Using natural language processing in emergency medicine health service research: A systematic review and meta-analysis.

OBJECTIVES: Natural language processing (NLP) represents one of the adjunct technologies within artificial intelligence and machine learning, creating structure out of unstructured data. This study aims to assess the performance of employing NLP to identify and categorize unstructured data within the emergency medicine (EM) setting. METHODS: We systematically searched publications related to EM research and NLP across databases including MEDLINE, Embase, Scopus, CENTRAL, and ProQuest Dissertations & Theses Global. Independent reviewers screened, reviewed, and evaluated article quality and bias. NLP usage was categorized into syndromic surveillance, radiologic interpretation, and identification of specific diseases/events/syndromes, with respective sensitivity analysis reported. Performance metrics for NLP usage were calculated and the overall area under the summary of receiver operating characteristic curve (SROC) was determined. RESULTS: A total of 27 studies underwent meta-analysis. Findings indicated an overall mean sensitivity (recall) of 82%-87%, specificity of 95%, with the area under the SROC at 0.96 (95% CI 0.94-0.98). Optimal performance using NLP was observed in radiologic interpretation, demonstrating an overall mean sensitivity of 93% and specificity of 96%. CONCLUSIONS: Our analysis revealed a generally favorable performance accuracy in using NLP within EM research, particularly in the realm of radiologic interpretation. Consequently, we advocate for the adoption of NLP-based research to augment EM health care management.

Multi-Task Learning for Post-transplant Cause of Death Analysis: A Case Study on Liver Transplant.

Organ transplant is the essential treatment method for some end-stage diseases, such as liver failure. Analyzing the post-transplant cause of death (CoD) after organ transplant provides a powerful tool for clinical decision making, including personalized treatment and organ allocation. However, traditional methods like Model for End-stage Liver Disease (MELD) score and conventional machine learning (ML) methods are limited in CoD analysis due to two major data and model-related challenges. To address this, we propose a novel framework called CoD-MTL leveraging multi-task learning to model the semantic relationships between various CoD prediction tasks jointly. Specifically, we develop a novel tree distillation strategy for multi-task learning, which combines the strength of both the tree model and multi-task learning. Experimental results are presented to show the precise and reliable CoD predictions of our framework. A case study is conducted to demonstrate the clinical importance of our method in the liver transplant.

Fast does not imply flawed: Analyzing emergency physician productivity and medical errors.

Objective: To determine whether emergency physician productivity is associated with the risk of medical errors. Methods: We retrospectively analyzed quality assurance (QA) and billing data over 3 years at 2 urban emergency departments. Faculty physicians working 400 hours or more at either site were included. We measured physician years of experience, age, gender, patients seen per hour (PPH), and relative value units billed per hour (RVU/h). From an established QA process, we obtained adjudicated medical errors to calculate rates of medical errors per 1000 patients seen as the outcome. We discretized numeric variables and used Kruskal-Wallis testing to examine relationships between independent variables and rates of medical errors. Results: We included data for 39 physicians at site A and 42 at site B. The median rate of errors per 1000 patients was 1.6 (interquartile range [IQR], 1.1-1.9) at site A and 3.3 (IQR, 2.4-3.9) at site B. At site A, RVU/h was associated with error rates (P = 0.03), with medians of 2.0, 1.2, 1.7, and 1.3 errors per 1000 patients, from slowest to fastest quartiles. At site B, PPH was associated with error rates (P < 0.01), with medians of 3.9, 3.7, 2.4, and 2.7 errors per 1000 patients, from slowest to fastest quartiles. There was no significant relationship between error rates and PPH at site A or RVU/h at site B. Conclusions: Rates of medical errors were associated with 1 metric of physician productivity at each site, with higher error rates seen among physicians with slower productivity.

Does crowding influence emergency department treatment time and disposition?

