Japanese group (National Cerebral and Cardiovascular Center, Kobe City Medical Center General Hospital, et al.) derived and evaluated a prehospital LVO screening scale from a prospective registry.
- Problem/setting: EMS routing — deciding which patients go to thrombolysis- and/or thrombectomy-capable centers.
- Paper checklist / EMS protocol; complements (does not replace) prehospital imaging or AI triage.
- Cohort: suspected-stroke patients transported to 6 stroke centers within 24 h of onset; assessed in the field by EMS and again on hospital arrival.
- Method: 7 candidate items → logistic regression weighting; final 6-item, 1-point-each scale (atrial fibrillation, conjugate eye deviation, unilateral spatial neglect, aphasia, facial weakness, upper-extremity weakness); dysarthria dropped (coefficient ≈0).
Evidence quality: prospective, multicenter, EMS-administered — operationally realistic.
- Performance (by EMS): PPV >50% at a 3-point cutoff; NPV 90% at 2 points; AUC ≈0.7 or higher for ICA, M1, M2 occlusions. EMS scoring slightly less accurate than in-hospital scoring. [Enrolled sample size (N) not reported in the published abstract; figures are as-reported, without CIs — no PMC full text available for deeper verification.]
- Failure modes: field inter-rater variability; lower accuracy for posterior (basilar) occlusion — the scale performed comparably (AUC ≈0.7+) for ICA, M1, and M2, and the source publishes no BA AUC; no imaging confirmation prehospital.
Commercial significance: reaffirms unmet need for objective prehospital LVO triage — the exact gap AI/MSU imaging aims to fill.
- Order: field assessment → routing decision — upstream of any imaging/AI inference.
- Operational friction: training and consistency across EMS crews.
IDs: PMID 42442625 · DOI 10.1016/j.jstrokecerebrovasdis.2026.108698 Sources: Primary https://doi.org/10.1016/j.jstrokecerebrovasdis.2026.108698 (PubMed / J Stroke Cerebrovasc Dis)