Skip to main content
. 2020 Nov 9;8(11):e22689. doi: 10.2196/22689

Table 5.

Our final Kaiser Permanente Southern California model in comparison with several previous models for forecasting hospitalizations and emergency department visits in patients with asthma.

Model Prediction target Number of features the model used Number of data instances Classification algorithm The undesirable outcome’s prevalence rate in the whole data set (%) AUCa Sensitivity (%) Specificity (%) PPVb (%) NPVc (%)
Our final KPSCd model Asthma-related hospital encounters 221 987,506 XGBooste 23,278 (2.36) 0.820 2259 (51.90) 182,176 (90.91) 2259 (11.03) 182,176 (98.86)
Our Intermountain Healthcare model [23] Asthma-related hospital encounters 142 334,564 XGBoost 12,144 (3.63) 0.859 436 (53.69) 16,955 (91.93) 436 (22.65) 16,955 (97.83)
Miller et al [15] Asthma-related hospital encounters 17 2821 Logistic regression 8.5 0.81 f
Loymans et al [10] Asthma exacerbation 7 611 Logistic regression 13 0.8
Lieu et al [3] Asthma-related hospitalization 7 16,520 Proportional hazards regression 1.8 0.79
Schatz et al [11] Asthma-related hospitalization in children 5 4197 Logistic regression 1.4 0.781 43.9 89.8 5.6 99.1
Yurk et al [17] Lost day or asthma-related hospital encounters 11 4888 Logistic regression 54 0.78 77 63 82 56
Eisner et al [12] Asthma-related EDg visit 3 2415 Logistic regression 18.3 0.751
Forno et al [22] Severe asthma exacerbation 17 615 Scoring 69.6 0.75
Schatz et al [11] Asthma-related hospitalization in adults 3 6904 Logistic regression 1.2 0.712 44.9 87.0 3.9 99.3
Lieu et al [3] Asthma-related ED visit 7 16,520 Proportional hazards regression 6.4 0.69
Eisner et al [12] Asthma-related hospitalization 1 2858 Logistic regression 32.8 0.689
Sato et al [13] Severe asthma exacerbation 3 78 Classification and regression tree 21 0.625
Schatz et al [20] Asthma-related hospital encounters 4 14,893 Logistic regression 6.5 0.614 25.4 92.0 22.0 93.2
Lieu et al [19] Asthma-related hospital encounters 4 7141 Classification and regression tree 6.9 49.0 83.6 18.5

aAUC: area under the receiver operating characteristic curve.

bPPV: positive predictive value.

cNPV: negative predictive value.

dKPSC: Kaiser Permanente Southern California.

eXGBoost: extreme gradient boosting.

fThe original paper presenting the model did not report the performance measure.

gED: emergency department.