Care delays in any hospital department can add unnecessary costs, decrease patient satisfaction, and, at worst, pose safety risks.
Paradoxically, imaging tests and radiology interpretations—which are necessary to support clinical decision-making and prevent errors—
can also lead to care delays, increased costs, reduced satisfaction and potential patient harm.
The quest for both urgency and efficiency is one of the reasons that ultrasound with artificial intelligence (AI)-driven interpretation is becoming more common in point-of-care settings such as the emergency department (ED), intensive care units (ICUs) and with emergency medical services (EMS) professionals. AI-driven ultrasound analysis in these critical care settings is an important clinical decision support tool, especially for providers with limited experience interpreting ultrasounds. The technology also helps overcome providers’ urgency challenges by reducing evaluation time and shortening the gaps that can occur in diagnosis and treatment.
Experience matters with portable ultrasound
Not all physicians, even those in critical care settings, are experienced in interpreting ultrasound imaging. Physicians routinely need to consult with a radiologist or request that a specialist (such as an echocardiologist, for cardiac ultrasound evaluation) perform the test. Yet, 74 percent of ED directors participating in a survey reported experiencing on-call problems with specialist physician coverage, which means longer waits for patients and higher costs for hospitals. Studies also have shown that patients who require imaging evaluations have longer ED visits and are 4.4 times more likely to remain there for more than four hours when imaging is involved.
To safely avoid delays during a potential cardiac issue, ultrasound evaluations in point-of-care settings have become an essential intervention. An ejection fraction (EF) evaluation, for example, is a key diagnostic criterion driving various treatment strategies. Yet nearly all point-of-care physicians, even most echocardiologists and radiologists, interpret the EF study mainly through visual estimation, which is subjective and relies mostly on their training and experience. AI-driven technology can fill some of those experience gaps and offer an objective and reliable consultative view for physicians with varying levels of specialized training.
By leveraging clinically validated machine-learning tools and portable ultrasound devices, physicians can gain a faster quantitative understanding of a patient’s condition that results in more informed clinical decisions. AI-powered and automated technology can also operate in the low-memory and processing-power environments of mobile-based ultrasound equipment, making the devices more practical in a wider variety of inpatient and outpatient settings.
Care delays yielding poorer outcomes
Chest pain sends 7 million Americans to EDs each year and about half of those patients are admitted for further observation, testing or treatment. Delaying the admission decision can be expensive, considering personnel costs per patient bed-hour in the ED are $58.20, but only $24.80 on an inpatient floor. Another study at one hospital found that delaying admission from the ED increased length of stay by 2,183 days for all patients in just one year and increased costs by $2.1 million.
To safely advance the care journey, cardiac ultrasound analysis with AI has become a fundamental instrument at the point-of-care for initial emergency treatment, diagnostic and triage decisions. Some of the latest AI tools that physicians are tapping into include image-processing technology with advanced pattern recognition and machine-learning algorithms. This means the technology is programmed to automatically imitate the way the human eye identifies borders and motion to produce relevant and actionable data and scoring on left ventricle function.
By substantially reducing the variability of the ultrasound evaluations, emergency and other critical care providers can more quickly support their decisions with real-time, accurate and objective data to evaluate heart function and formulate the next steps in the care plan.
AI-powered technology offers life-saving insights
AI-powered cardiac ultrasound decision-support technology has numerous potentially life-saving applications. For example, consider a 76-year-old man who presents in the ED with chest pain and swollen legs. This could be due to a blood clot, kidney problem, heart problem or many other reasons.
An AI-powered ultrasound analysis application that provides a fast and accurate EF evaluation gives the clinician the necessary objective information and support for a differential diagnosis. Once the source of the problem is identified—in this case, heart failure—an appropriate treatment is initiated. This clinical decision support tool helps saves evaluation time and may prevent unnecessary costly and longer-lasting diagnostic tests.
Using only visual estimation to determine EF in this scenario requires a clinician with deep experience in interpreting cardiac ultrasound images. It is a more time-consuming process that potentially exposes the patient to greater risk and discomfort. An AI-driven tool supports such critical care scenarios and gives a broader range of clinicians additional objective instruments in their toolbox to help them decide what to do next with greater confidence at the point-of-care and more promptly advance the patient toward effective treatment.
A fully AI-automated clinical decision support tool that can quantitatively evaluate the heart’s left ventricle function can also be valuable in situations such as cardiac arrest after CPR is delivered, and in patients with symptoms such as chest pain, shortness of breath, shock, low blood pressure and heart palpitations and arrhythmias.
Enhancing patient safety and improving outcomes
Medicine is and will always be an art and science, and one that requires collaboration with other skilled and experienced physicians and health professionals. That is why AI will never replace point-of-care clinicians, as their expertise extends far beyond image interpretation. Likewise, one of the most encouraging aspects of AI and machine-learning in healthcare is that it empowers clinicians and other specialists to focus on the work that calls upon their vast experience, training and skill sets while also supporting their decision-making with objective data.
By using AI-powered ultrasound analysis tools, hospitals and health systems can more efficiently and accurately diagnose and decide on treatments or next steps. Care can then progress to the next step, whether that is additional testing, medication, or observation and discharge. Regardless, an effective decision reached more rapidly will increase patient satisfaction and improve outcomes while also controlling costs for the organization.
About the author:
Hila Goldman-Aslan is co-founder and chief executive officer of DiA Imaging Analysis Ltd., an AI-powered ultrasound analysis software company providing fully automated, turnkey decision-support tools that enable quick, objective and accurate imaging analysis, with an initial focus on cardiac ultrasound. DiA will be offering demos of its AI technology at the Machine Learning Showcase (Booth #8168) at the Radiological Society of North America (RSNA) Annual Meeting from November 25-29 in Chicago.