Technological Integration in Triage With AI and Machine Learning Compared With Conventional Triage and Narrative Synthesis of Telehealth for Patient Prioritization in Emergency and Acute Care Settings: Systematic Review and Meta-Analysis.
Faii O M Ashour, Waqas Sami, Intima Alrimawi et al.
✦ AI-curated · Sources linked
In 30 seconds
This systematic review and meta-analysis assessed AI- and machine-learning-assisted triage alongside tele-triage and digital tools in emergency and acute care. It included 27 comparative studies: 13 in quantitative analyses and 14 in narrative synthesis. Across four adult emergency department studies assessing critical care or deterioration outcomes, AI-assisted triage showed pooled discrimination of AUC 0.874 (95% CI 0.845-0.898), with substantial heterogeneity (I²=94.4%). Tele-triage evidence suggested operational improvements, but the review cautioned that discrimination does not establish better outcomes or implementation readiness.
Key findings
- Four adult ED studies reported pooled AI-assisted triage discrimination for critical care or deterioration outcomes of AUC 0.874 (95% CI 0.845-0.898; I²=94.4%).
- Four within-study comparisons favored AI-assisted models, with AUC differences of 0.097, 0.120, 0.108, and 0.065; paired variance data prevented pooling.
- Narrative evidence reported tele-triage reductions of 15%-40% in time to clinical assessment and 2-8 percentage points in left-without-being-seen rates.
- Exploratory modeling projected 12-38 fewer missed critical illness cases and 45-120 additional false-positive escalations per 10,000 ED visits under specified assumptions.
Why it matters
Triage tools may help identify patients at risk and improve front-door workflow, but their performance must be interpreted in context. AUC measures discrimination, not whether a tool improves decisions at clinically relevant thresholds or patient outcomes. The review highlights the need to assess calibration, external validity, workflow effects, equity, and costs before relying on these systems in routine care.
What to keep in mind
The evidence was highly heterogeneous, with incomplete threshold-level performance and calibration reporting, limited external validation, and sparse pediatric evidence. Economic and equity findings were projections rather than observed outcomes.