OBJECTIVE: The objective of this study was to determine whether crowding influences treatment times and disposition decisions for emergency department (ED) patients. METHODS: We conducted a retrospective cohort study at 2 hospitals from January 1, 2014, to July 1, 2014. Adult ED visits with dispositions of discharge, admission, or transfer were included. Treatment times were modeled by linear regression with log-transformation; disposition decisions (admission or transfer vs discharge) were modeled by logistic regression. Both models adjusted for chief complaint, Emergency Severity Index (ESI), and 4 crowding metrics in quartiles: waiting count, treatment count, boarding count, and National Emergency Department Overcrowding Scale. RESULTS: We included 21,382 visits at site A (12.9% excluded) and 29,193 at site B (15.0% excluded). Respective quartiles of treatment count increased treatment times by 7.1%, 10.5%, and 13.3% at site A (P < 0.001) and by 4.0%, 6.5%, and 10.2% at site B (P < 0.001). The fourth quartile of treatment count increased estimates of treatment time for patients with chest pain and ESI level 2 from 2.5 to 2.9 hours at site A (20 minutes) and from 3.0 to 3.3 hours at site B (18 minutes). Treatment times decreased with quartiles of waiting count by 5.6%, 7.2%, and 7.3% at site B (P < 0.001). Odds of admission or transfer increased with quartiles of waiting count by 8.7%, 9.6%, and 20.3% at site A (P = 0.011) and for the third (11.7%) and fourth quartiles (27.3%) at site B (P < 0.001). CONCLUSIONS: Local crowding influenced ED treatment times and disposition decisions at 2 hospitals after adjusting for chief complaint and ESI.

Understanding Emergency Care Delivery Through Computer Simulation Modeling.

In 2017, Academic Emergency Medicine convened a consensus conference entitled, "Catalyzing System Change through Health Care Simulation: Systems, Competency, and Outcomes." This article, a product of the breakout session on "understanding complex interactions through systems modeling," explores the role that computer simulation modeling can and should play in research and development of emergency care delivery systems. This article discusses areas central to the use of computer simulation modeling in emergency care research. The four central approaches to computer simulation modeling are described (Monte Carlo simulation, system dynamics modeling, discrete-event simulation, and agent-based simulation), along with problems amenable to their use and relevant examples to emergency care. Also discussed is an introduction to available software modeling platforms and how to explore their use for research, along with a research agenda for computer simulation modeling. Through this article, our goal is to enhance adoption of computer simulation, a set of methods that hold great promise in addressing emergency care organization and design challenges.

Dashboard visualizations: Supporting real-time throughput decision-making.

Providing timely and effective care in the emergency department (ED) requires the management of individual patients as well as the flow and demands of the entire department. Strategic changes to work processes, such as adding a flow coordination nurse or a physician in triage, have demonstrated improvements in throughput times. However, such global strategic changes do not address the real-time, often opportunistic workflow decisions of individual clinicians in the ED. We believe that real-time representation of the status of the entire emergency department and each patient within it through information visualizations will better support clinical decision-making in-the-moment and provide for rapid intervention to improve ED flow. This notion is based on previous work where we found that clinicians' workflow decisions were often based on an in-the-moment local perspective, rather than a global perspective. Here, we discuss the challenges of designing and implementing visualizations for ED through a discussion of the development of our prototype Throughput Dashboard and the potential it holds for supporting real-time decision-making.

Factors Associated With the Likelihood of Hospitalization Following Emergency Department Visits for Behavioral Health Conditions.

OBJECTIVES: Behavioral health-related emergency department (ED) visits have been linked with ED overcrowding, an increased demand on limited resources, and a longer length of stay (LOS) due in part to patients being admitted to the hospital but waiting for an inpatient bed. This study examines factors associated with the likelihood of hospital admission for ED patients with behavioral health conditions at 16 hospital-based EDs in a large urban area in the southern United States. METHODS: Using Andersen's Behavioral Model of Health Service Use for guidance, the study examined the relationship between predisposing (characteristics of the individual, i.e., age, sex, race/ethnicity), enabling (system or structural factors affecting healthcare access), and need (clinical) factors and the likelihood of hospitalization following ED visits for behavioral health conditions (n = 28,716 ED visits). In the adjusted analysis, a logistic fixed-effects model with blockwise entry was used to estimate the relative importance of predisposing, enabling, and need variables added separately as blocks while controlling for variation in unobserved hospital-specific practices across hospitals and time in years. RESULTS: Significant predisposing factors associated with an increased likelihood of hospitalization following an ED visit included increasing age, while African American race was associated with a lower likelihood of hospitalization. Among enabling factors, arrival by emergency transport and a longer ED LOS were associated with a greater likelihood of hospitalization while being uninsured and the availability of community-based behavioral health services within 5 miles of the ED were associated with lower odds. Among need factors, having a discharge diagnosis of schizophrenia/psychotic spectrum disorder, an affective disorder, a personality disorder, dementia, or an impulse control disorder as well as secondary diagnoses of suicidal ideation and/or suicidal behavior increased the likelihood of hospitalization following an ED visit. CONCLUSION: The block of enabling factors was the strongest predictor of hospitalization following an ED visit compared to predisposing and need factors. Our findings also provide evidence of disparities in hospitalization of the uninsured and racial and ethnic minority patients with ED visits for behavioral health conditions. Thus, improved access to community-based behavioral health services and an increased capacity for inpatient psychiatric hospitals for treating indigent patients may be needed to improve the efficiency of ED services in our region for patients with behavioral health conditions. Among need factors, a discharge diagnosis of schizophrenia/psychotic spectrum disorder, an affective disorder, a personality disorder, an impulse control disorder, or dementia as well as secondary diagnoses of suicidal ideation and/or suicidal behavior increased the likelihood of hospitalization following an ED visit, also suggesting an opportunity for improving the efficiency of ED care through the provision of psychiatric services to stabilize and treat patients with serious mental illness.

