Table 4

Number of patients identified as cases (both true and false positives) and the number of false positives called up for the detection of a single case if 10% of the cases are to be found in the 1:45 ratio set (true incidence)

Logistic regressionRandom forestsXGBoost
Demographic dataset2577 (14.16)3377 (18.86)2568 (14.11)
Non-sequential dataset1288 (6.58)1451 (7.54)1386 (7.15)
Extended non-sequential dataset1196 (6.04)1387 (7.16)1266 (6.45)
Sequential dataset1572 (8.25)1519 (7.94)1766 (9.39)
Extended sequential dataset1269 (6.46)1450 (7.53)1591 (8.36)
Complete dataset1189 (5.99)1277 (6.51)1406 (7.27)
  • The best performance is found for the complete dataset with logistic regression.