemMeta categorizes a hypothesis as certified if a human analyst determines that there are at least two in vitro studies or one human or animal study in the peer-reviewed scientific literature that tests the hypothesis. General acceptance scores are computed for all certified hypotheses.
ChemMeta now reports general acceptance scores for a subset of "uncertified" hypotheses as well. Uncertified hypotheses are hypotheses that ChemMeta's algorithmic scoring method indicates may have been tested in the peer-reviewed literature, but have yet to be "certified" by a human analyst.
An uncertified hypothesis is included in ChemMeta if at least ten percent of the articles that the algorithm indicates could be investigating the hypothesis have a relevance score of 75 or above. The relevance score, which is reported in the literature table for every article, indicates the degree of confidence the algorithm has that the article is in fact investigating the hypothesis. Articles that have a relevance score of less than 50 are excluded from ChemMeta (a hypothesis or chemical with no relevant articles by this measure is categorized as having "no relevant literature").
As of this release, ChemMeta reports general acceptance risks scores for 4,551 certified hypotheses and 5,254 uncertified hypotheses.
Users can choose to display only certified hypotheses or both certified and uncertified hypotheses by selecting or deselecting the "include uncertified hypotheses" checkbox in the upper-right corner of every page. Uncertified hypotheses are displayed in italics to differentiate them from certified hypotheses.
A new literature explorer modal allows users to view summary statistics for the literature contributing to the general acceptance risk score for a given hypothesis. The literature explorer modal can be accessed from the general acceptance risk score card on the chemical profile page by clicking on the general acceptance risk score for a hypothesis of interest.
The literature explorer modal reports article counts by study type (human, animal, in vitro) and whether the article's metadata was extracted algorithmically or by a human analyst. Clicking on any count in this table filters the literature table below accordingly. Users can also see a list of "top" authors and journals, where "top" is determined by an author's or journal's influence-weighted sum of articles in the literature.
- Literature card. The literature card on the chemical profile page has been revised to be in line with the literature explorer modal described in the section above.
- Sorting. Clicking on column headers for numerical columns sorts tables in descending order.
- Table pagination. The default number of records per table is now set to 100 with options to view 50, 200, or 500 records.