Acetylfentanyl: An Emerging Drug of Abuse.

BACKGROUND: Opioid analgesics are widely used in health care, yet have significant potential for abuse. High doses are associated with potentially fatal respiratory depression, which caused 21,314 deaths in the United States in 2011. Acetylfentanyl, a synthetic opioid agonist closely related to fentanyl, recently emerged as a drug of abuse linked to numerous deaths in North America. CASE REPORT: A 36-year-old male developed the habit of using a propylene glycol electronic cigarette filled with acetylfentanyl to aid relaxation. He purchased the drug online in a manner that appeared legal to him, which compromised his insight about the danger of the substance. He had been using the e-cigarette with increasing frequency while on medical leave, and his wife reported finding him weakly responsive on more than one occasion. At approximately 3 am, the family activated 911 for altered mental status. His presentation included respiratory depression, pinpoint pupils, hypoxemia, and a Glasgow Coma Scale score of 6. He responded to serial doses of intravenous naloxone with improvement in his mental status and respiratory condition. Due to the need for repeated dosing, he was placed on a naloxone infusion and recovered uneventfully in intensive care. WHY SHOULD AN EMERGENCY PHYSICIAN BE AWARE OF THIS?: Complications from emerging drugs of abuse, like acetylfentanyl, frequently present first to emergency departments. Prompt recognition and treatment can help avoid morbidity and mortality. Acetylfentanyl can be managed effectively with naloxone, although higher than conventional dosing may be required to achieve therapeutic effect.

Voluntary Medical Incident Reporting Tool to Improve Physician Reporting of Medical Errors in an Emergency Department.

INTRODUCTION: Medical errors are frequently under-reported, yet their appropriate analysis, coupled with remediation, is essential for continuous quality improvement. The emergency department (ED) is recognized as a complex and chaotic environment prone to errors. In this paper, we describe the design and implementation of a web-based ED-specific incident reporting system using an iterative process. METHODS: A web-based, password-protected tool was developed by members of a quality assurance committee for ED providers to report incidents that they believe could impact patient safety. RESULTS: The utilization of this system in one residency program with two academic sites resulted in an increase from 81 reported incidents in 2009, the first year of use, to 561 reported incidents in 2012. This is an increase in rate of reported events from 0.07% of all ED visits to 0.44% of all ED visits. In 2012, faculty reported 60% of all incidents, while residents and midlevel providers reported 24% and 16% respectively. The most commonly reported incidents were delays in care and management concerns. CONCLUSION: Error reporting frequency can be dramatically improved by using a web-based, user-friendly, voluntary, and non-punitive reporting system.

Hemodynamically unstable: accidental atenolol toxicity?

BACKGROUND: Antihypertensive medications, including ß-blockers, are widely used in patients with chronic kidney disease. Unlike most ß-blockers, atenolol is excreted primarily by the kidney, and its clearance by peritoneal dialysis is poor. These pharmacokinetic factors may predispose patients to gradual accumulation of the drug over time. OBJECTIVES: To review the management of a diagnostic dilemma, the role of glucagon therapy, and the clinical implications of atenolol clearance. CASE REPORT: A young woman with end-stage renal disease requiring peritoneal dialysis presented with sudden onset of abdominal pain and hemodynamic instability with hypotension and relative bradycardia. The patient reported that she took her regular four antihypertensive agents, including atenolol, with no excess ingestion or recent dose changes. After resuscitation and consideration of a broad differential diagnosis, the most likely cause of the patient's illness was determined to be unintentional atenolol toxicity, with secondary mesenteric ischemia due to a low-flow state that caused her abdominal pain. Glucagon therapy led to rapid correction of the patient's hemodynamic instability and pain. CONCLUSION: The unique pharmacokinetics of long-term medications must be considered in patients with impaired clearance, such as this patient with end-stage renal disease treated by peritoneal dialysis. Medications may gradually accumulate to supratherapeutic levels, which over time may lead to symptoms of significant toxicity.

