Table 3 Characterization of included portable stroke technologies and associated diagnostic accuracy metrics
From: Portable stroke detection devices: a systematic scoping review of prehospital applications
No | Author | Year | Technology Name | Purpose | Specificity | Sensitivity | AUC | PPV | NPV | Comparator | Time to results |
|---|---|---|---|---|---|---|---|---|---|---|---|
1 | Michelson | 2015 | Hand-held EEG | Stroke detection | 50.40% (43.0–58.7%) | 91.7% (80.5–96.7%) | – | 39.6% (30.6–49.4%) | 94.40% (85.7–98.2%) | CT/MRI | 10 min |
2 | Wilkinson | 2020 | Muse | Stroke detection | 86% | 63% | – | – | – | CT, MRI | 3 min |
3 | Herzberg | 2014 | SonoSite Micromaxx; Philips CX50 | Stroke detection | 48% (29–67%) | 94% (86–98%) | – | 82% (72–89%) | 77% (52–93%) | CT, CTA, MRA | 5.6 min |
4 | Schlachetzki | 2012 | SonoSite Micromaxx; Philips CX50 | MCA occlusion | 98% (92.89–99.97%) | 90% (55.5–99.75%) | – | 90% (55.5–99.75%) | 98% (92.89–99.97%) | CT, MRA, neuro-Sonography | 5.6 min |
5 | Antipova | 2020 | SonoSite M-Turbo; Philips Sparq; Philips CX50 | LVO detection | 97% | 55% | 0.93 (0.865–0.996) | 75% | 93% | CT | 20 min |
ICH detection | 99% | 63% | 0.912 (0.829–0.996) | 91% | 92% | ||||||
6 | Erani | 2020 | Quick-20 | LVO detection | 80% | 41% | 68.9 | – | – | – | 3 min |
Acute stroke/TIA | 80% | 65% | 78.2 | ||||||||
7 | Kellner | 2018 | Cerebro-tech Visor | Severe vs. small stroke | 92% (75–99%) | 93% (83–98%) | 0.93 (0.85–0.97) | 96% (88–99%) | 86% (70–94%) | Triage scales, imaging | 30 sec |
Severe stroke detection | 87% (81–92%) | 93% (83–98%) | 0.93 (0.89–0.96) | 70% (61–77%) | 98% (94–99%) | ||||||
8 | Thorpe | 2018 | 2 MHz hand-held ultrasound | LVO detection | 82% (VAI) | 82% (VAI) | 0.88 (VAI) | – | – | CTA | 30 sec |
88% (VCI) | 91% (VCI) | 0.94 (VCI) | |||||||||
9 | Sergot | 2021 | PLD | LVO detection | 80% (77–83%) | 80% (74–85%) | – | – | – | CTA | 4.6 min |
10 | Persson | 2014 | 2 helmet-design prototypes | ICH/IS (S1) | – | – | 0.88 | – | – | Clinical, radiography | – |
ICH/IS (S2) | 0.85 | ||||||||||
0.87 | |||||||||||
ICH/Control | |||||||||||
11 | Robertson | 2010 | Infra-scanner | Any intracranial hemorrhage detection | 90.70% (86.4–93.7%) | 68.70% (58.3–77.6%) | – | 72.50% (62.0–81.1%) | 89% (84.6–92.3%) | CT | < 2 min |
Detection in Infrascanner limits | 90.70% (86.4–93.7%) | 88% (74.9–95.0%) | – | 63.70% (51.2–74.7%) | 97.6% (94.6–99.0%) | ||||||
12 | Liang | 2018 | Infra-scanner 2000 | Intracranial hematoma detection in Infrascanner limits | 93.6% (85–97.6%) | 100% (82.8–100%) | – | 82.8% (63.5–93.5%) | 100% (93.8–100%) | CT | < 3 min |
13 | Xu | 2017 | Infra-scanner 2000 | Intracranial hematoma detection in Infrascanner limits | 92.50% (78.5–98%) | 95.60% (83.6–99.2%) | 0.97 | 93.50% (81.1–98.3%) | 94.90% (81.4–99.1%) | CT, MRI | < 3 min |
14 | Peters | 2017 | Infra-scanner 2000 | Intracranial hematoma detection | 78.60% | 93.30% | – | – | – | CT | 4 min |
15 | Yuksen | 2020 | Infra-scanner 2000 | Intracranial hematoma detection | 44.4% (35.8–44.40%) | 100% (71.90–100%) | 0.722 | 35.5% (25.5–35.5%) | 100% (80.7–100%) | CT | 3 min |
16 | Kontojannis | 2019 | Infra-scanner 2000 | Any intracranial hematoma detection | 50.43% (41.03–59.80%) | 75% (64.63–83.62%) | – | 53.23% (47.76–58.62%) | 72.84% (64.16–80.07%) | CT | 3.74 min |
Intracranial hematoma detection in Infrascanner limits | 48.73% (40.71–56.80%) | 89.36% (76.90–96.45%) | – | 34.15% (30.2–38.33) | 93.90% (86.88–97.28) |