TABLE 2.
Food groups important for the modeling of LSI in the PLS regression analysis (n = 354)1
| Food groups | Regression coefficient for standardized food item in the PLS regression model | Variable importance in the projection statistic | Variation in food groups explained by exploratory dietary pattern score |
| % | |||
| Inverse association with LSI | |||
| Fruit and herbal tea | −0.09 | 1.89 | 7.05 |
| Green and black tea | −0.08 | 1.77 | 7.82 |
| Sugar and confectionary | −0.08 | 1.74 | 18.41 |
| Other fats | −0.07 | 1.46 | 1.59 |
| Bread | −0.07 | 1.36 | 10.15 |
| Breakfast cereals | −0.06 | 1.17 | 6.27 |
| Cheese | −0.05 | 1.04 | 9.35 |
| Positive association with LSI | |||
| Soups | 0.12 | 2.47 | 23.98 |
| Beer | 0.09 | 1.96 | 15.27 |
| Wine | 0.08 | 1.59 | 9.70 |
| Poultry | 0.07 | 1.40 | 8.41 |
| Juices | 0.07 | 1.38 | 3.64 |
| Eggs | 0.06 | 1.34 | 7.97 |
1
The variable importance in the projection statistic (40) was used to assess the importance of a food group to the PLS model. Food groups with a variable importance in the projection statistic ≤1.0 in the PLS model were excluded from the table. LSI, liver signal intensity; PLS, partial least squares.