An incidental finding? Pneumatosis intestinalis after minor trauma.

BACKGROUND: Pneumatosis intestinalis (PI) refers to the identification of air within the wall of the gastrointestinal tract. This finding often marks serious underlying pathology, which can be potentially surgical in nature. However, this process may also occur within a benign context, for example, in patients who are chronically immunosuppressed. The prevalence of benign PI may be greater than previously anticipated, because its discovery is facilitated by the increasingly widespread use of computed tomography (CT) scanning. OBJECTIVES: We will illustrate how widespread use of CT scanning after trauma leads to incidental findings, some of which are difficult to distinguish from acute pathologic findings. We will also discuss the differential diagnosis for PI and the associated clinical significance. CASE REPORT: A female patient with two autoimmune disorders requiring immunosuppression presented after minor trauma. Her clinical stability and benign examination led us to refrain from ordering a full radiographic evaluation, including an abdominal CT scan. She was safely discharged; however, per CT several days later, the incidental finding was made of PI with free intraperitoneal air. These findings after trauma commonly prompt an exploratory laparotomy. However, given her persistent stability, we attributed this to immunosuppression rather than to recent trauma. CONCLUSION: The indications for ordering CT scans after minor trauma must be carefully considered, and incidental findings must be interpreted in the context of the overall clinical scenario.

Forecasting emergency department crowding: an external, multicenter evaluation.

STUDY OBJECTIVE: We apply a previously described tool to forecast emergency department (ED) crowding at multiple institutions and assess its generalizability for predicting the near-future waiting count, occupancy level, and boarding count. METHODS: The ForecastED tool was validated with historical data from 5 institutions external to the development site. A sliding-window design separated the data for parameter estimation and forecast validation. Observations were sampled at consecutive 10-minute intervals during 12 months (n=52,560) at 4 sites and 10 months (n=44,064) at the fifth. Three outcome measures-the waiting count, occupancy level, and boarding count-were forecast 2, 4, 6, and 8 hours beyond each observation, and forecasts were compared with observed data at corresponding times. The reliability and calibration were measured following previously described methods. After linear calibration, the forecasting accuracy was measured with the median absolute error. RESULTS: The tool was successfully used for 5 different sites. Its forecasts were more reliable, better calibrated, and more accurate at 2 hours than at 8 hours. The reliability and calibration of the tool were similar between the original development site and external sites; the boarding count was an exception, which was less reliable at 4 of 5 sites. Some variability in accuracy existed among institutions; when forecasting 4 hours into the future, the median absolute error of the waiting count ranged between 0.6 and 3.1 patients, the median absolute error of the occupancy level ranged between 9.0% and 14.5% of beds, and the median absolute error of the boarding count ranged between 0.9 and 2.8 patients. CONCLUSION: The ForecastED tool generated potentially useful forecasts of input and throughput measures of ED crowding at 5 external sites, without modifying the underlying assumptions. Noting the limitation that this was not a real-time validation, ongoing research will focus on integrating the tool with ED information systems.

Forecasting emergency department crowding: a prospective, real-time evaluation.

