Characteristics of AMI patients
After screening of inclusion and exclusion criteria, the final sample for the study was 4610 patients, of which 894 had died within 90 days (Table 1) . The only index that differed significantly between the two groups was AG. Among the 1383 patients in the modeling group, 76 (5.5%) had a previous history of myocardial infarction and 488 (35.3%) had undergone PCI.
Table1
Charateristics of patients with AMI
Variables(N=4610)
|
training group(N=1383)
|
validation group(N=3227)
|
P value
|
age (years)
|
71.00 [60.00, 79.00]
|
70.00 [60.00, 79.00]
|
0.329
|
AST(IU/L)
|
43.00 [26.00, 98.00]
|
42.00 [25.00, 101.00]
|
0.731
|
troponinT(ng/mL)
|
1.10 [0.33, 3.45]
|
1.19 [0.35, 3.32]
|
0.740
|
CK(IU/L)
|
315.00 [86.00, 755.50]
|
260.00 [84.00, 770.00]
|
0.884
|
CK-MB(ng/mL)
|
9.00 [7.00, 64.50]
|
9.00 [7.00, 65.00]
|
0.377
|
INR
|
1.20 [1.10, 1.40]
|
1.20 [1.10, 1.40]
|
0.669
|
platlet(k/uL)
|
236.00 [183.00, 299.00]
|
234.00 [185.00, 295.00]
|
0.661
|
PT(s)
|
13.50 [12.70, 15.00]
|
13.50 [12.70, 14.90]
|
0.806
|
APTT(s)
|
32.30 [26.70, 54.10]
|
31.70 [26.40, 50.90]
|
0.110
|
AG(mEq/L)
|
16.00 [13.00, 18.00]
|
15.00 [13.00, 18.00]
|
0.026
|
SOFA
|
4.00 [2.00, 6.00]
|
4.00 [2.00, 6.00]
|
0.878
|
APS-III
|
40.00 [29.00, 54.00]
|
40.00 [29.00, 53.00]
|
0.646
|
sex=Male/Female
|
898/485 (64.9/35.1)
|
2081/1146 (64.5/35.5)
|
0.788
|
complications
|
|
|
|
Hypertension = 0/1
|
754/629 (54.5/45.5)
|
1701/1526 (52.7/47.3)
|
0.273
|
Obesity = 0/1
|
1329/54 (96.1/3.9)
|
3069/158 (95.1/4.9)
|
0.146
|
congestive heart failure = 0/1
|
742/641 (53.7/46.3)
|
1649/1578 (51.1/48.9)
|
0.115
|
AFI = 0/1
|
986/397 (71.3/28.7)
|
2291/936 (71.0/29.0)
|
0.859
|
AFL = 0/1
|
1359/24 (98.3/1.7)
|
3147/80 (97.5/2.5)
|
0.130
|
VFI = 0/1
|
1323/60 (95.7/4.3)
|
3079/148 (95.4/4.6)
|
0.757
|
VT = 0/1
|
1253/130 (90.6/9.4)
|
2939/288 (91.1/8.9)
|
0.615
|
AVB = 0/1
|
1332/51 (96.3/3.7)
|
3112/115 (96.4/3.6)
|
0.863
|
MI_his = 0/1
|
1307/76 (94.5/5.5)
|
3059/168 (94.8/5.2)
|
0.720
|
IABP = 0/1
|
1348/35 (97.5/2.5)
|
3129/98 (97.0/3.0)
|
0.388
|
drug use = 0/1
|
682/701 (49.3/50.7)
|
1582/1645 (49.0/51.0)
|
0.872
|
PCI = 0/1
|
895/488 (64.7/35.3)
|
2025/1202 (62.8/37.2)
|
0.217
|
bypass = 0/1
|
1055/328 (76.3/23.7)
|
2435/792 (75.5/24.5)
|
0.574
|
LDH(IU/L)
|
|
|
0.096
|
<109
|
8 (0.6)
|
6 (0.2)
|
|
109-245
|
455 (32.9)
|
1013 (31.4)
|
|
245-1000
|
826 (59.7)
|
2014 (62.4)
|
|
1000-5000
|
86 (6.2)
|
173 (5.4)
|
|
>5000
|
8 (0.6)
|
21 (0.7)
|
|
Creatinine(mg/dL)
|
|
|
>0.999
|
0.00-0.39
|
100 (7.2)
|
211 (6.5)
|
|
0.40-0.79
|
162 (11.7)
|
378 (11.7)
|
|
0.80-1.19
|
124 (9.0)
|
250 (7.7)
|
|
1.20-1.59
|
259 (18.7)
|
587 (18.2)
|
|
1.60-1.99
|
531 (38.4)
|
1338 (41.5)
|
|
2.00-3.99
|
203 (14.7)
|
455 (14.1)
|
|
≥4.00
|
4 (0.3)
|
8 (0.2)
|
|
meanHR(min-1)
|
|
|
0.096
|
<50
|
5 (0.4)
|
10 (0.3)
|
|
50-69
|
245 (17.7)
|
601 (18.6)
|
|
70-89
|
699 (50.5)
|
1676 (51.9)
|
|
90-109
|
331 (23.9)
|
767 (23.8)
|
|
110-140
|
103 (7.4)
|
173 (5.4)
|
|
SysBP(mmHg)
|
|
|
>0.999
|
<80
|
6 (0.4)
|
22 (0.7)
|
|
80-99
|
183 (13.2)
|
375 (11.6)
|
|
100-119
|
725 (52.4)
|
1691 (52.4)
|
|
120-139
|
376 (27.2)
|
899 (27.9)
|
|
140-159
|
79 (5.7)
|
211 (6.5)
|
|
160-190
|
14 (1.0)
|
29 (0.9)
|
|
