Artificial intelligence in Emergency Medical Services dispatching: assessing the potential impact of an automatic speech recognition software on stroke detection taking the Capital Region of Denmark as case in point

Table 1 Baseline characteristics stroke patients in EMS of the Capital Region of Denmark 2016–2018

Characteristic	N	%
Stroke patients	9049	100.00
Type
Ischaemic stroke	8215	90.78
Haemorrhagic stroke	834	9.22
Sex
Male	4827	53.34
Female	4222	46.66
Age—mean (Min.\|Q1\|Median\|Q3\|Max.)	71.28 (18\|63\|73\|81\|104)
EMS Number
1-1-2	5180	57.24
1813-Medical Helpline	3869	42.76
Stroke criteria
Stroke relevant criteria	5973	66.01
Stroke nonrelevant criteria	1861	20.57
Missing criteria	1215	13.43
Response [53]
A (life threatening)	6128	67.72
B (urgent, no immediate risk to life)	1028	11.36
C (non-urgent/planned transport)	24	0.27
D (no treatment/medical care required)	4	0.04
F (referral/advice)	45	0.50
Non-Urgent Response*	1820	20.11
Year
2016	2807	31.02
2017	3021	33.38
2018	3221	35.59
Weekday (Monday–Friday)	5854	64.69
Weekend (Saturday–Sunday)	3195	35.31
Incident time of day
1–6	677	7.48
7–12	3542	39.14
13–18	3096	34.21
19–0	1734	19.16
Treatment^†	1652	18.26
Thrombolysis	1449	16.01
Reperfusion	217	2.40
Endovascular	47	0.52
Surgical	43	0.48
Thrombectomy	239	2.64
Time-to-call
≤ 4.5 h	6013	66.45
≤ 24 h	8095	89.46

*A patient with a “Non-Urgent Response” typically received a referral to a physician or was advised to drive to the hospital
^†A patient who received multiple treatments was considered individually for the individual treatments but cumulated for the total number of treatments

ISSN: 1757-7241