Lighthouse AdvisorySLED AI Adoption Intelligence
← Back to results

From the Emergency Services edition of September 11, 2026

Academic researchMixedNewly relevant · Dec 2024

Sim911 training deployment reports favorable user feedback while retaining hallucination limits

Vanderbilt University and Metro Nashville Department of Emergency Communications · 911 dispatcher training · Nashville, Tennessee, United States

Publisher
Vanderbilt University and Metro Nashville Department of Emergency Communications
Original publication
December 26, 2024, inspected preprint version 3
Source retrieved
2026-09-12
Read original source

What happened

Sim911 supports supervised dispatcher practice, but reported satisfaction and simulation quality do not establish operational competence.

Why it matters

Useful for emergency medical, fire and severe-weather call training; the study also includes routine policing scenarios, without separate emergency-services effect estimates.

Evidence and measured results

The study uses 2,641 historical-call configurations for component comparisons and reports 228 completed deployment simulations. Nine of ten respondents rated it at least comparable to human-led training; respondents included two trainees, two dispatchers and six training officers. Tables assess linguistic, location and scenario proxies, not retained skills or patient outcomes. Logged simulation time is not a controlled net labor-saving estimate.

Limitations and uncertainty

Single-center developer/operator study with a small convenience survey. Hallucinations remain. The accessible version is a 2024 preprint; the 2025 proceedings PDF could not be retrieved.

Put this evidence to work

Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-09-12; this does not change the original publication date. Labels below come from the analysis itself.

Sales

Role takeaway

Engage the training manager, dispatch director, medical-dispatch lead and workforce development staff about practice capacity and rare-call exposure. Ask how trainees are currently assessed, which scenarios are difficult to stage and how much instructor time is actually constrained. Offer a bounded curriculum and evaluation pilot focused on medical, fire and disaster calls. The value hypothesis is more supervised practice opportunities without proportionate role-playing demand. Do not convert favorable user ratings into proven competence or assume logged simulation hours equal net savings. Include scenario maintenance and coaching costs. Applicability depends on local protocol fit and staff willingness to review imperfect generated calls.

Pre-sales engineering

Role takeaway

Create an isolated training environment with approved scenario inputs and no permission to route calls or modify operational CAD. Prerequisites include current local protocols, validated locations, a privacy-reviewed corpus and a pinned model configuration. Challenge retrieval and evaluator components with misleading instructions embedded in call text, invalid addresses and contradictory incident details. Measure undetected errors independently rather than relying solely on the system's own evaluator. Compare cloud APIs with local options for data handling and service continuity; no hosting choice is validated by this study. A useful proof of value pairs instructor scoring with delayed trainee assessment, covering missing data, language variation and service failure.

Delivery

Role takeaway

A training officer should own rollout with a dispatch quality specialist, data steward and IT support. Select a limited emergency scenario set, cleanse source material, test instructor controls and train users to stop and flag implausible output. Dependencies include review time and a maintained protocol inventory. Approve corpus access and evaluation design before the pilot, then review results before expanding. Proposed acceptance includes zero connections to live dispatch, logged instructor interventions, successful shutdown and recovery, and prespecified delayed-skill and workload comparisons with existing training. Monitor stereotype complaints and scenario drift. Proposed criteria are not study findings; substitution for experienced coaching requires separate evidence.

Implementation considerations

Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.

Architecture and integration

What must connect, and where does the AI sit in the workflow?

Separate training from emergency call routing; maintain versioned local protocols, scenario data and evaluator components.

Governance

Who approves, reviews and stays accountable for outcomes?

Require instructor review and separate authorization for any expansion beyond simulation.

Security and privacy

What data, permissions and controls need testing?

Audit de-identification and external API payloads; derived scenarios can still expose sensitive source details.

Accessibility and workforce

Who is affected, and what skills or accommodations follow?

Involve language-access and disability advisers in checking simulated caller profiles; proxy tag matching cannot establish equitable real-world service.

Procurement

What should contracts, pricing and exit terms secure?

Require independently scored skill retention, instructor workload accounting and local protocol support before promising training savings.

Operating model

Which teams own the service once it runs?

Training management owns curriculum; clinical/dispatch specialists validate scenarios; IT manages service dependencies and data boundaries.

What changed

Previously absent from all 217 archive records. Adds historical deployed training evidence, distinct from the already archived DispatchMAS emergency dialogue simulation.

Publication history

  1. 2026-09-11Emergency Services · Issue 063 resources
Read preserved resource versions (JSON)

Stable resource ID: sim911-nashville-training-preprint-2024