status = 0/1
|
1099/284 (79.5/20.5)
|
2617/610 (81.1/18.9)
|
0.208
|
Nonnormal continuous variables were presented as median[IQR]. Categorical variables were presented as number(precentage ). Abbreviations: AST, aspartate transaminase; CK, creatine kinase; CK-MB, creatine kinase isoenzymes; INR, international normalized ratio; PT, prothrombin time; APTT, activated partial thromboplastin time; AG, anion gap; SOFA, sequential organ failure assessment; APSIII, acute physiology score III; AFI, atrial fibrillation; AFL, atrial flutter; VFI, ventricular fibrillation; VT, ventricular tachycardia; AVB, atrioventricular block; MI_his, previous history of myocardial infarction; IABP, intra-aortic balloon pump; drug use, aspirin, clopidogrel, atorvastatin, or streptokinase use or not; PCI, percutaneous coronary intervention; bypass, heart bypass surgery; LDH, lactate dehydrogenase; meanHR, mean heart rate; SysBP, systolic blood pressure.
|
Variable selection and nomogram development
Figure 1 shows the process and results of the random forest feature selection, which identified that the following 19 risk factors affected the prognosis of AMI patients: APS-III, bypass, age, PCI, AG, PT, INR, LDH, AST, troponinT, CK, platelets, congestive heart failure, APTT, MI_his, CK_MB, AFI, VT, and VFI. Multivariate logistic regression analysis was conducted on these risk factors, and this results are listed in Table 2. We developed our new risk model by identifying eight significant risk factors using multivariate logistic regression analysis. An interactive nomogram was developed based on this model for predicting the 90-day mortality of AMI patients (Figure 2) .
Table 2
The results of multivariate logistic regression
|
OR
|
2.5%
|
97.5%
|
P value
|
age (years)
|
1.04
|
1.03
|
1.06
|
<0.001
|
APSIII
|
1.03
|
1.03
|
1.04
|
<0.001
|
bypass
|
0.18
|
0.12
|
0.25
|
<0.001
|
PCI
|
0.30
|
0.23
|
0.39
|
<0.001
|
AG(mEq/L)
|
1.01
|
0.99
|
1.04
|
0.235
|
PT(s)
|
1.02
|
1.00
|
1.05
|
0.088
|
INR
|
0.93
|
0.80
|
1.07
|
0.373
|
LDH(IU/L)
|
|
|
|
|
<109
|
Ref.
|
|
|
|
109-245
|
0.47
|
0.07
|
4.54
|
0.461
|
245-1000
|
0.78
|
0.12
|
7.47
|
0.805
|
1000-5000
|
1.38
|
0.20
|
13.69
|
0.756
|
>5000
|
2.29
|
0.26
|
27.12
|
0.475
|
troponinT(ng/mL)
|
1.07
|
1.03
|
1.11
|
<0.001
|
platlet(k/uL)
|
1.00
|
1.00
|
1.00
|
0.895
|
congestive heart failure
|
1.10
|
0.88
|
1.36
|
0.399
|
APTT(s)
|
1.00
|
1.00
|
1.00
|
0.634
|
MI_his
|
1.58
|
1.04
|
2.37
|
0.028
|
CK-MB(ng/mL)
|
1.00
|
1.00
|
1.00
|
0.174
|
AFI
|
1.03
|
0.82
|
1.30
|
0.774
|
VT
|
1.59
|
1.13
|
2.24
|
0.008
|
VFI
|
1.70
|
1.05
|
2.71
|
0.027
|
Abbreviations: APSIII, acute physiology score III; bypass, heart bypass surgery; PCI, percutaneous coronary intervention; AG, anion gap; PT, prothrombin time; INR, international normalized ratio; LDH, lactate dehydrogenase; APTT, activated partial thromboplastin time; MI_his, previous history of myocardial infarction; CK-MB, creatine kinase isoenzymes; AFI, atrial fibrillation; VT, ventricular tachycardia; VFI, ventricular fibrillation.
|
Based on the GRACE Score table, we extracted and processed all variables except ST segment reduction to produce a similar GRACE Score and this results are listed in Table 3.
