Table 4.

Grading System

Grade	Model	Rules	Attribute
A	100% accordance Example: If rs5 = BB and rs10 = AA then Case; If rs15 = AA and AUC >105 then Case; Else Control	100% accordance Example: If rs5 = BB and rs10 = AA then Case	All attributes were present (i.e. detected) and ranked as most frequent, i.e. top 2 for model A and top 4 for model B. Example: rs5 → Nr. 1; rs10 → Nr. 3; rs15 → Nr.2; AUC → Nr. 4
B	One attribute was missing or an additional attribute was identified by the generated model Example: If rs5 = BB and rs10 = AA and rs234 = BB then Case; If rs15 = AA and AUC >105 then Case; Else Control	One attribute was missing or an additional attribute was identified by the generated model Example: If rs5 = BB and rs10 = AA and rs234 = BB then Case	All attributes were present but not ranked as most frequent Example: rs5 → Nr. 5; rs10 → Nr. 3; rs15 → Nr.7; AUC → Nr. 4
C	Two attributes were different between the generated and the true model, Example: If rs5 = BB and rs10 = AA and rs234 = BB then Case; If rs15 = AA and AUC >105 and rs56 = AA then Case; Else Control	Two attributes were different between the generated and the true model Example: If rs5 = BB and rs234 = BB then Case	One attribute was not present, remaining attributes were present and rank was not considered Example: rs5 → not detected; rs10, rs15 and AUC → detected
D	Three attributes were different between the generated and the true model Example: If rs5 = BB and rs234 = BB and rs56 = AA then Case; If rs15 = AA and AUC >105 then Case; Else Control	Three attributes were different between the generated and the true model Example: If rs5 = BB and rs56 = AA and rs234 = BB then Case	Two attributes were not present, remaining attributes were present and ranked as most frequent Example: rs5 and AUC → not detected; rs10 and rs15 → detected

Comparison of the generated versus true model, rules and attributes. (Model B was used for the example.)