Stratified by centre
Overall a
n 562 141
PAPS_11nov (median [IQR]) 36.0 [30.0, 42.0] 36.0 [30.0, 41.0]
gender_1_men = 1 (%) 233 (41.5) 67 (47.5)
ageatprocedure (median [IQR]) 83.0 [80.0, 87.0] 83.0 [80.0, 86.0]
NHYA_baseline (%)
1 14 ( 2.5) 13 ( 9.2)
2 265 (47.2) 57 (40.4)
3 270 (48.0) 63 (44.7)
4 13 ( 2.3) 8 ( 5.7)
Test_cognitivo (%)
0 145 (25.8) 39 (27.7)
1 112 (19.9) 48 (34.0)
2 131 (23.3) 40 (28.4)
3 62 (11.0) 5 ( 3.5)
4 40 ( 7.1) 6 ( 4.3)
5 39 ( 6.9) 2 ( 1.4)
6 31 ( 5.5) 0 ( 0.0)
7 2 ( 0.4) 1 ( 0.7)
BADL (%)
1 36 ( 6.4) 4 ( 2.8)
2 5 ( 0.9) 3 ( 2.1)
3 24 ( 4.3) 2 ( 1.4)
4 68 (12.1) 6 ( 4.3)
5 144 (25.7) 33 (23.4)
6 283 (50.5) 93 (66.0)
IADL (median [IQR]) 6.0 [5.0, 8.0] 7.0 [5.0, 8.0]
MNA_sh (median [IQR]) 11.0 [10.0, 12.0] 11.0 [10.0, 12.0]
LVEF (median [IQR]) 56.0 [48.0, 62.0] 59.0 [54.0, 64.0]
Grad_picco (median [IQR]) 73.0 [58.7, 88.0] 76.0 [67.0, 87.0]
Grad_medio (median [IQR]) 45.0 [33.0, 59.0] 47.0 [39.0, 55.0]
AVAplan (median [IQR]) 0.5 [0.4, 0.6] 0.4 [0.4, 0.5]
Crea_pre_op_feb2024 (median [IQR]) 0.9 [0.7, 1.2] 0.8 [0.7, 1.0]
CKDEPI_Syn (median [IQR]) 65.5 [47.8, 79.8] 75.6 [60.6, 83.8]
RHYTHM_1FA2PM3RS (%)
1 74 (13.2) 34 (24.1)
2 31 ( 5.5) 12 ( 8.5)
3 457 (81.3) 95 (67.4)
RHYTHM_FA = 1 (%) 105 (18.7) 46 (32.6)
BAV_1 (%)
0 56 (10.0) 0 ( 0.0)
1 477 (84.9) 126 (89.4)
2 29 ( 5.2) 15 (10.6)
LBBB_10 (%)
0 60 (10.7) 0 ( 0.0)
1 488 (86.8) 127 (90.1)
2 14 ( 2.5) 14 ( 9.9)
Composite_rechek = 1 (%) 78 (13.9) 10 ( 7.1)
Decesso_10 = 1 (%) 75 (13.3) 10 ( 7.1)
Stratified by centre
b c
n 37 384
PAPS_11nov (median [IQR]) 36.0 [30.0, 41.0] 36.0 [30.0, 42.0]
gender_1_men = 1 (%) 14 (37.8) 152 (39.6)
ageatprocedure (median [IQR]) 84.0 [82.0, 87.0] 83.0 [80.0, 87.0]
NHYA_baseline (%)
1 0 ( 0.0) 1 ( 0.3)
2 21 (56.8) 187 (48.7)
3 14 (37.8) 193 (50.3)
4 2 ( 5.4) 3 ( 0.8)
Test_cognitivo (%)
0 31 (83.8) 75 (19.5)
1 6 (16.2) 58 (15.1)
2 0 ( 0.0) 91 (23.7)
3 0 ( 0.0) 57 (14.8)
4 0 ( 0.0) 34 ( 8.9)
5 0 ( 0.0) 37 ( 9.6)
6 0 ( 0.0) 31 ( 8.1)
7 0 ( 0.0) 1 ( 0.3)
BADL (%)
1 1 ( 2.9) 31 ( 8.1)
2 0 ( 0.0) 2 ( 0.5)
3 1 ( 2.9) 21 ( 5.5)
4 2 ( 5.7) 60 (15.6)
5 15 (42.9) 96 (25.0)
6 16 (45.7) 174 (45.3)
IADL (median [IQR]) NA [NA, NA] 6.0 [5.0, 7.0]
MNA_sh (median [IQR]) 12.0 [11.0, 12.0] 11.0 [10.0, 13.0]
LVEF (median [IQR]) 60.0 [55.0, 62.0] 55.0 [46.0, 60.0]
Grad_picco (median [IQR]) 83.5 [70.0, 95.5] 71.0 [56.0, 88.0]
Grad_medio (median [IQR]) 52.0 [45.0, 62.0] 44.0 [29.0, 61.0]
AVAplan (median [IQR]) NA [NA, NA] 0.5 [0.4, 0.6]
