Ambulance dispatch prioritisation for traffic crashes using machine learning: A natural language approach

BackgroundCalls for emergency medical assistance at the scene of a motor vehicle crash (MVC) substantially contribute to the demand on ambulance services. Triage by emergency medical dispatch systems is therefore important, to ensure the right care is provided to the right patient, in the right amount of time. A lights and sirens (L&S) response is the highest priority ambulance response, also known as a priority one or hot response. In this context, over triage is defined as dispatching an ambulance with lights and sirens (L&S) to a low acuity MVC and under triage is not dispatching an ambulance with L&S to those who require urgent medical care. We explored the potential for crash characteristics to be used during emergency ambulance calls to identify those MVCs that required a L&S response.MethodsWe conducted a retrospective cohort study using ambulance and police data from 2014 to 2016. The predictor variables were crash characteristics (e.g. road surface), and Medical Priority Dispatch System (MPDS) dispatch codes. The outcome variable was the need for a L&S ambulance response. A Chi-square Automatic Interaction Detector technique was used to develop decision trees, with over/under triage rates determined for each tree. The model with an under/over triage rate closest to that prescribed by the American College of Surgeons Committee on Trauma (ACS COT) will be deemed to be the best model (under triage rate of ≤ 5% and over triage rate of between 25–35%.ResultsThe decision tree with a 2.7% under triage rate was closest to that specified by the ACS COT, had as predictors—MPDS codes, trapped, vulnerable road user, anyone aged 75 + , day of the week, single versus multiple vehicles, airbag deployment, atmosphere, surface, lighting and accident type. This model had an over triage rate of 84.8%.ConclusionsWe were able to derive a model with a reasonable under triage rate, however this model also had a high over triage rate. Individual EMS may apply the findings here to their own jurisdictions when dispatching to the scene of a MVC.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12873-022-00609-5.

Effect of a Medical Priority Dispatch System key question addition in the seizure/convulsion/fitting protocol to improve recognition of ineffective (agonal) breathing

Diabetes mellitus, although a chronic disease, also can cause acute, sudden symptoms requiring emergency intervention. In these cases, Emergency Medical Dispatchers (EMDs) must identify true diabetic complaints in order to determine the correct care. In 911 systems utilizing the Medical Priority Dispatch System (MPDS), International Academies of Emergency Dispatch-certified EMDs determine a patient's chief complaint by matching the caller's response to an initial pre-scripted question to one of 37 possible chief complaints protocols. The ability of EMDs to identify true diabetic-triggered events reported through 911 has not been studied. The primary objective of this study was to determine the percentage of EMD-recorded patient cases (using the Diabetic Problems protocol in the MPDS) that were confirmed by either attending paramedics or the hospital as experiencing a diabetic-triggered event. This was a retrospective study involving six hospitals, one fire department, and one ambulance service in Salt Lake City, Utah USA. Dispatch data for one year recorded under the Diabetic Problems protocol, along with the associated paramedic and hospital outcome data, were reviewed/analyzed. The outcome measures were: the percentage of cases that had diabetic history, percentage of EMD-identified diabetic problems cases that were confirmed by Emergency Medical Services (EMS) and/or hospital records as true diabetic-triggered events, and percentage of EMD-identified diabetic patients who also had other medical conditions. A diabetic-triggered event was defined as one in which the patient's emergency was directly caused by diabetes or its medical management. Descriptive statistics were used for categorical measures and parametric statistical methods assessed the differences between study groups, for continuous measures. Three-hundred ninety-three patient cases were assigned to the Diabetic Problems Chief Complaint protocol. Of the 367 (93.4%) patients who had a documented history of diabetes, 279 (76%) were determined to have had a diabetic-triggered event. However, only 12 (3.6%) initially assigned to this protocol did not have a confirmed history of diabetes. Using the MPDS to select the Diabetic Problems Chief Complaint protocol, the EMDs correctly identified a true diabetic-triggered event the majority of the time. However, many patients had other medical conditions, which complicated the initial classification of true diabetic-triggered events. Future studies should examine the associations between the five specific Diabetic Problems Chief Complaint protocol determinant codes (triage priority levels) and severity measures, eg, blood sugar level and Glasgow Coma Score.

