BMC Medical Informatics and Decision Making

Table 4 AUC and accuracy by LSTM using all 5 CVH submetric measures using multi-label (Method 1), multiclass (Method 2), and single submetric measures for the subpopulation

From: Predicting cardiovascular health trajectories in time-series electronic health records with LSTM models

CVH	AUC			Accuracy (%)
CVH category	Method 1	Method 2	Single	Method 1	Method 2	Single
A1C ideal	0.97	0.92	0.91	99.7	90.7	90.3
A1C intermediate	0.81	0.83	0.84	94.7	78.7	79.7
A1C poor	0.90	0.92	0.93	50.7	86.2	87.4
LDL ideal	0.77	0.78	0.78	60.7	84.0	82.3
LDL intermediate	0.69	0.71	0.73	99.9	88.0	88.4
LDL poor	0.82	0.86	0.81	1.0	92.2	93.1
BMI ideal	0.97	0.97	0.97	97.6	96.6	97.0
BMI intermediate	0.96	0.95	0.94	97.6	92.0	91.8
BMI poor	0.98	0.97	0.97	69.3	94.6	94.3
BP ideal	0.76	0.77	0.77	99.5	75.9	76.1
BP intermediate	0.59	0.61	0.63	99.0	59.3	58.8
BP poor	0.78	0.77	0.79	99.9	81.2	80.2
SMK ideal	0.95	0.94	0.93	45.4	94.7	94.3
SMK intermediate	1.0	0.78	1.0	1.0	99.7	99.9
SMK poor	0.95	0.95	0.93	98.3	94.7	94.3

Back to article page

ISSN: 1472-6947

Contact us

General enquiries: journalsubmissions@springernature.com