Table 9 Statistics of covariates, from the external cohort samples, categorized by GBDT as Non-ACS, Uncertain and ACS. The percentage is calculated w.r.t the number of samples in the category.

From: Selective classification with machine learning uncertainty estimates improves ACS prediction: a retrospective study in the prehospital setting

Characteristic

Type

Non-ACS (n=784)

Uncertain (n=193)

ACS (n=150)

Age

Numerical

\(58(\pm 32)\)

\(64(\pm 25)\)

\(64(\pm 26)\)

Gender(male)

Binary

421(53%)

114(59%)

94(62%)

Medical history

Hypercholesterolemia

Binary

253(32%)

109(56%)

123(82%)

Hypertension

Binary

504(64%)

157(81%)

142(94%)

Current Smoker

Binary

198(25%)

42(21%)

43(28%)

Diabetes

Binary

204(26%)

78(40%)

72(48%)

Prior MI

Binary

0(0%)

100(51%)

145(96%)

Angina

Binary

1(<1%)

24(12%)

55(36%)

Prior CABG

Binary

29(3%)

70(36%)

81(54%)

Prior PCI

Binary

1(<1%)

4(2%)

1(<1%)

CAD

Binary

47(5%)

107(55%)

117(78%)

Family history of CV disease

Binary

44(5%)

18(9%)

19(12%)

Other

Binary

781(99%)

193(100%)

150(100%)

Symptoms

Chestpain

Binary

396(50%)

119(61%)

129(86%)

Syncope

Binary

48(6%)

15(7%)

6(4%)

Shortness of breath

Binary

198(25%)

42(21%)

42(28%)

Diaphoresis

Binary

62(7%)

13(6%)

14(9%)

Nausea and/or vomiting

Binary

76(9%)

17(8%)

20(13%)

Palpitations

Binary

135(17%)

24(12%)

5(3%)

Other symptoms

Binary

450(57%)

96(49%)

72(48%)

ECG Interpretation

ST elevation

Binary

119(15%)

23(11%)

28(18%)

ST depression

Binary

117(14%)

43(22%)

57(38%)

T wave inversion

Binary

124(15%)

37(19%)

19(12%)

NSTE-ACS outcome

NSTEMI

Binary

3(<1%)

7(3%)

52(34%)

Unstable Angina

Binary

4(<1%)

8(4%)

19(12%)

Non-ACS condition

Binary

778(99%)

179(92%)

89(59%)