From the Emergency Services edition of October 24, 2026
MDPI conference paper: 99.6% accuracy for an EMS hospital-destination model built from 30 real records expanded with synthetic data
Universidad Bolivariana del Ecuador and Universidad de Guayaquil · Emergency medical services dispatch · Guayaquil, Ecuador
- Publisher
- MDPI Engineering Proceedings
- Original publication
- 2026-10-21
- Source retrieved
- 2026-10-25
What happened
The authors built a dispatcher decision-support prototype to recommend one of five destination hospitals. A seed dataset of 30 real records (March to June 2020) was expanded by Monte Carlo sampling and randomized coordinates to 3,572 instances, split 70:30. Random forest reached 0.996 accuracy and KNN 0.855. The abstract says the model diminishes response times, but no response-time outcome was measured, no case-level data or code is released, and the authors note random forest requires further validation and broader datasets.
Why it matters
A caution for EMS and 911 buyers: near-perfect accuracy figures in published dispatch AI can rest on synthetic data generated from tiny real samples, so they say little about field performance.
Evidence and measured results
Conference proceedings paper; synthetic augmentation from n=30; accuracy only; no clinical or operational outcome; no external validation.
Limitations and uncertainty
Tiny real sample; synthetic data likely makes the task easier; response-time benefit asserted, not tested; single city context.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-10-24; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- EMS agencies want faster, correct destination decisions at peak demand but face vendor and research accuracy claims of unclear quality.
- Stakeholders
- EMS medical director
- Dispatch center manager
- Hospital liaison
- Procurement officer
- Discovery questions
- What share of the model's training and test data was real?
- Were response times or outcomes measured?
- Does the model account for hospital diversion and specialty capability?
- Value hypothesis
- Rigorous validation requirements can screen out overstated dispatch AI before purchase (interpretation).
- Potential engagement
- A bounded evidence review of dispatch AI claims against agency data.
- Unsupported claims
- Reduced response times
- Field accuracy of 99.6%
- Clinical benefit
Pre-sales engineering
Role takeaway
- Fit
- Limited; illustrates a prototype, not a deployable system.
- Architecture and integration
- Would need CAD location feeds, hospital status data and ePCR clinical fields to be realistic.
- Prerequisites
- Real historical destination data
- Hospital capacity feeds
- Clinical protocol rules
- Constraints
- Synthetic data from 30 records
- Single city
- No code or data released
- Security
- Patient location and clinical data require privacy controls even in model development.
- Proof of value
- External validation on real held-out periods
- Clinical review of mismatches
- Measured transport time change
Delivery
Role takeaway
- Work
- Define validation gates and data requirements for any destination-recommendation pilot.
- Dependencies
- Historical CAD and ePCR data
- Medical direction
- Hospital data sharing
- Ownership
- EMS medical director with the dispatch center manager.
- Skills and adoption
- Dispatchers treat recommendations as advisory and document overrides.
- Governance checkpoints
- Data provenance review
- Medical director protocol approval
- Post-pilot outcome audit
- Acceptance criteria
- Proposed: agreement with protocol-correct destinations on real data above agreed level
- Proposed: no increase in transport time
- Risks
- Inflated accuracy expectations
- Ignoring capacity and specialty
- Privacy gaps
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?
A simple classifier on location, sector, times and distance; any real deployment would need CAD integration and live hospital capacity data the model lacks.
Governance
Who approves, reviews and stays accountable for outcomes?
Shows why procurement and medical direction should demand real-world validation data before accepting accuracy claims.
Security and privacy
What data, permissions and controls need testing?
Real patient location data was used for the seed set; synthetic data generation is not a substitute for privacy review in deployment.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Destination decisions depend on clinical and capacity factors dispatchers know that this model does not include.
Procurement
What should contracts, pricing and exit terms secure?
Ask vendors and researchers what share of training and test data is real, and whether outcomes were measured.
Operating model
Which teams own the service once it runs?
Any hospital-recommendation tool should stay advisory under medical-direction protocols.
Publication history
- 2026-10-24Emergency Services · Issue 484 resources
Stable resource ID: mdpi-engproc-159-25-prehospital-hospital-destination-rf-0996-synthetic-30-records-20261021