Research Article

Sequential Pattern Mining to Predict Medical In-Hospital Mortality from Administrative Data: Application to Acute Coronary Syndrome

Table 6

Internal validation: AURC, error rate, numbers of predicted, and observed deaths by context according to the type of trajectory.

ContextDRG trajectoriesICD-10 code trajectories
AURCError rateObservedPredictedAURCError rateObservedPredicted

Man and >65 years and ≤5 stays0.930.111612.60.980.081614
Woman and 45–65 years0.960.051514.80.990.071513.4
>65 years and ≤5 stays0.910.091614.60.850.141613.4
Man and 45–65 years and ≤5 stays0.990.041615.40.990.051615.6
45–65 years and ≤5 stays0.990.031615.20.980.051615.6
Woman and >65 years and >5 stays0.870.222016.80.930.22018.4
Woman and >65 years0.920.162622.60.870.182624.6
Man and ≤5 stays0.970.0714140.970.021413.8
≤5 stays0.970.0320190.990.042019
Woman and >5 stays0.930.192416.80.960.202420
Woman0.930.133024.20.940.143026.8
Man and 45–65 years and >5 stays0.890.272220.40.860.272217.2
Man and >65 years and >5 stays0.820.2946320.840.284634.2
45–65 years and >5 stays0.860.272419.80.780.392416.2
Man and 45–65 years0.930.071514.20.780.352620.4
Man and >65 years0.840.255642.40.800.285644.2
45–65 years0.820.2630220.820.293018.6
>65 years and >5 stays0.710.326631.60.710.336663.6
>65 years0.760.268256.20.740.278248.4
Man and >5 stays0.790.317056.40.740.337059.2
Man0.780.308494.60.920.238479.4
>5 stays0.820.2892880.820.269275.2
General0.810.2511469.40.810.2511488.6

Best results are in bold.