Crea_pre_op_feb2024 (median [IQR]) NA [NA, NA] 0.9 [0.8, 1.3]
CKDEPI_Syn (median [IQR]) 75.0 [55.0, 88.0] 61.0 [43.2, 77.6]
RHYTHM_1FA2PM3RS (%)
1 4 (10.8) 36 ( 9.4)
2 1 ( 2.7) 18 ( 4.7)
3 32 (86.5) 330 (85.9)
RHYTHM_FA = 1 (%) 5 (13.5) 54 (14.1)
BAV_1 (%)
0 30 (81.1) 26 ( 6.8)
1 7 (18.9) 344 (89.6)
2 0 ( 0.0) 14 ( 3.6)
LBBB_10 (%)
0 33 (89.2) 27 ( 7.0)
1 4 (10.8) 357 (93.0)
2 0 ( 0.0) 0 ( 0.0)
Composite_rechek = 1 (%) 6 (16.2) 62 (16.1)
Decesso_10 = 1 (%) 6 (16.2) 59 (15.4) EDA TAVI
Caratteristiche del campione, divise per setting. Ho diviso basandomi sul ID (non so se ha senso): A = PD, B = FIN, C = niente
Uso A+B come derivation cohort. C sarà la mia validation cohort. Su questa faccio factor analysis e alleno un modello di regressione logistica penalizzato (LASSO), un modello di ensamble (Gradient boosting) simile a random forest ottimizzato, un support vector machine e una regressione logistica fittata secondo le mie intuizioni (mMod - my model).
Decido di utilizzare 3 fattori (elbow method sul grafico sotto). Identifico i pesi delle variabili in 3 fattori (tabella)e costruisco gli score
Loadings:
ML1 ML2 ML3
PAPS_11nov -0.393
gender_1_men 0.354 0.447
ageatprocedure
NHYA_baseline
BADL 0.575
MNA_sh 0.645
LVEF -0.560
CKDEPI_Syn
RHYTHM_FA
BAV_1 1.005
LBBB_10 0.569 0.301
ML1 ML2 ML3
SS loadings 1.373 1.109 0.675
Proportion Var 0.125 0.101 0.061
Cumulative Var 0.125 0.226 0.287Warning in ci.auc.roc(roc, ...): ci.auc() of a ROC curve with AUC == 1 is
always 1-1 and can be misleading.
| model_name | aucTRAIN | X95CI_LB | X95CI_UB | |
|---|---|---|---|---|
| XGB | XGB | 1.0000000 | 1.0000000 | 1.0000000 |
| SVM | SVM | 0.9722222 | 0.9073854 | 1.0000000 |
| LASSO | LASSO | 0.9043210 | 0.8333223 | 0.9620811 |
| ML2 | ML2 | 0.8417108 | 0.7407297 | 0.9224096 |
| myMod | myMod | 0.7843915 | 0.6551808 | 0.9023589 |
| ML3 | ML3 | 0.5515873 | 0.3776235 | 0.7096561 |
| ML1 | ML1 | 0.4129189 | 0.3116953 | 0.5092593 |
Valuto la capacità discriminativa - nel test - dei vari score
| model_name | aucTEST | X95CI_LB | X95CI_UB | |
|---|---|---|---|---|
| LASSO | LASSO | 0.8988179 | 0.8519297 | 0.9403013 |
| ML2 | ML2 | 0.8700661 | 0.8101564 | 0.9245173 |
| XGB | XGB | 0.8199760 | 0.7668766 | 0.8692772 |
| SVM | SVM | 0.8194751 | 0.7712282 | 0.8649081 |
| myMod | myMod | 0.7783510 | 0.7161384 | 0.8320032 |
| ML3 | ML3 | 0.6073432 | 0.5300967 | 0.6772315 |
| ML1 | ML1 | 0.4578241 | 0.4106291 | 0.5016035 |