Fig 1.
Open (left) and closed (right) discovery process defined by Weeber at al. [16].
Fig 2.
Extension of Swanson’s ABC model.
Table 1.
Resources and statistics of the dictionaries for named entity recognition.
Fig 3.
Overview of our proposed approach.
Fig 4.
Visualization of a portion of the directed network generated by literature mining.
Table 2.
Top ranked candidates of the developed hypothesis.
Table 3.
Top ranked candidates with multiple B terms of metabolites.
Fig 5.
Statistical relations of lactosylceramide, nitric oxide, malondialdehyde, and ba-PWV.
Relationship of lactosylceramide (d18:1/12:0), nitric oxide, malondialdehyde, and ba-PWV in male subjects under 50 yrs.*Tested by log-transformed. Tested by Pearson correlation (r0: smoker, r1: non-smoker, r2: total). (A) r0 = -0.739, P0<0.001; r1 = -0.388, P1 = 0.061; r2 = -0.551, P2<0.001. (B) r0 = -0.751, P0<0.001; r1 = -0.400, P1 = 0.053; r2 = -0.612, P2<0.001. (C) r0 = 0.526, P0 = 0.012; r1 = 0.628, P1 = 0.001; r2 = 0.570, P2<0.001. (D) r0 = 0.527, P0 = 0.012; r1 = 0.414, P1 = 0.044; r2 = 0.470, P2 = 0.001.
Fig 6.
Overall view of nitric oxide, malondialdehyde, and ba-PWV with lactosylceramade.
Table 4.
Clinical and biochemical characteristics in male subjects under 50 yrs.
Fig 7.
Network relations of lactosylceramade, nitric oxide, malondialdehyde, and arterial stiffness.
Fig 8.
Scatterplot of database-based versus semantic relatedness score (both normalized).