Artificial intelligence (AI) has been used in elementary forms in emergency departments worldwide for decades. It shows promise in its more developed forms from prehospital to disposition.
Issues with AI
AI does not always work well. If I had a dollar for every time a patient said, “I Googled my symptoms, and I’m pretty sure I have…” I’d be a millionaire! Searching online symptom checkers has been proven inconsistent; they frequently recommend urgent and emergent care more than needed, or worse, they do not recommend care when warranted.
Putting AI to Work in the ER
AI could help link electronic health records from prehospital to those in the hospital. For example, kiosks are already set up at entrances to many primary care facilities and specialty clinics for scheduled appointments. Kiosks at the emergency department entrance with screening questions could be integrated into the triage process. Companies are marketing AI as the future of triage, and, why not? Pattern recognition and case-based reasoning are something AI is supposed to excel at. There are criticisms that the subjective nature of the triage process leads to inconsistencies in identifying patients with critical illnesses. Several studies have revealed that triage staff often under-triage critically ill patients. While chatbots and avatars may not be staffing triage desks anytime soon, it is certainly possible that AI supported technology will someday. A UCSF-led study found that a large language model AI was able to identify the more serious condition when evaluating two patients 89% of the time. In a 500-patient sub-sample, AI was correct 88% of the time, and the physician sampling was accurate at 86%. Despite being able to show good accuracy, the study ultimately found the AI was not ready for deployment. Critically, the study’s lead author noted that there are still larger issues to address and to eliminate bias from the models. “First, we need to know if it works and understand how it works, and then be careful and deliberate in how it is applied.”
Nursing Considerations: Treating AI Ethically
The ethics of AI in nursing is being championed by the American Nursing Association. Their position statement on AI notes the following: “ANA believes the appropriate use of AI in nursing practice supports and enhances the core values and ethical obligations of the profession. AI that appears to impede or diminish these core values and obligations must be avoided or incorporated only in such a way that these values and obligations are protected.”
Integrating AI in healthcare practice must not alter patient care goals. The ANA recommends that AI be adopted or integrated into nursing practice within the nursing and ethical care elements of compassion, trust, and caring so that nursing practice remains relevant in the changing landscape. Nurses must consider how AI is integrated into their practice and be conscious of how it can positively or negatively impact individual and population health outcomes. Nurses must be knowledgeable on appropriate use of AI and must be able to educate patients and families on how to use AI to dispel myths and alleviate fears.
Nowhere do speed, efficiency, and accuracy matter more than in the emergency department. AI may have the potential to improve patient outcomes, allow for better patient management, and reduce emergency room overcrowding. While most AI for emergency nursing is currently directed towards assisting at the triage level, many other areas can ultimately benefit from AI, such as with identifying and analyzing trends and anomalies in vital signs, improving communication and charting, and reducing medication errors, to name a few.
As the practice of nursing evolves, along with the use of AI in an emergency department, it will be our role as nurses to ensure its use is applied mindfully and ethically.
Author: Ricky Knack, RN, CCM
