Lighthouse AdvisorySLED AI Adoption Intelligence

Public Sector & Government · Issue 03 ·

Emergency Services

Three newly catalogued historical sources cover U.S. EMS workflow concerns, a Chinese integrated EMS deployment and Arlington's non-emergency storm-call diversion. Evidence includes reported operational improvement, material attribution limits and frontline constraints. No new September 8 deployment or AI-specific survival benefit is claimed. Role interpretations address integration, continuity, governance, privacy, procurement and workforce. One supported cross-source pattern; gaps include current hurricane/fire results, independent U.S. clinical efficacy and inaccessible supplementary tables/dashboard visuals.

Evidence records
3
Cross-source patterns
1
Evidence classes
2 academic research1 government evaluation
Outcomes
1 cautionary1 mixed1 emerging
Source freshness
2 recent1 undated
Research completed
2026-09-09

Choose a role to see its takeaway beside every record in the ledger.

Synthesis · Lighthouse Advisory interpretation

Patterns across the evidence

1 pattern, each supported by at least two sources

  1. Pair operational improvement with unresolved and unsafe cases

    Arlington's reduced call workload and Anyang's reported response-time improvement describe different service gains. Neither alone establishes safe resolution for every caller or patient. Evaluation should include misroutes, repeat contacts, false alerts and excluded cases alongside throughput, with separate denominators and attribution to non-AI changes.

    Operating questionCan the agency account for unsuccessful or excluded journeys as rigorously as the work it reports saving?

    Supporting evidenceArlington County Public Safety Communications and Emergency Management; Timothy KaneXiaopeng Liu and colleagues; Henan clinical institutions and technology collaborators

Full record · every source keeps its link and limitations

Evidence ledger

3 records
  1. Academic researchCautionaryRecent

    U.S. EMS interviews identify workflow and autonomy constraints for AI support

    Interviews identify coordination, attention and autonomy concerns surrounding prospective EMS AI.

    Emily Hou and colleagues; Wellesley College, University of New Hampshire and University of HaifaUnited StatesJune 15, 2026 arXiv version; interviews June–July 2025

    Why it matters, evidence and limitations
    Why it matters
    Directly relevant to U.S. municipal, private and volunteer EMS workflow discovery; no deployment efficacy is established.
    Evidence and measured results
    Researchers interviewed 25 U.S. EMS clinicians in June–July 2025 and used thematic analysis. Tables map technology interactions and hypothetical applications across response stages. No controlled baseline or measured AI response-time benefit.
    Limitations and uncertainty
    Interview and hypothetical-system evidence, not direct observation. Some ideation assumed perfectly accurate AI. Section 5.5.4 conflicts with 5.4.1 about assessment/treatment demand; avoid that stage-specific conclusion.
    CHIWORK 2026 paper, author manuscript on arXiv Role takeaways & implementation
  2. Academic researchMixedRecent

    Anyang EMS system reports faster response, with bundled-intervention and validation limits

    An integrated EMS rollout was associated with improvement; the study cannot isolate AI's contribution.

    Xiaopeng Liu and colleagues; Henan clinical institutions and technology collaboratorsAnyang, China; three countiesJuly 6, 2026; page lists version of record August 28, 2026

    Why it matters, evidence and limitations
    Why it matters
    Useful for examining regional EMS-to-hospital integration; Chinese staffing, infrastructure and clinical pathways limit U.S. transfer.
    Evidence and measured results
    Main text reports 1,208 cases (587 before, 621 after), historical controls and interrupted time series. Median call-to-scene time fell from 9.8 to 6.7 minutes. Deterioration alerts had reported 41% positive predictive value. Baseline was 2022; full-operation evaluation July–December 2023.
    Limitations and uncertainty
    No concurrent control, six-month follow-up, local model validation and no cost-effectiveness or long-term survival result. Excludes patients declared dead before hospital arrival. Separate tables/supplements could not be inspected; figures retained are corroborated in main text and abstract.
    Peer-reviewed before-and-after implementation study; includes technology-company affiliation Role takeaways & implementation
  3. Government evaluationEmergingUndated source

    Arlington describes non-emergency call diversion with opt-out and reporting-portal dependencies

    Arlington reports non-emergency workload reductions after call routing and related process changes.

    Arlington County Public Safety Communications and Emergency Management; Timothy KaneArlington County, Virginia, United StatesUndated PDF; includes operational data through February 15, 2025

    Why it matters, evidence and limitations
    Why it matters
    Storm-damage and fire-alarm administrative workloads make this relevant to emergency communications; towing is ancillary, with no Public Safety cross-tag.
    Evidence and measured results
    Operator describes approximately 30 days of utterance collection, AWS routing and SMS links, with human opt-out. It reports 22,167 fewer administrative calls comparing the first six months of 2022 and 2024. No controlled comparison, misrouting rate or confirmed emergency-response improvement.
    Limitations and uncertainty
    Operator report with concurrent towing outsourcing; reduction cannot be attributed solely to AI. Visual dashboard pages did not yield inspectable images through tools. Future multilingual bot capabilities are not evaluated results.
    Local government operator case report Role takeaways & implementation

How to read this edition

Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.

Academic research
Research produced through an academic institution or peer-reviewed venue.
Government evaluation
A public body’s measured evaluation or documented pilot.