Use of dispatch codes for obvious/expected deaths: Maintaining patient safety while reducing the number of lights-and-sirens responses

ObjectiveEmergency medical dispatch is intended to ensure that emergency medical services (EMS) allocate appropriate resources for the treatment of patients with status epilepticus (SE). However, it is unclear whether dispatch algorithms accurately identify those patients having a seizure-related medical emergency and how dispatch algorithms influence what prehospital resources are allocated for the encounter.MethodsWe performed a cross-sectional analysis of prehospital encounters for SE using data from the 2019 ESO Data Collaborative. We included patients who were ≥18 years of age, had an EMS diagnostic impression of SE, and did not have a cardiac arrest. We examined the dispatch-determined complaint designated by the emergency medical dispatch (EMD) code, dispatch-determined level of acuity (A, B, C, D), ambulance response, and training level of the responding prehospital professional.ResultsOf the 18,515 patient encounters for SE with an EMD code, 8,279 (44.9%) were women, and the mean age was 40.0 years (SD 19.7). There were 13,829 (75%) encounters that received a dispatch code for seizures/convulsions and 4,686 (25%) with a dispatch code for a non-seizure-related condition. Among encounters for SE identified by dispatch as seizures/convulsions, 6,412 (46.4%) were designated high acuity, 6,626 (63.6%) were designated low acuity, and the majority received emergent ambulance responses (98.1% among those designated high acuity and 81.8% among those designated low acuity) and an Advanced Life Support-trained responder (93.7% among those designated high acuity and 92.7% among those designated low acuity). Median response times were similar for all acuity levels (9.1, 8.8, 9.1, and 8.3 minutes for A-D, respectively).ConclusionApproximately one-fourth of SE cases were categorized as a non-seizure related condition at dispatch, and fewer than half received the highest acuity determinant code. Despite this, dispatch-assigned acuity did not have a strong relationship with the ambulance response or training level of the EMS responder or response time, suggesting that use of dispatch algorithms might be further optimized and highlighting a potential area to improve quality of EMS care.

Although the Medical Priority Dispatch System (MPDS) is widely used by emergency medical services (EMS) dispatchers to determine dispatch priority, there is little evidence that it reflects patient acuity. The Canadian Triage and Acuity Scale (CTAS) is a standard patient acuity scale widely used by Canadian emergency departments and EMS systems to prioritize patient care requirements. To determine the relationship between MPDS dispatch priority and out-of-hospital CTAS. All emergency calls on a large urban EMS communications database for a one-year period were obtained. Duplicate calls, nonemergency transfers, and canceled calls were excluded. Sensitivity and specificity to detect high-acuity illness, as well as positive predictive value (PPV) and negative predictive value (NPV), were calculated for all protocols. Of 197,882 calls, 102,582 met inclusion criteria. The overall sensitivity of MPDS was 68.2% (95% confidence interval [CI] = 67.8% to 68.5%), with a specificity of 66.2% (95% CI = 65.7% to 66.7%). The most sensitive protocol for detecting high acuity of illness was the breathing-problem protocol, with a sensitivity of 100.0% (95% CI = 99.9% to 100.0%), whereas the most specific protocol was the one for psychiatric problems, with a specificity of 98.1% (95% CI = 97.5% to 98.7%). The cardiac-arrest protocol had the highest PPV (92.6%, 95% CI = 90.3% to 94.3%), whereas the convulsions protocol had the highest NPV (85.9%, 95% CI = 84.5% to 87.2%). The best-performing protocol overall was the cardiac-arrest protocol, and the protocol with the overall poorest performance was the one for unknown problems. Sixteen of the 32 protocols performed no better than chance alone at identifying high-acuity patients. The Medical Priority Dispatch System exhibits at least moderate sensitivity and specificity for detecting high acuity of illness or injury. This performance analysis may be used to identify target protocols for future improvements.

To assess whether repositioning of ambulance stations in a rural county of Alabama can improve emergency medical services (EMS) response time to motor vehicle crashes (MVCs) without adversely affecting response time to non-MVC-related emergencies. Using geographical information system software, locations of MVCs during a 9-month period in a rural county of Alabama were plotted on a map. A single ambulance station provided EMS for the entire county. Based on the number of ambulances serving the county and concentrated areas of MVCs, the county was geographically divided into two regions. A new ambulance station was assigned to each region based on high MVC concentrations and access to a major thoroughfare. The number of ambulances in-service did not change. Following establishment of both ambulance stations (redeployment), data were prospectively collected for EMS miles to scene, EMS time to scene, fatalities, and type of call (MVC vs. non-MVC) during a 9-month period (January 2006 to September 2006). The prospective data were compared with historical data (non-redeployment) from a similar time period (January 2005 to September 2005). During the redeployment period, 597 EMS calls were documented, 106 (17.8%) of which were MVCs. In all, 764 EMS calls were documented before the redeployment period, 62 (8.1%) of which were MVCs. During the redeployment period, the mean miles EMS traveled to an MVC scene was 8.6 miles versus 10.7 miles before redeployment (p=0.038). The mean time to an MVC scene was 8.0 minutes during redeployment versus 9.5 minutes before redeployment (p=0.03). During the redeployment period, the mean time to non-MVC emergencies was 8.6 minutes versus 9.2 minutes during the period before redeployment (p=0.27). Utilizing geographical information system software, EMS response time to MVCs could be improved in rural areas by optimal location of ambulance stations based on geographical highest concentration of MVCs and vicinity of major thoroughfares. This can be accomplished without adversely affecting response time to non-MVC-related emergencies.