Table 3
The results of similar GRACE Score by multivariate logistic regression
|
OR
|
2.5%
|
97.5%
|
P value
|
age(years)
|
|
|
|
|
<30
|
Ref.
|
|
|
|
30-39
|
0.37
|
0.03
|
9.46
|
0.468
|
40-49
|
1.23
|
0.17
|
25.79
|
0.858
|
50-59
|
1.16
|
0.16
|
23.89
|
0.896
|
60-69
|
2.23
|
0.32
|
45.48
|
0.484
|
70-79
|
3.93
|
0.57
|
79.93
|
0.233
|
80-89
|
6.07
|
0.88
|
123.39
|
0.117
|
meanHR(min-1)
|
|
|
|
|
<50
|
Ref.
|
|
|
|
50-69
|
0.66
|
0.13
|
5.11
|
0.642
|
70-89
|
0.66
|
0.13
|
5.10
|
0.645
|
90-109
|
1.10
|
0.22
|
8.46
|
0.918
|
110-140
|
1.19
|
0.23
|
9.36
|
0.849
|
SysBP(mmHg)
|
|
|
|
|
<80
|
Ref.
|
|
|
|
80-99
|
0.22
|
0.06
|
0.67
|
0.013
|
100-119
|
0.08
|
0.02
|
0.24
|
<0.001
|
120-139
|
0.07
|
0.02
|
0.22
|
<0.001
|
140-159
|
0.10
|
0.03
|
0.32
|
<0.001
|
160-190
|
0.07
|
0.01
|
0.30
|
<0.001
|
Creatinine(mg/dL)
|
|
|
|
|
0.00-0.39
|
Ref.
|
|
|
|
0.40-0.79
|
0.68
|
0.46
|
1.00
|
0.052
|
0.80-1.19
|
0.59
|
0.38
|
0.92
|
0.019
|
1.20-1.59
|
0.38
|
0.26
|
0.56
|
<0.001
|
1.60-1.99
|
0.25
|
0.17
|
0.37
|
<0.001
|
2.00-3.99
|
0.30
|
0.19
|
0.47
|
<0.001
|
≥4.00
|
NA
|
NA
|
NA
|
0.959
|
congestive heart failure
|
1.22
|
0.99
|
1.50
|
0.060
|
PCI
|
0.38
|
0.29
|
0.49
|
<0.001
|
MI_his
|
1.70
|
1.14
|
2.50
|
0.008
|
AST(IU/L)
|
1.00
|
1.00
|
1.00
|
0.484
|
LDH(IU/L)
|
1.00
|
1.00
|
1.00
|
<0.001
|
CK(IU/L)
|
1.00
|
1.00
|
1.00
|
0.660
|
CK-MB(ng/mL)
|
1.00
|
1.00
|
1.00
|
0.032
|
troponinT(ng/mL)
|
1.08
|
1.05
|
1.12
|
<0.001
|
Abbreviations: meanHR, mean heart rate; SysBP, systolic blood pressure; PCI, percutaneous coronary intervention; MI_his, previous history of myocardial infarction; AST, aspartate transaminase; LDH, lactate dehydrogenase; CK, creatine kinase; CK-MB, creatine kinase isoenzymes.
|
Nomogram performance
AUCs were used to evaluate the differences between our risk model and similar GRACE Score, APS-III, and SOFA as shown in Figure 3. Our risk model had the highest AUC values (0.826 and 0.818 in the training and validation groups, respectively) .
Pair-wise AUC comparisons of the different models revealed all P values to be less than 0.001, indicating that there were statistically significant differences in the predictive ability between the different models.
Nomogram calibration
The Hosmer-Lemeshow goodness-of-fit test revealed a high consistency between the prediction and observation probabilities of the training (chi-square=16.91, P=0.051) and validation groups (chi-square=11.72, P=0.230) . The calibration diagram also revealed good consistency between the predicted and observed results of the training and validation groups (Figure 4) .
The NRI values for the risk model in the training group were 0.29 (95% CI=0.13–0.37) , 0.70 (95% CI=0.59–0.82) , and 0.59 (95% CI=0.50–0.69) versus the similar GRACE Score, SOFA, and APS-III models, respectively; the corresponding values in the validation group were 0.32 (95% CI=0.05–0.41) , 0.51 (95% CI=0.34–0.73) , and 0.45 (95% CI= 0.31–0.61) .
Similarly, the IDI values for the risk model in the training group were 0.05 (95% CI=0.04–0.07, P<0.001) , 0.15 (95% CI=0.13–0.16, P<0.001) , and 0.10 (95% CI=0.09–0.11, P<0.001) versus the GRACE, SOFA, and APS-III models, respectively; the corresponding values in the validation group were 0.06 (95% CI=0.03–0.09, P<0.001) , 0.12 (95% CI=0.09–0.15, P<0.001) , and 0.07 (95% CI=0.05–0.08, P<0.001) . These results indicate that the nomogram developed in this study exhibits improved prediction ability compared with the other models.
Clinical application
The DCA plots indicated the high clinical net benefit of our risk model and demonstrated its clinical applicability and impact on clinical decision-making (Figure 5) .