Skip to main content

Table 1 Baseline characteristicsa

From: Artificial intelligence algorithm to predict the need for critical care in prehospital emergency medical services

Characteristics

Development data

(n = 8,981,181)

Test data

(n = 2604)

p-valueb

Data

 Data type

National Emergency Department Infromation System (NEDIS)

Emergency Medical Service (EMS) Run Sheets

 

 Data source

Emergency department visit data

EMS run sheets

 

 Data period

1 January 2014–30 June 2016

1 September 2018–28 February 2019

 

Age

49.9 ± 18.9

61.5 ± 18.6

< 0.001

Female, No.(%)

4,511,654 (50.2%)

1411 (54.2%)

< 0.001

Initial vital signs, mean ± SD

 Systolic BP (mmHg)

131.2 ± 23.3

132.0 ± 24.6

0.271

 Diastolic BP (mmHg)

79.3 ± 13.9

83.7 ± 17.4

< 0.001

 Heart rate (/min)

83.8 ± 16.2

85.5 ± 20.5

< 0.001

 Respiratory rate (/min)

19.6 ± 2.7

17.7 ± 3.3

< 0.001

 Body temperature (°C)

36.7 ± 0.7

36.7 ± 0.8

< 0.001

Mental status, No.(%)

< 0.001

 Alert

8,674,058 (96.6%)

2513 (96.5%)

 

 Reacting to voice

161,624 (1.8%)

20 (0.8%)

 

 Reacting to pain

113,192 (1.3%)

46 (1.8%)

 

 Unresponsive

32,310 (0.3%)

25 (1.0%)

 

Trauma, No.(%)

2,536,815 (28.2%)

550 (21.1%)

< 0.001

Symptome onset to visit (contact) time, No.(%)

< 0.001

 –24 h

5,394,527 (60.1%)

2105 (80.8%)

 

 24 h–72 h

2,666,179 (29.7%)

448 (17.2%)

 

 72 h–7 Days

536,525 (6.0%)

38 (1.5%)

 

 7 Days–30 Days

258,641 (2.9%)

12 (0.5%)

 

 30 Days–

125,312 (1.4%)

1 (0.0%)

 

Outcomes, No.(%)

 Critical care

511,342 (5.7%)

319 (12.3%)

0.006

 In-hospital mortality

125,219 (1.4%)

30 (1.2%)

< 0.001

 Hospitalization

2,443,994 (27.1%)

1003 (38.5%)

< 0.001

  1. aBP denotes blood pressure
  2. bThe alternative hypothesis for this p-value was that there is a difference between the development and test data group for each variable