Skip to main content
. 2022 Feb 15;9:69–81. doi: 10.2147/JHC.S341045

Figure 2.

Figure 2

Area under the receiver operating characteristic curve (AUC-ROC) for prediction of hepatocellular carcinoma risk among participants of the NIH-AARP Diet and Health Study prospective cohort. A total of 377,065 participants (cases: n=647, controls, n=376,418) were included in the analysis, among whom 226,239 were used for training (cases: 388, controls: 225,851) and 150,826 participants were used for independent validation (cases: n=259, control: n=150,567). (A) (Elaborate Base Model) was based on 14 variables: age (continuous), sex, BMI (continuous), height (continuous) diabetes (yes, no), general health status (excellent/very good, good, fair/poor), alcohol (continuous), moderate-to-vigorous physical activity (< 3 times/week, ≥3 times/week), dietary cholesterol (continuous), saturated fat (continuous), trans-fatty acids (continuous), ounce equivalents of lean meat from eggs per day (continuous), dietary vitamin B6 (mg/d, continuous), and healthy eating index scores (continuous). (B) (Parsimonious Base Model) was based on 9 of the above variables: age, sex, BMI, diabetes, general health status, alcohol, moderate-to-vigorous physical activity, ounce equivalents of lean meat from eggs per day, and healthy eating index scores.