Table 4. The most contributing features belonging to at least five of the six models.
| Features | Frequency | Rank | Descriptions | |||||
|---|---|---|---|---|---|---|---|---|
| ANN | Random forest | AdaBoost | Stochastic gradient boosting | XGBoost | SVM | |||
| The ratio of family income to poverty | 6 | 1 | 1 | 2 | 1 | 1 | 2 | The ratio of family income to poverty guidelines |
| Triglycerides | 6 | 14 | 4 | 10 | 4 | 3 | 10 | Triglycerides, refrigerated serum (mmol/L) |
| White blood cell count | 6 | 13 | 7 | 13 | 11 | 5 | 13 | White blood cell count (1000 cells/uL) |
| Age | 6 | 9 | 9 | 11 | 14 | 8 | 11 | Age in years of the participant at the time of screening |
| Sleep disorder | 5 | 2 | 2 | 1 | 2 | NA | 1 | Ever told a doctor that you had trouble sleeping |
| Arthritis | 5 | 8 | 16 | 3 | 3 | NA | 3 | Ever told by a doctor that you had arthritis |
| Hemoglobin | 5 | NA | 6 | 12 | 16 | 6 | 12 | Hemoglobin (g/dL) |
| Marital status | 5 | NA | 18 | 5 | 10 | 16 | 5 | Marital status of the participants |
| Education level | 5 | 15 | 17 | 8 | 8 | NA | 8 | The highest grade or level of schooling or the highest degree |
ANN: artificial neural network; SVM: support vector machine
NA: This feature was not ranked in the top 20 features for that model.