OBJECTIVE: Emergency department crowding threatens quality and access to health care, and a method of accurately forecasting near-future crowding should enable novel ways to alleviate the problem. The authors sought to implement and validate the previously developed ForecastED discrete event simulation for real-time forecasting of emergency department crowding. DESIGN AND MEASUREMENTS: The authors conducted a prospective observational study during a three-month period (5/1/07-8/1/07) in the adult emergency department of a tertiary care medical center. The authors connected the forecasting tool to existing information systems to obtain real-time forecasts of operational data, updated every 10 minutes. The outcome measures included the emergency department waiting count, waiting time, occupancy level, length of stay, boarding count, boarding time, and ambulance diversion; each forecast 2, 4, 6, and 8 hours into the future. RESULTS: The authors obtained crowding forecasts at 13,239 10-minute intervals, out of 13,248 possible (99.9%). The R(2) values for predicting operational data 8 hours into the future, with 95% confidence intervals, were 0.27 (0.26, 0.29) for waiting count, 0.11 (0.10, 0.12) for waiting time, 0.57 (0.55, 0.58) for occupancy level, 0.69 (0.68, 0.70) for length of stay, 0.61 (0.59, 0.62) for boarding count, and 0.53 (0.51, 0.54) for boarding time. The area under the receiver operating characteristic curve for predicting ambulance diversion 8 hours into the future, with 95% confidence intervals, was 0.85 (0.84, 0.86). CONCLUSIONS: The ForecastED tool provides accurate forecasts of several input, throughput, and output measures of crowding up to 8 hours into the future. The real-time deployment of the system should be feasible at other emergency departments that have six patient-level variables available through information systems.

Forecasting emergency department crowding: a discrete event simulation.

STUDY OBJECTIVE: To develop a discrete event simulation of emergency department (ED) patient flow for the purpose of forecasting near-future operating conditions and to validate the forecasts with several measures of ED crowding. METHODS: We developed a discrete event simulation of patient flow with evidence from the literature. Development was purely theoretical, whereas validation involved patient data from an academic ED. The model inputs and outputs, respectively, are 6-variable descriptions of every present and future patient in the ED. We validated the model by using a sliding-window design, ensuring separation of fitting and validation data in time series. We sampled consecutive 10-minute observations during 2006 (n=52,560). The outcome measures--all forecast 2, 4, 6, and 8 hours into the future from each observation--were the waiting count, waiting time, occupancy level, length of stay, boarding count, boarding time, and ambulance diversion. Forecasting performance was assessed with Pearson's correlation, residual summary statistics, and area under the receiver operating characteristic curve. RESULTS: The correlations between crowding forecasts and actual outcomes started high and decreased gradually up to 8 hours into the future (lowest Pearson's r for waiting count=0.56; waiting time=0.49; occupancy level=0.78; length of stay=0.86; boarding count=0.79; boarding time=0.80). The residual means were unbiased for all outcomes except the boarding time. The discriminatory power for ambulance diversion remained consistently high up to 8 hours into the future (lowest area under the receiver operating characteristic curve=0.86). CONCLUSION: By modeling patient flow, rather than operational summary variables, our simulation forecasts several measures of near-future ED crowding, with various degrees of good performance.

Systematic review of emergency department crowding: causes, effects, and solutions.

Emergency department (ED) crowding represents an international crisis that may affect the quality and access of health care. We conducted a comprehensive PubMed search to identify articles that (1) studied causes, effects, or solutions of ED crowding; (2) described data collection and analysis methodology; (3) occurred in a general ED setting; and (4) focused on everyday crowding. Two independent reviewers identified the relevant articles by consensus. We applied a 5-level quality assessment tool to grade the methodology of each study. From 4,271 abstracts and 188 full-text articles, the reviewers identified 93 articles meeting the inclusion criteria. A total of 33 articles studied causes, 27 articles studied effects, and 40 articles studied solutions of ED crowding. Commonly studied causes of crowding included nonurgent visits, "frequent-flyer" patients, influenza season, inadequate staffing, inpatient boarding, and hospital bed shortages. Commonly studied effects of crowding included patient mortality, transport delays, treatment delays, ambulance diversion, patient elopement, and financial effect. Commonly studied solutions of crowding included additional personnel, observation units, hospital bed access, nonurgent referrals, ambulance diversion, destination control, crowding measures, and queuing theory. The results illustrated the complex, multifaceted characteristics of the ED crowding problem. Additional high-quality studies may provide valuable contributions toward better understanding and alleviating the daily crisis. This structured overview of the literature may help to identify future directions for the crowding research agenda.

The challenge of predicting demand for emergency department services.

