Close to half of all hospital-based health care delivered annually in the US is performed in emergency departments (ED), and this fraction is rising [1].

Emergency medicine (EM) physicians must simultaneously care for multiple patients, many of whom are critical, making rapid access to patient and clinical information for acute decision making necessary.

At the same time, however, the human capacity to multitask and quickly synthesize information is very limited.

However, there has never been a time in history where more information is available to health care providers to make accurate diagnoses and treatments than now.

The necessary tool to assist the EM provider is artificial intelligence (AI).

While the application of AI in healthcare has made rapid inroads in many fields, such as imaging and dermatology, its use in acute care medicine has been limited, and in the emergency department, nearly non-existent.

When built with clinical workflow in mind, AI in the emergency department for augmented decision will improve care, decrease errors, and increase efficiency.

While the components are available to accomplish this, as yet EMAI does not currently exist in any meaningful way.

The Unique Nature of emergency medicine

The delivery of emergency medicine has unique challenges that make it unlike any other medical specialty and most amenable to AI.

Factors such as the unpredictability of patient influx, criticality of the conditions, 24/7 service, resource intense evaluations and more all lead a strain on providers and resources.  These in turn lead to the high-risk of human error and exhaust human capabilities.

In the ED, there are no scheduled patients, and this is reflected in the fact that most emergency departments work at or over maximum capacity. Thus, from before a patient is even seen, the pressure on work flow starts.

Statistics from the CDC reveal that only 35% of patients presenting to the ED are seen in less than 15 minutes [2]. Appropriate patient triage is an initial challenge that continues to be a struggle for every ED.

Once in the ED, length of stay (LOS) affects outcomes. The longer the ED LOS, the higher the mortality [3]. The application from the initial moment of contact allows for appropriate triage, quicker patient to physician times and shorter stays.

Another unique aspect of EM is the intensity of the work. In the ED, multiple patients must be seen simultaneously. In fact, an EM provider cares for more than three at any given moment compared to the primary care provider who sees only a single scheduled patient at a time [4].

Beyond the patient load, the average ED patient is complex. The most acutely and critically ill patients are cared for in the ED.

In addition, many patients are directed to the ED by non-ED providers because of the ability to have diagnostics are typically performed more quickly than in an outpatient setting. Unfortunately, these evaluations take time which consequentially lead to longer stays for all ED patients.

How AI impacts emergency medicine

AI augmented care uses data driven decision making to not only assist in critical care patient evaluation and help decide which diagnostics need to be done while the patient is in the ED and which ones can be done as an outpatient.

When should a patient be discharged from the ED? Although seemingly simple, this is a very complex question that AI helps answer.

Whether to admit or discharge is not always straight forward, and there are many factors that go into this decision. For example, there are numerous variables including patient social resources, time of day, day of the week, environmental factors, living situation, availability of specialists, age, comorbidities and many more.

AI integration into clinical workflow can also help discharge patients by quickly assisting in routine, time-consuming tasks such as treatment planning (e.g., prescriptions, therapy and follow-up scheduling).

The current state of EM delivery is also fragmented and inconsistent.

According to a recent study published in the Annals of Emergency Medicine, fewer than two thirds of EM physicians are EM trained, and this is most pronounced in rural areas and critical access hospitals [5].

As a result, emergency care is variable and often substandard in these ED’s. The AIEM becomes a useful clinical decision-making tool to both the EM trained and non-EM trained provider as it combines patient data with current research and accepted treatment guidelines seamlessly into the workflow.

The positive results of EMAI leads to improved standardization of care regardless of whether the hospital is urban or rural.

Additionally, transferring to a regional medical center is common in rural areas with over and under transfers being an issue. EMAI improves early decision making for these transfers.

There are several medical conditions where time is critically important and every second counts. These “time sensitive emergencies” include acute stroke, ST elevation myocardial infarction (STEMI), sepsis and trauma.

Rapid diagnosis and the ability of the ED to immediately shift limited resources impacts outcomes. The implementation of AI assistance improves this process and reduces morbidly and mortality.

References:

  1. David Marcozzi, Brendan Carr, Alisha Liferidge, Nicole Baehr, Brian Browne. Trends in the Contribution of Emergency Departments to the Provision of Health Care in the USA. International Journal of Health Services, 2017; 002073141773449 DOI: 1177/0020731417734498
  2. Rui P, Kang K. National Hospital Ambulatory Medical Care Survey: 2015 Emergency Department Summary Tables. Available from: http://www.cdc.gov/nchs/data/ahcd/nhamcs_emergency/2015_ed_web_tables.pdf
  3. Singer AJ, et al. The association between length of emergency department boarding and mortality. Acad Emerg Med. 2011;18:1324–9.
  4. Chisholm CD, et al. Work interrupted: a comparison of workplace interruptions in emergency departments and primary care offices. Ann Emerg Med. August 2001;38:146-151.
  5. Hall, M. Kennedy et al. State of the National Emergency Department Workforce: Who Provides Care Where? Annals of Emergency Medicine, 2018 Article in Press

BIO

 

artificial intelligence medicine healthcare emergency medicineRourke M. Yeakley, MD, MHA

Board certified and practicing emergency medicine physician for over 15 years. Founder of Precision EM. Medical Director, Air St. Luke’s since 2007. Boise Regional Dean for Pacific Northwest University of Health Science. Clinical faculty at the University of Washington School of Medicine. Inventor of the Avepax medical device for single dose delivery of liquid medications, vaccines and nutritionals that are mixed at the point of care. Seven patents have been issued worldwide for this device.

 

 

 

artificial intelligence healthcare medicine emergency medicine

Mayur Saxena, Ph.D

Healthcare strategy enthusiast || Passionate for artificial Intelligence implementations on noisy datasets ||Always curious about B2B technology strategy || Healthcare market and policy dynamics