Research Article

Machine Learning-Based Models for Prediction of Critical Illness at Community, Paramedic, and Hospital Stages

Table 1

Demographics and clinical characteristics between patients with critical illness and no critical illness.

VariablesCritical illnessNo critical illness

Age (years), median (IQR)70 (57–79)51 (34–66)
Gender
 Female141,957 (41.10)3,048,163 (54.08)
 Male203,444 (58.90)2,588,161 (45.92)
Mode of arrival
 Public/air-ambulance129,754 (37.57)951,941 (16.89)
 Private ambulance100,543 (29.11)229,180 (4.07)
 Police/public officer car/other transportation/walking114,030 (33.01)4,445,218 (78.87)
 Others1,061 (0.31)9,741 (0.17)
 Unknown13 (0.00)244 (0.00)
ED visit date
 On weekday255,705 (74.03)3,703,819 (65.71)
 On weekend89,696 (25.97)1,932,505 (34.29)
ED visit time
 Night (0–8 hr)52,871 (15.31)1,225,776 (21.75)
 Day (8−16 hr)168,215 (48.70)2,112,811 (37.49)
 Evening (16−24 hr)124,315 (35.99)2,297,737 (40.77)
Symptom onset to ED arrival time (min), median (IQR)347 (80−1,746)494 (120−2,105)
Mental status (AVPU)
 Alert258,331 (74.79)5,508,812 (97.74)
 Verbal response38,168 (11.05)80,213 (1.42)
 Painful response38,694 (11.20)43,376 (0.77)
 Unresponsive10,208 (2.96)3,923 (0.07)
Vital signs
 Body temperature (°C), median (IQR)36.6 (26.2–37.0)36.6 (36.4–37.1)
 Heart rate (bpm per min), median (IQR)90 (76–108)82 (74–95)
 Respiratory rate (per min), median (IQR)20 (18–22)20 (18–20)
 Systolic blood pressure (mmHg), median (IQR)128 (102–150)130 (116–147)
 Diastolic blood pressure (mmHg), median (IQR)76 (60–90)80 (70–90)
KTAS level
 Level 1 (resuscitation)33,301 (9.64)21,562 (0.38)
 Level 2 (emergent)140,934 (40.80)385,822 (6.85)
 Level 3 (urgent)150,884 (43.68)2,543,145 (45.12)
 Level 4 (less urgent)17,009 (4.92)2,126,146 (37.72)
 Level 5 (nonurgent)3,273 (0.95)559,649 (9.93)
Number of diagnostic codes, median (IQR)4 (3–7)1 (1–2)
Total345,401 (5.77)5,636,324 (94.23)

ED: emergency department; IQR: interquartile range; and KTAS: Korean Triage and Acuity Scale.