{"resourceId":"mdpi-engproc-159-25-prehospital-hospital-destination-rf-0996-synthetic-30-records-20261021","versions":[{"version":"external-ff98749a045c772fd5fefec6ea4159d001ddc19b613ee538eea108d11d8a8c15","resource":{"id":"mdpi-engproc-159-25-prehospital-hospital-destination-rf-0996-synthetic-30-records-20261021","title":"MDPI conference paper: 99.6% accuracy for an EMS hospital-destination model built from 30 real records expanded with synthetic data","organization":"Universidad Bolivariana del Ecuador and Universidad de Guayaquil","sector":"Emergency medical services dispatch","geography":"Guayaquil, Ecuador","publishedAt":"2026-10-21","publicationDate":"2026-10-21","eventDate":null,"sourceName":"MDPI Engineering Proceedings","sourceLabel":"Engineering Proceedings 159(1):25: Machine Learning for Prediction and Optimization of Response Time in Prehospital Clinical Emergencies","sourceUrl":"https://www.mdpi.com/2673-4591/159/1/25","evidenceClass":"academic-research","outcomeClass":"cautionary","topics":["knowledge-work","data-security","governance-procurement"],"finding":"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.","sledRelevance":"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":"Conference proceedings paper; synthetic augmentation from n=30; accuracy only; no clinical or operational outcome; no external validation.","architectureImplications":"A simple classifier on location, sector, times and distance; any real deployment would need CAD integration and live hospital capacity data the model lacks.","governanceImplications":"Shows why procurement and medical direction should demand real-world validation data before accepting accuracy claims.","securityPrivacyImplications":"Real patient location data was used for the seed set; synthetic data generation is not a substitute for privacy review in deployment.","caveats":"Tiny real sample; synthetic data likely makes the task easier; response-time benefit asserted, not tested; single city context.","streamIds":["emergency-services"],"roles":{"sales":"For sales, the framing point is credibility: a 99.6% accuracy claim built on 30 real records plus synthetic data is the kind of figure EMS and 911 buyers should question. Use this as a discovery aid, asking what data any destination or routing AI was trained and tested on. It cannot support claims of faster response times, better patient outcomes or deployable accuracy, because none were measured.","engineering":"For pre-sales engineering, the model is a random forest using coordinates, sector, times and distance to pick among five hospitals. Real destination decisions also depend on patient condition, specialty capability and hospital diversion status, which the model lacks. Synthetic augmentation from a 30-record seed means test accuracy mostly reflects the generator. Proof of value would require real CAD and ePCR history, external validation and measured response-time effects.","delivery":"For delivery, the lesson is to set validation gates before any destination-recommendation tool goes live: real historical data, a held-out period, clinical review of disagreements and measured transport times. Ownership sits with the EMS medical director and the dispatch center. Acceptance should rest on real-world agreement with protocol-correct destinations. Risks include trusting inflated accuracy, ignoring hospital capacity and specialty needs, and privacy gaps in data handling."},"roleDetails":{"sales":{"customerProblem":"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"],"discoveryQuestions":["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?"],"valueHypothesis":"Rigorous validation requirements can screen out overstated dispatch AI before purchase (interpretation).","potentialEngagement":"A bounded evidence review of dispatch AI claims against agency data.","unsupportedClaims":["Reduced response times","Field accuracy of 99.6%","Clinical benefit"]},"engineering":{"fit":"Limited; illustrates a prototype, not a deployable system.","architectureIntegration":"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.","proofOfValue":["External validation on real held-out periods","Clinical review of mismatches","Measured transport time change"]},"delivery":{"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.","skillsAdoption":"Dispatchers treat recommendations as advisory and document overrides.","governanceCheckpoints":["Data provenance review","Medical director protocol approval","Post-pilot outcome audit"],"acceptanceCriteria":["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"]}},"retrievedAt":"2026-10-24T21:59:30-05:00","enrichedAt":"2026-10-24T22:02:19-05:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Destination decisions depend on clinical and capacity factors dispatchers know that this model does not include.","procurementImplications":"Ask vendors and researchers what share of training and test data is real, and whether outcomes were measured.","operatingModelImplications":"Any hospital-recommendation tool should stay advisory under medical-direction protocols.","sourceVerification":{"responseId":"webfetch-mdpi-2026-10-24","openedUrl":"https://www.mdpi.com/2673-4591/159/1/25","referenceExcerpt":"Because the original operational dataset was limited (seed dataset of n = 30 records containing patient location and assigned hospital destination)","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}