Skip to main content

Table 2 Predictive performance of models using pre-assigned (P) or learned (L) weights in the weighted aggregation (WA) strategy

From: Learning temporal weights of clinical events using variable importance

  Accuracy   AUC   AUPRC
ADE PWA LWA   PWA LWA   PWA LWA
D611 76.46 76.68   0.882 0.882   0.956 0.954
D642 97.79 97.79   0.973 0.965   0.750 0.688
D695 76.28 76.69   0.863 0.872   0.771 0.768
E273 81.77 81.78   0.681 0.669   0.309 0.323
G620 91.64 91.64   0.809 0.803   0.316 0.326
I952 91.71 91.71   0.583 0.564   0.110 0.107
L270 73.07 72.12   0.847 0.841   0.807 0.793
L271 84.18 84.18   0.774 0.735   0.386 0.342
M804 95.17 95.17   0.652 0.615   0.098 0.075
M814 98.71 98.71   0.732 0.740   0.040 0.050
O355 72.24 71.55   0.978 0.980   0.934 0.942
R502 99.16 99.16   0.800 0.776   0.218 0.168
T782 94.81 94.81   0.712 0.700   0.156 0.152
T783 83.33 83.33   0.749 0.758   0.362 0.363
T784 66.15 66.15   0.734 0.741   0.809 0.811
T801 90.35 90.35   0.891 0.892   0.493 0.458
T808 77.93 77.72   0.882 0.879   0.958 0.958
T886 94.95 94.95   0.689 0.692   0.117 0.109
T887 61.35 60.97   0.751 0.752   0.776 0.773
Average 84.58 84.50   0.788 0.782   0.493 0.482
p-value 0.2936   0.1564   0.08743
  1. Bold indicates winning