AI-ECHO RCT
Improving workflow of sonographers using Artificial Intelligence-based automated ECHOcardiographic measurements and the workflow of sonographers (AI-ECHO): Randomized Crossover Trial [Late Breaking Science, AHA 2024]
A research team from Juntendo University, led by Dr. Nobuyuki Kagiyama, recently presented their findings from the Randomized Crossover Trial: Artificial Intelligence-based Automated Echocardiographic Measurements and the Workflow of Sonographers (AI-ECHO) at the AHA Late-Breaking Science Session.
Dr. Nobuyuki Kagiyama, an Associate Professor at Juntendo University, is a renowned medical doctor and clinician-scientist, recognized for his commitment to advancing artificial intelligence in medicine. With a deep passion for AI research in healthcare, Dr. Kagiyama is dedicated to designing and implementing innovative AI applications that enhance patient care and transform clinical workflows.
We had the opportunity to speak with Dr. Kagiyama ahead of his presentation at AHA 2024, where he shared insights on the study, offering valuable perspectives on AI’s impact in clinical practice.
—––––––––––––––––
The primary endpoint of your study was the number of echocardiography exams performed per day. On completion of the study, the results showed that AI assistance helped increase the average number of echo exams conducted per day, from approximately 14 exams on non-AI-assisted days to 16 on AI-assisted days. While some may consider this increase modest, could you elaborate on the broader impact of AI in echocardiography, in terms of time and volume efficiency in relation to patient experience?
I wouldn’t consider this increase modest. If each sonographer can perform two additional echocardiograms per day, it translates to 10 extra exams with a team of five, significantly reducing echocardiography wait times in our hospital. Previously, long waiting lists often required sonographers to work overtime. By shortening examination times, the AI system alleviates the need for sonographers to work extended hours, enabling them to focus on more meaningful tasks, such as discussing hemodynamics with colleagues or engaging with patients to better understand their conditions.
Overwork and burnout remain pressing issues for sonographers in both Japan and the U.S. I am optimistic that AI-driven efficiency improvements can not only enhance productivity but also make sonographer’s roles more fulfilling, ultimately supporting their motivation and well-being in the long term.
What were some of the key AI features that you felt helped streamline the examination process and reduce overall examination time, and how does this impact patient experience or clinical efficiency?
The automatic measurements performed by AI are notably faster than manual measurements, significantly enhancing the efficiency of echocardiographic examinations. One notable benefit for patients is the reduced duration of image acquisition, resulting in less time spent actively scanning. Prolonged bedside examinations can sometimes cause anxiety for patients, as they may worry that extended scanning indicates a serious heart condition. Furthermore, the experience of having their upper body exposed for an extended period can be uncomfortable. By shortening scanning times, AI contributes to a more positive patient experience, making the procedure both more comfortable and less distressing.
How valuable do you find the ability to collect an expanded range of measurements from patient echocardiograms?
Over the past few decades, the introduction of numerous new echocardiographic parameters has significantly expanded the scope of measurements required per examination. Given the increasing volume of exams, it would be unrealistic to expect sonographers to perform additional measurements without increasing their workload and stress. Consequently, many echocardiography labs are unable to routinely incorporate advanced and clinically valuable parameters, such as strain analysis, into every case, despite their well-documented benefits.
A comprehensive AI system, such as Us2.ai, addresses this challenge by providing automated measurements, including strain analysis, as a standard part of every examination. This automated expansion of parameters enables more precise diagnoses and deeper insights into patient pathology, ultimately enhancing the quality of patient care.
Were there any results from your study that you found unexpected or surprising, and yet very impactful and beneficial?
One of the most fascinating findings from our study was that the quality of echocardiographic images, as assessed by blinded observers, improved with the implementation of AI. This outcome was somewhat unexpected; however, I believe a key reason is that AI allows sonographers to concentrate more on image acquisition rather than measurements. Another contributing factor is that sonographers became increasingly aware of the impact of image quality on AI performance, which encouraged them to capture clearer images.
When sonographers perform measurements manually, they may sometimes accept “adequate” image quality, knowing they can adjust tracings as necessary. With AI, however, optimal image quality becomes essential, and this awareness appears to foster a stronger commitment to obtaining the highest-quality images possible. From a diagnostic standpoint, this improvement in image quality is immensely beneficial, as it enhances the precision of echocardiographic interpretation, ultimately supporting improved clinical outcomes.
Can you elaborate on how AI facilitates a more uniform and standardized workflow for sonographers? Did it address some of the challenges related to workflow standardization prior to the integration of AI?
Before implementing AI, we took several measures to standardize workflows and ensure accuracy within our hospital. For instance, we prepared reference images and tracings for different ranges of left ventricular ejection fraction (LVEF), reviewed them as a team at regular intervals, and conducted periodic training sessions focused on measurement techniques. Additionally, an echocardiography specialist will review all scans, providing feedback on any clearly incorrect tracings.
However, even among experienced echocardiography specialists, individual variations in approach are inevitable. Given the large volume of cases reviewed daily, it is unrealistic to offer detailed feedback on every examination. AI provides a consistent “gold standard” that can be uniformly applied to all examinations, significantly reducing variability between sonographers. This constant reference standard has proven to be a valuable tool in promoting a more uniform, reliable workflow and minimizing discrepancies across the team.
Has your research demonstrated the anticipated benefits of incorporating AI into your clinical practice?
Yes! While we initially projected a 20% improvement in examination efficiency, the results fell slightly short of this target. However, we still observed clear clinical utility. Additionally, our secondary outcomes, particularly a significant reduction in mental fatigue among sonographers and notable improvements in image quality, were highly impactful from a clinical perspective. These findings are critical for maintaining a high standard of care in clinical practice.
Access Dr. Kagiyama’s AHA presentation slides and view the exclusive interview below.
Read more here about Dr. Kagiyama’s experiences integrating Us2.ai into his echo lab at Juntendo University.
