This Content is Brought to you in Partnership with:

Improve patient safety in the emergency department through AI

By Dr. Daniel J. Sullivan

A 2020 study by Newman-Toker, et al. identified that 15 diseases account for about half of all serious misdiagnosis-related harms. The research focused on the “Big Three” categories of vascular events, infections, and cancers and found that 10% of patients with a “Big Three” disease are misdiagnosed.

A likely explanation for this result is that practitioners encounter these conditions infrequently, are not familiar with the variable presentations, and are prone to cognitive errors in the diagnostic process.

These diseases require a consistent, thorough evaluation that focuses on the history and exam’s diagnostic features that should prompt the practitioner to include the condition in the differential diagnosis. The authors of this study concluded succinctly that their findings allow us to “target diagnostic improvement initiatives to diseases with the highest error and harm rates.”

Additional studies outline the failure to diagnose profile demonstrates omission of elements in history taking, physical exams, and medical decision-making.

Reducing diagnostic error

Leveraging 30 years of data into the root causes of the failure to diagnose, Nuance partnered with The Sullivan Group to develop ED Guidance for Dragon Medical Advisor, an AI-based decision support tool. ED Guidance for Dragon Medical Advisor specifically targets the most common diagnosis-related errors.

ED Guidance for Dragon Medical Advisor drives clinical alignment around elements in the history, physical exam, and medical decision-making that improve diagnostic certainty. Additionally, the algorithms help identify possible high-risk conditions and passively notifies the practitioner of a “Risk Identified.” Importantly, ED Guidance for Dragon Medical Advisor also fits comfortably within the practitioner workflow. This real-time clinical feedback helps physicians avoid medical errors before they become an adverse event or malpractice claim.

In emergency medicine

According to industry statistics, failure to diagnose accounts for 60% of medical errors and malpractice claims in emergency medicine, with 76% of ED physicians stating they experience diagnostic uncertainty on a daily basis.

Regarding patient safety, failure to evaluate and diagnose potential adverse events is driving high rates of medical errors and malpractice claims in the emergency department. A recent study from Johns Hopkins highlights the challenges around failure to diagnose and evaluate, and concludes that when focus and investment are applied to this challenge, lower error and harm rates are seen.

The advice and resources provided by ED Guidance are focused on the most common and impactful patient safety and risk issues in emergency medicine.

For the last 30 years, I have taught risk and safety to 4,000 medical directors and tens of thousands of emergency practitioners. While I enjoy lecturing, unfortunately, it does not lead to a sustained change in clinical practice that keeps patients and providers safe. Utilizing AI allows us to bring these risk and safety elements to the bedside. Since wrong or delayed diagnoses cause more wrong serious harm to patients than any other type of medical error, leveraging AI is the most promising way to reduce diagnostic error.


Dr. Daniel J. Sullivan, Founder and CEO, The Sullivan Group