The Medical Priority Dispatch System's ability to predict cardiac arrest outcomes and high acuity pre-hospital alerts in chest pain patients presenting to 9-9-9

Objectives Emergency medical services (EMS) systems increasingly grapple with rising call volumes and workforce shortages, forcing systems to decide which responses may be delayed. Limited research has linked dispatch codes, on-scene findings, and emergency department (ED) outcomes. This study evaluated the association between dispatch categorizations and time-critical EMS responses defined by prehospital interventions and ED outcomes. Secondarily, we proposed a framework for identifying dispatch categorizations that are safe or unsafe to hold in queue. Methods This retrospective, multi-center analysis encompassed all 9-1-1 responses from 8 accredited EMS systems between 1/1/2021 and 06/30/2023, utilizing the Medical Priority Dispatch System (MPDS). Independent variables included MPDS Protocol numbers and Determinant levels. EMS treatments and ED diagnoses/dispositions were categorized as time-critical using a multi-round consensus survey. The primary outcome was the proportion of EMS responses categorized as time-critical. A non-parametric test for trend was used to assess the proportion of time-critical responses Determinant levels. Based on group consensus, Protocol/Determinant level combinations with at least 120 responses (∼1 per week) were further categorized as safe to hold in queue (<1% time-critical intervention by EMS and <5% time-critical ED outcome) or unsafe to hold in queue (>10% time-critical intervention by EMS or >10% time-critical ED outcome). Results Of 1,715,612 EMS incidents, 6% (109,250) involved a time-critical EMS intervention. Among EMS transports with linked outcome data (543,883), 12% had time-critical ED outcomes. The proportion of time-critical EMS interventions increased with Determinant level (OMEGA: 1%, ECHO: 38%, p-trend < 0.01) as did time-critical ED outcomes (OMEGA: 3%, ECHO: 31%, p-trend < 0.01). Of 162 unique Protocols/Determinants with at least 120 uses, 30 met criteria for safe to hold in queue, accounting for 8% (142,067) of incidents. Meanwhile, 72 Protocols/Determinants met criteria for unsafe to hold, accounting for 52% (883,683) of incidents. Seven of 32 ALPHA level Protocols and 3/17 OMEGA level Protocols met the proposed criteria for unsafe to hold in queue. Conclusions In general, Determinant levels aligned with time-critical responses; however, a notable minority of lower acuity Determinant level Protocols met criteria for unsafe to hold. This suggests a more nuanced approach to dispatch prioritization, considering both Protocol and Determinant level factors.

Objective. To determine the association between pediatric restraint use during motor vehicle crashes and patient transports by emergency medical services (EMS) providers. Methods. Children under 16 years of age who were involved in motor vehicle crashes in Houston or Harris County, Texas, in 1997 were identified from Texas Department of Public Safety (DPS) crash data. DPS data were linked probabilistically to City of Houston Fire Department EMS data to identify whether the children were transported from the scene by EMS. Odds ratios were calculated to determine the association between restraint use and EMS transport. Logistic regression was used to control for age and injury severity in this relationship. Results. From the linked data for 1997, 1,696 children were involved in reported automobile or truck crashes to which EMS responded. Of the 1,580 children about whom information concerning restraint use was available, 1,309 (83%) were wearing some type of safety restraint and 1,342 (85%) were transported by EMS. Adjusting for injury severity, children wearing safety restraints during motor vehicle crashes were 60% (95% confidence interval 34−75%) less likely to be transported by EMS than those not wearing safety restraints during such crashes. Conclusions. The use of safety restraints during motor vehicle crashes is associated with a reduction in the number of children transported from the scene by EMS providers. This effect is independent of age and injury severity.

Statistical modeling of emergency medical services’ response and rescue times to road traffic crashes in the Kingdom of Saudi Arabia

