Table 2.
Overview of top predictive topics using the 2-LDA. The left column describes the stemmed words (in Dutch), followed by an expert interpretation of the terms. The rightmost columns show the weight the topics obtained in the final logistic regression equations that resulted from that specific setting (Table 1 fourth row, third column for the Weight AG/T column and the twelfth row, third column for the Weight AG/C/M/R/L/T column). Note that merely one topic is present in the final equation with the coded data
| Key topic terms, stemmed (top 2) |
Expert description |
Weight AG/T | Weight AG/C/M/R/L/T |
|---|---|---|---|
| coloncarcinom (0.22), malais (0.12) |
Suspicion of CRC |
0.067 | 0.337 |
| algemen (0.50), journal (0.49), brak (0.002) |
General journal text |
0.058 | - |
| anemie (0.16), chronisch (0.12) |
Anemia | 0.051 | - |
| met (0.18), rectumcarcinom (0.07) |
Suspicion of CRC/cancer |
0.048 | - |
| malign (0.10), duizel (0.10) |
Suspicion of cancer |
0.043 | - |