OBJECTIVES: The objective was to develop methodology for predicting demand for emergency department (ED) services by characterizing ED arrivals. METHODS: One year of ED arrival data from an academic ED were merged with local climate data. ED arrival patterns were described; Poisson regression was selected to represent the count of hourly ED arrivals as a function of temporal, climatic, and patient factors. The authors evaluated the appropriateness of prediction models by whether the data met key Poisson assumptions, including variance proportional to the mean, positive skewness, and absence of autocorrelation among hours. Model accuracy was assessed by comparing predicted and observed histograms of arrival counts and by how frequently the observed hourly count fell within the 50 and 90% prediction intervals. RESULTS: Hourly ED arrivals were obtained for 8,760 study hours. Separate models were fit for high- versus low-acuity patients because of significant arrival pattern differences. The variance was approximately equal to the mean in the high- and low-acuity models. There was no residual autocorrelation (r = 0) present after controlling for temporal, climatic, and patient factors that influenced the arrival rate. The observed hourly count fell within the 50 and 90% prediction intervals 50 and 90% of the time, respectively. The observed histogram of arrival counts was nearly identical to the histogram predicted by a Poisson process. CONCLUSIONS: At this facility, demand for ED services was well approximated by a Poisson regression model. The expected arrival rate is characterized by a small number of factors and does not depend on recent numbers of arrivals.

Measuring and forecasting emergency department crowding in real time.

STUDY OBJECTIVE: We quantified the potential for monitoring current and near-future emergency department (ED) crowding by using 4 measures: the Emergency Department Work Index (EDWIN), the National Emergency Department Overcrowding Scale (NEDOCS), the Demand Value of the Real-time Emergency Analysis of Demand Indicators (READI), and the Work Score. METHODS: We calculated the 4 measures at 10-minute intervals during an 8-week study period (June 21, 2006, to August 16, 2006). Ambulance diversion status was the outcome variable for crowding, and occupancy level was the performance baseline measure. We evaluated discriminatory power for current crowding by the area under the receiver operating characteristic curve (AUC). To assess forecasting power, we applied activity monitoring operating characteristic curves, which measure the timeliness of early warnings at various false alarm rates. RESULTS: We recorded 7,948 observations during the study period. The ED was on ambulance diversion during 30% of the observations. The AUC was 0.81 for the EDWIN, 0.88 for the NEDOCS, 0.65 for the READI Demand Value, 0.90 for the Work Score, and 0.90 for occupancy level. In the activity monitoring operating characteristic analysis, only the occupancy level provided more than an hour of advance warning (median 1 hour 7 minutes) before crowding, with 1 false alarm per week. CONCLUSION: The EDWIN, the NEDOCS, and the Work Score monitor current ED crowding with high discriminatory power, although none of them exceeded the performance of occupancy level across the range of operating points. None of the measures provided substantial advance warning before crowding at low rates of false alarms.

Physician predictions of graft survival following liver transplantation.

INTRODUCTION: Due to the scarcity of cadaveric livers, clinical judgment must be used to avoid futile transplants. However, the accuracy of human judgment for predicting outcomes following liver transplantation is unknown. The study aim was to assess expert clinicians' ability to predict graft survival and to compare their performance to published survival models. MATERIALS AND METHODS: Pre-transplant case summaries were prepared based on 16 actual, randomly selected liver transplants. Clinicians specializing in the care of liver transplant patients were invited to assess the likelihood of 90-day graft survival for each case using (1) a 4-point Likert scale ranging from poor to excellent, and (2) a visual analog scale denoting the probability of survival. Four published models were also used to predict survival for the 16 cases. RESULTS. Completed instruments were received from 50 clinicians. Prognostic estimates on the two scales were highly correlated (median r=0.88). Individual clinicians' predictive ability was 0.61+/-0.13, by area under the receiver operating characteristic curve. The performance of published models was MELD 0.59, Desai 0.66, Ghobrial 0.61, and Thuluvath 0.45. For three cases, clinicians consistently overestimated the probability of survival (87+/-10%, 89+/-9%, 86+/-9%); these patients had early graft failures caused by postoperative complications. DISCUSSION. Clinicians varied in their ability to predict survival for a set of pre-transplant scenarios, but performed similarly to published models. When clinicians overestimated the chance of transplant success, either sepsis or hepatic artery thrombosis was involved; such events may be hard to predict before surgery.

Visualizing temporal patterns of demand, throughput and crowding in an emergency department.

Emergency department (ED) operational data were calculated at 10-minute intervals throughout 2006 (n = 52561) in the adult ED of an academic medical center. Several operational parameters per observation were measured to better understand temporal patterns of input, throughput, and output of medical services. This may allow for improvement of predictive models of overcrowding. Visualization of this dataset is structured by a calendar template, facilitating discovery of cyclic patterns at diurnal, weekly, and monthly scales.