Background: Accurate identification of ischemic stroke from large vessel occlusion (LVO) by 9-1-1 professionals can facilitate prompt patient transfer to stroke centers that perform endovascular thrombectomy. We sought to determine the proportion of LVO stroke patients that were evaluated using the Medical Priority Dispatch System's (MPDS) Stroke Diagnostic Tool (SDxT). Methods: Structured, retrospective review of our institution’s stroke registry, EMS Records, and dispatch codes between January 1, 2019 and November 30, 2022. Inclusion: Adult (age &gt;17), accessed the 9-1-1 system and transported by a ground EMS agency for an acute stroke. Exclusion: age &lt;18, pregnant, incarcerated, cases with missing 9-1-1 data, and intracranial hemorrhage. Two trained/monitored abstractors, blinded to the study purpose, used a standardized data collection tool/dictionary to extract data. Data included MPDS codes, SDxT outcome, EMS care, demographics, hospital NIHSS score, advanced imaging, tPA administration, and mechanical thrombectomy. Primary outcome: the proportion of LVO cases that had an acute stroke (28 protocol) dispatch code. Secondary outcome: the proportion of LVO cases that that a SDxT score indicating “strong or clear evidence of a stroke”. We report proportions and 95% CI's as appropriate. Results: 60 patients met inclusion/exclusion criteria. 11/60, 18.3 % (95%CI= 10.4, 30) of the patients had complete data and a LVO. 54% were male, the median (IQR) age was 69.5 (61.3-77.5), and the median (IQR) NIHSS was 16 (5-19.75). Primary outcome: 45.5 % (95% CI= 21.3.72.0) of the LVO’s were coded as an Acute Stroke by the 9-1-1 center. Of those, 3/5 utilized the stroke diagnostic tool, all 3 of these were determined to have "Clear Evidence of Stroke". Limitations: Retrospective, unlinked data, small number of LVO's. Conclusion: When utilized, the SDxT correctly identified all LVO’s. There is an opportunity to increase awareness and utilization of the SDxT by 9-1-1 centers. Future research should focus on identifying barriers to stroke recognition and provide additional tools to address the real-life challenges our 9-1-1 communications specialists face when rapidly assessing for an acute stroke over the phone.

BackgroundAccurate dispatch prioritization for emergency medical services (EMS) is essential for optimizing resource allocation and ensuring timely emergency response. In the Province of Quebec, Canada, a locally adapted dispatch system was implemented using the standardized codes of the Medical Priority Dispatch System (MPDS) but with regional priority definitions. Despite periodic reviews, the system’s performance has not been formally assessed. This study evaluates the effectiveness of this prioritization system by comparing priority levels assigned at call-taking with on-scene paramedic assessments and by examining how the system’s performance has evolved over three years and across chief complaints.MethodsIn this retrospective observational study, we analyzed EMS dispatches in the Capitale-Nationale administrative region of the Province of Quebec, Canada, between July 15 and December 15 over three consecutive years (2021, 2022, and 2023). We assessed system performance using sensitivity, specificity, overtriage, undertriage, predictive values, and accuracy. Statistical analyses included chi-square tests for priority consistency and pairwise t-tests for performance changes over time. Additionally, we examined variations across chief complaints to identify high overtriage and undertriage medical conditions.ResultsThis study analyzed 96,099 EMS dispatches over a three-year period. While 61.8% of these dispatches were classified as urgent at call-taking, paramedics later determined that 79.7% of all cases were stable and required non-urgent transport, indicating a high level of overtriage. Conditions such as abdominal pain, falls, and psychiatric issues were the chief complaints that showed high overtriage rates (> 90%), whereas allergic reactions, diabetic problems, and heart conditions had the highest undertriage rates (> 10%). Over the three-year period, priority modifications led to a 2.5% decrease in undertriage but a 3.7% increase in overtriage (p < 0.05), highlighting the ongoing challenge of balancing accuracy with an adequate response in dispatch prioritization.ConclusionThe studied prioritization system effectively identifies non-urgent dispatches but exhibits a high overtriage rate, which strains EMS resources. The recent priority modifications further increased overtriage, underscoring the challenge of balancing resource allocation with timely intervention. Refining dispatch criteria and integrating secondary triage or AI-based decision support could potentially improve accuracy and system efficiency.

This paper analyses the recent evolution of the electricity price of one of the EastEuropean countries’ Balancing Markets (BM) – Romania, aiming to understand the prices trend and predict them in the current economic and geopolitical context. This is especially important as the electricity producers have to allocate their output between wholesale electricity market, ancillary services markets and BM targeting to maximize value and achieve a sustainable economic development. Therefore, in this paper, we propose an AI-powered electricity price forecast using several types of standout Machine Learning (ML) algorithms such as classifiers and regressors to predict the electricity price on BM. This approach, consisting of two steps, identifies the imbalance sign and significantly enhances the performance of the price forecast. The proposed method offers valuable insights into the market participants’ trading opportunities using two prediction solutions. The first prediction solution consists of averaging the results of five ensemble ML algorithms. The second one consists in weighting the results of the five forecasting ML algorithms using either a linear regression or a decision tree algorithm. Thus, we propose to combine supervised and unsupervised ML algorithms and find the fundamentals for creating optimal bidding strategies for electricity market players.

Pre-hospital trauma care resources for road traffic injuries in a middle-income country—A province based study on need and access in Iran