Dataset Information

Artificial intelligence algorithm to predict the need for critical care in prehospital emergency medical services.

ABSTRACT: BACKGROUND:In emergency medical services (EMSs), accurately predicting the severity of a patient's medical condition is important for the early identification of those who are vulnerable and at high-risk. In this study, we developed and validated an artificial intelligence (AI) algorithm based on deep learning to predict the need for critical care during EMS. METHODS:We conducted a retrospective observation cohort study. The algorithm was established using development data from the Korean national emergency department information system, which were collected during visits in real time from 151 emergency departments (EDs). We validated the algorithm using EMS run sheets from two EDs. The study subjects comprised adult patients who visited EDs. The endpoint was critical care, and we used age, sex, chief complaint, symptom onset to arrival time, trauma, and initial vital signs as the predicted variables. RESULTS:The number of patients in the development data was 8,981,181, and the validation data comprised 2604 EMS run sheets from two hospitals. The area under the receiver operating characteristic curve of the algorithm to predict the critical care was 0.867 (95% confidence interval, [0.864-0.871]). This result outperformed the Emergency Severity Index (0.839 [0.831-0.846]), Korean Triage and Acuity System (0.824 [0.815-0.832]), National Early Warning Score (0.741 [0.734-0.748]), and Modified Early Warning Score (0.696 [0.691-0.699]). CONCLUSIONS:The AI algorithm accurately predicted the need for the critical care of patients using information during EMS and outperformed the conventional triage tools and early warning scores.

SUBMITTER: Kang DY

PROVIDER: S-EPMC7057604 | biostudies-literature | 2020 Mar

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Artificial intelligence algorithm to predict the need for critical care in prehospital emergency medical services.

Kang Da-Young DY Cho Kyung-Jae KJ Kwon Oyeon O Kwon Joon-Myoung JM Jeon Ki-Hyun KH Park Hyunho H Lee Yeha Y Park Jinsik J Oh Byung-Hee BH

Scandinavian journal of trauma, resuscitation and emergency medicine 20200304 1

<h4>Background</h4>In emergency medical services (EMSs), accurately predicting the severity of a patient's medical condition is important for the early identification of those who are vulnerable and at high-risk. In this study, we developed and validated an artificial intelligence (AI) algorithm based on deep learning to predict the need for critical care during EMS.<h4>Methods</h4>We conducted a retrospective observation cohort study. The algorithm was established using development data from th ...[more]

PMID: 32131867

Similar Datasets

Project description:BACKGROUND: As the United States embraces electronic health records (EHRs), improved emergency medical services (EMS) information systems are also a priority; however, little is known about the experiences of EMS agencies as they adopt and implement electronic patient care report (e-PCR) systems. We sought to characterize motivations for adoption of e-PCR systems, challenges associated with adoption and implementation, and emerging implementation strategies. METHODS: We conducted a qualitative study using semi-structured in-depth interviews with EMS agency leaders. Participants were recruited through a web-based survey of National Association of EMS Physicians (NAEMSP) members, a didactic session at the 2010 NAEMSP Annual Meeting, and snowball sampling. Interviews lasted approximately 30 minutes, were recorded and professionally transcribed. Analysis was conducted by a five-person team, employing the constant comparative method to identify recurrent themes. RESULTS: Twenty-three interviewees represented 20 EMS agencies from the United States and Canada; 14 EMS agencies were currently using e-PCR systems. The primary reason for adoption was the potential for e-PCR systems to support quality assurance efforts. Challenges to e-PCR system adoption included those common to any health information technology project, as well as challenges unique to the prehospital setting, including: fear of increased ambulance run times leading to decreased ambulance availability, difficulty integrating with existing hospital information systems, and unfunded mandates requiring adoption of e-PCR systems. Three recurring strategies emerged to improve e-PCR system adoption and implementation: 1) identify creative funding sources; 2) leverage regional health information organizations; and 3) build internal information technology capacity. CONCLUSION: EMS agencies are highly motivated to adopt e-PCR systems to support quality assurance efforts; however, adoption and implementation of e-PCR systems has been challenging for many. Emerging strategies from EMS agencies and others that have successfully implemented EHRs may be useful in expanding e-PCR system use and facilitating this transition for other EMS agencies.

