# Table 1 Statistical criteria and diagnostic accuracy of the fitted models

Variancea G Modelb Log- likelihood BIC AIC ICL Sensitivity (%) Specificity (%) PPV (%) NPV (%) Classification rate (%)c Group sizesd
σ2 4 A −4861 9836 9751 9656 36.1 96.8 65.9 89.8 87.8 110-460-398-85
σ2 4 A × t −4932 9997 9892 9794 43.9 94.9 59.7 90.7 87.4 112-439-388-114
σ2 3 A −5105 10300 10233 10174 55.5 91.3 52.4 92.2 86.0 256-633-164
σ2 3 A × t −5117 10324 10257 10196 56.8 90.3 50.3 92.4 85.4 245-633-175
σ2 2 A −5517 11098 11050 11026 76.8 73.7 33.5 94.8 74.2 698-355
σ2 2 A × t −5509 11084 11035 11011 74.2 77.7 36.5 94.6 77.2 738-315
σg 2 4 A −4552 9241 9138 8886 65.2 86.5 45.5 93.5 83.4 153-324-354-222
σg 2 4 A × t −4699 9535 9432 9236 61.3 79.0 33.5 92.2 76.4 329-379-61-329
σg 2 3 A −4924 9954 9875 9749 74.2 79.1 38.0 94.7 78.3 292-459-303
σg 2 3 A × t −4923 9951 9872 9748 71.6 80.4 38.7 94.3 79.1 298-468-287
σg 2 2 A −5519 11110 11055 11032 82.6 64.7 28.8 95.6 67.3 608-445
σg 2 2 A × t −5511 11095 11040 11020 76.8 73.9 33.7 94.9 74.4 700-353
σ2 4 Non-biological −4830 9823 9701 9632 41.9 95.9 63.7 90.5 87.9 111-445-395-102
1. a σ for models with constant residual variances and σ g for models with different residual variance between groups.bType of residual hCG production: constant (r(t) = A), proportional to time (r(t) = A × t), and non-biological model. cPercentage of well-classified subjects. dNumber of subjects per group from the lowest to the highest hCG trajectory group