Skip to main content

November/December 2024 Common Sense

Page 22

OPERATIONS MANAGEMENT SECTION

Artificial Intelligence and Emergency Medicine: Current Applications and Beyond Akiva Dym, MD MBA FAAEM

T

he concept of artificial intelligence (AI) has become ubiquitous in nearly every industry, with its implementation already beginning to transform countless industries, including healthcare. One proposed definition of AI is “the capability of a computer program to perform tasks or reasoning processes that we usually associate with intelligence in a human being.”1 The scope of artificial intelligence is wide-ranging and encompasses various subdomains, including machine learning, neural networks, natural language models (NLM), and natural language processing (NLP). In healthcare, AI is driving innovation in multiple areas, including diagnostics, treatment plans, risk stratification, and clinical care coordination. Emergency medicine, uniquely positioned at the crossroads of healthcare, is already experiencing many of the impacts of AI on clinical operations and patient care. As the function and role of AI continues to evolve, it is crucial for emergency medicine physicians to remain informed about the current applications and implications of AI, as well as potential future developments which will directly impact clinical practice and patient care.

Emergency departments across the country continue to struggle with increasing patient volumes every year, often associated with prolonged wait times, an increased number of patients left without being seen, and increased admissions and ED boarding. It is thus more important than ever that EDs function as efficiently as possible to meet these challenges. There are a multitude of complex operational tasks required for the efficient functioning of an emergency department, many of which are well suited for the application of AI.

ED Throughput System efficiency relies heavily on the appropriate throughput of patients through all steps of their care, including diagnostic testing, treatment, and disposition. Some of the major challenges to patient throughput in the ED include the performance and interpretation of diagnostic imaging, and the time to patient disposition and bed assignment. Machine learning models can utilize real-time clinical information as well as availability of radiology resources to determine the optimal prioritization for the performance and interpretation of patient imaging. This can lead to reduced times required for imaging to be performed and interpreted, and improve times to ED disposition. In addition, AI can be utilized to help with the challenge of bed management. Hospitals typically utilize a manual system of bed management, relying on a Tetris-style approach to try to find appropriate placement for admitted patients. This is an inefficient process which leads to significant time delays and contributes to the much-discussed boarding crisis. The incorporation of predictive modeling using historical discharge patterns and real-time clinical information can improve the bed management process and help reduce boarding gridlock in the ED.3

ED Triage Patient triage is one of the earliest operational processes in the ED, and one which is critical to develop an appropriate acuity-based prioritization

AI can also be utilized to improve time from workup completion to discharge, notifying physicians in real-time when ED workups are complete as well as assisting with automated discharge

Emergency Department System Operations

Emergency medicine, uniquely positioned at the crossroads of healthcare, is already experiencing many of the impacts of AI on clinical operations and patient care.”

20

of patient care. However, nursing triage is a human-driven process which may be subject to variability, as well as often prone to bottleneck during periods of high patient volume. The introduction of AI machine learning can expedite the triage process, as well as improve ESI designations and appropriate clinical assignment within an ED. Software such as Mednition’s KATE triage has demonstrated significant improvements in speed and accurate ESI designations as compared to nursing driven triage.2

COMMON SENSE NOVEMBER/DECEMBER 2024

>>


Turn static files into dynamic content formats.

Create a flipbook
November/December 2024 Common Sense by American Academy of Emergency Medicine - Issuu