Project description:BackgroundPrehospital obstetric events encountered by emergency medical services (EMS) can be high-risk patient presentations for which suboptimal care can cause substantial morbidity and mortality. The frequency of prehospital obstetric events is unclear because existing descriptions have reported obstetric and gynecological conditions together, without delineating specific patient presentations. Our objective was to identify the types, frequency, and acuity of prehospital obstetric events treated by EMS personnel in the US.MethodsWe conducted a cross-sectional analysis of EMS patient care records in the 2018 National EMS Information System dataset (n=22,532,890). We focused on EMS activations (i.e., calls for service) for an emergency scene response for patients aged 12-50 years with evidence of an obstetric event. Type of obstetric event was determined by examining patient symptoms, the treating EMS provider's impression (i.e., field diagnosis), and procedures performed. High patient acuity was ascertained by EMS documentation of patient status and application of the modified early obstetric warning system (MEOWS) criteria, with concordance assessed using Cohen's kappa. Descriptive statistics were calculated to describe the primary symptoms, impressions, and frequency of each type of obstetric event among these activations.ResultsA total of 107,771 (0.6%) of EMS emergency activations were identified as involving an obstetric event. The most common presentation was early or threatened labor (15%). Abdominal complaints, including pain and other digestive/abdomen signs and symptoms, was the most common primary symptom (29%) and primary impression (18%). We identified 3,489 (3%) out-of-hospital deliveries, of which 1,504 were preterm. Overall, EMS providers documented 34% of patients as being high acuity, similar to the MEOWS criteria (35%); however, there were high rates of missing data for EMS documented acuity (19%), poor concordance between the two measures (Cohen's kappa=0.12), and acuity differences for specific conditions (e.g., high acuity of non-cephalic presentations, 77% in EMS documentation versus 53% identified by MEOWS).ConclusionPrehospital obstetric events were infrequently encountered by EMS personnel, and about one-third were high acuity. Additional work to understand the epidemiology and clinical care of these patients by EMS would help to optimize prehospital care and outcomes.

Project description:BackgroundEmergency Care has previously been identified as an area of significant concern regarding the prevalence of Adverse Events (AEs). However, the majority of this focus has been on the in-hospital setting, with little understanding of the identification and incidence of AEs in the prehospital environment.MethodThe early development and testing of Emergency Medical Services (EMS) specific triggers for the identification of AEs and Harm has been previously described. To operationalise the Emergency Medical Services Trigger Tool (EMSTT), the processes developed by the Institute for Healthcare Improvement for use with the Global Trigger Tool were adapted to a prehospital emergency care setting. These were then applied using a stepwise approach to the analysis of 36 consecutive samples of patient care records over an 18-month period (n =?710). Inter-rater reliability was measured for each trigger item and level of Harm classification. Total Triggers per 10,000 Patient Encounters, AEs per 10,000 Patient Encounters and Harm per 10,000 Patient Encounters were measured. All measures were plotted on Statistical Process Control Charts.ResultsThere was a high level of inter-rater agreement across all items (range: 85.6-100%). The EMSTT found an average rate of 8.20 Triggers per 10,000 Patient Encounters, 2.48 AEs per 10,000 Patient Encounters and 0.34 Harm events per 10,000 Patient Encounters. Three triggers: Change in Systolic Blood Pressure Greater Than 20%; Temp?>?38 °C without subsequent reduction; and SpO2 <?94% without supplemental Oxygen or SpO2 <?85% without assisted ventilation accounted for 93% (n =?180) of the triggers found throughout the longitudinal analysis.DiscussionWith sufficient focus on implementation and data collection, as well as the inclusion of a contextually relevant system for classifying AE/Harm, the EMSTT represents a potentially successful strategy towards identifying the rate of AEs within EMS across a large patient population with limited commitment of time and resources.

Dataset Information

Artificial intelligence algorithm to predict the need for critical care in prehospital emergency medical services.

Publications

Artificial intelligence algorithm to predict the need for critical care in prehospital emergency medical services.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure