How AI can increase the effectiveness of point-of-care ultrasounds
For patients who need care in rural areas, getting diagnostic testing can be a complicated process. In some cases, individuals may need to travel long distances to larger facilities to get access to necessary technology.
To address those issues, many developers have focused innovation efforts on point-of-care-ultrasound, or POCUS, devices.
As Dr. Mark Favot explained to Healthcare IT News, such tools – especially handheld ones – can help make care more broadly available and less expensive in remote regions.
Favot, associate professor and director of EM Ultrasound Education at the Wayne State University School of Medicine in Detroit, spoke with us about how artificial intelligence guidance can make a difference with POCUS, and shared what excites him about the technology in the longer term.
Q. Why does having access to point-of-care ultrasound devices make a difference for patient care, especially in rural areas?
A. Healthcare in rural areas often involves patients seeking medical care at facilities that are not equipped with the full array of diagnostic testing resources as large medical centers in urban population centers. Often, the physician caring for the patient has to arrange for transfer to larger hospitals in order for the patient to gain access to things like computed tomography scanning, magnetic resonance imaging and echocardiography.
Providing high-quality healthcare in a setting without these resources can be challenging for physicians and frustrating for patients.
Delays in undergoing diagnostic testing (either because the patient is being transferred to another facility, or urgent testing is deferred to a later date in an outpatient setting) not only add to the frustrations felt by both physicians and patients, but are often associated with adverse outcomes related to the disease that caused the patient to seek care in the rural setting in the first place.
Traditional ultrasound imaging that occurs in a suite in the radiology department is usually available during daytime hours, but may have limitations in terms of which parts of the body they are able to scan.
With the advent of point-of-care ultrasound, or POCUS, the imaging paradigm has shifted away from the traditional model of the images being acquired by a trained sonographer and then sent to a radiologist for interpretation to a model where the patient’s treating physician is performing and interpreting the POCUS in real time while developing a treatment plan for the patient.
Larger cart-based portable POCUS machines have been around emergency departments for the past 20 years, but they are often quite expensive – and in a low-volume healthcare setting, such as a rural ED, they may not be a financially viable solution if they are not being used often enough.
Handheld POCUS machines have further shifted the imaging paradigm because they are often affordable enough to be owned by physicians (rather than owned by hospitals) and can be brought to any care setting where that physician works. This development has important implications for patients in rural healthcare settings, because often the physicians that are staffing these hospitals do not live in the community where the hospital is located. If the hospital does not have cart-based POCUS equipment the physician can bring their own handheld POCUS machine to the hospital for their clinical shifts in the ED.
Armed with this technology, physicians can now use POCUS to aid in diagnostic testing for patients which can lead to improvements in the ability to rule in or rule out specific diagnoses for patients, rather than be forced to transfer them to centers where they may end up having completely normal imaging, only to be sent home from a hospital that is 60 miles from where they live.
Q. How can having artificial intelligence guidance make a difference with these devices?
A. Artificial intelligence has the potential to dramatically increase the effectiveness of POCUS, primarily by reducing the impact of poor confidence in image interpretation, which is one of the most common barriers to POCUS implementation.
Many physicians learned POCUS by attending one- or two-day Continuing Medical Education courses that are hosted by a large group of POCUS experts. These courses offer many advantages to the attendees, including favorable faculty-to-learner ratios, a wide variety of high-end state-of-the-art POCUS equipment, simulated patient models with excellent “windows” for ultrasound imaging and access to lectures with a large array of pathology.
The “problem” for people that attend these courses begins when they return to their own institution and now have to scan with their older, often outdated equipment, on patients that may have issues that make ultrasound imaging challenging and without the expert right there next to them coaching them on probe movements to improve the images or assisting with interpretation when things don’t look exactly like the textbook or lectures they recently attended.
Typically, if the physician works at an institution with a robust POCUS program, that ultrasound exam will get reviewed by an expert in POCUS one to three days later, and if there were issues with the exam that expert will be able to offer them feedback and suggestions for improvement. While this feedback can help that physician next time they use POCUS, it does nothing to help the patient they have just scanned, and the impact of feedback that is provided after the fact, and not in real time, is limited.
AI can be that important bridge between the expert feedback a couple days later, and no feedback whatsoever at the time of the exam. A robust AI on POCUS machines can give the user immediate actionable feedback to improve the image, such as “Tilt the probe toward the patient’s head to acquire the proper apical 4 chamber image of the heart that includes both atria and both ventricles,” rather than the image the user had acquired which only had right and left ventricle.
This type of AI feedback has obvious immediate benefits for patient care now that a proper diagnosis can be made, because the physician was able to acquire a standard, more easily interpreted POCUS image. However, the long-term impact of immediate AI feedback might be even more impactful.
Often, physicians will be enthusiastic about using POCUS in their practice immediately after undergoing training, but over time that enthusiasm wanes because their confidence level also begins to wane the further they are removed from the training session. When AI is built into POCUS systems, it can function in a role like the expert POCUS practitioner standing at the bedside. However, because it is a machine and not a human standing beside you, it creates a lower-pressure environment that makes it easier for the new POCUS user to use the machine and gradually advance their skills over time. POCUS is very humbling for physicians, and the shame of not being strong in this skill can lead physicians to avoid using it. AI is one way that this can be overcome.
Q. What excites you about the potential for POCUS in the near and long term?
A. I am most excited about what is happening at the medical-school level when it comes to the demand for POCUS. Today’s medical students are very technology-savvy, and they will not stand idly by and accept that 200-plus year-old technology, like a stethoscope, is the most effective diagnostic tool for their patients. They have and will continue to demand more out of their education.
Institutions with strong medical student POCUS curricula, like Medical University of South Carolina, Wayne State University, and University of California-Irvine, have seen increases in the numbers of applications as prospective students seek out POCUS training. Schools that do not currently have a POCUS curriculum are scrambling to catch up.
The students are driving change from the bottom up and forcing leaders to respond. The earlier that physicians can have access to POCUS training in their careers, the more likely that these skills will become durable and stay with them for the long term.
The advent of handheld POCUS systems at an affordable price point that are now marketed toward physicians rather than healthcare institutions will help to accelerate this shift. No matter how effective a POCUS curriculum or educator is, the best learning often happens when a curious physician picks up a POCUS system and learns by trial and error. When the machine is handheld and can be brought home, the physician, or the physician’s family, can become the de facto ultrasound model. All of us in the field have had this experience ourselves, and although it can be a slower and more arduous way to learn, it is a tried-and-true method that is extremely valuable.
While POCUS is an important disruptive technology that allows for more precise and more timely diagnoses by physicians and allows for safer performance of needle-guided procedures, it is important to realize that as the price point and availability of handheld POCUS equipment continues to improve, it is imperative that physicians are receiving proper training. The training should ideally begin as early as possible in one’s career, be ongoing and targeted toward areas of identified deficiencies, and be aided by modern technology such as AI.
The explosion of growth in the affordable handheld POCUS market has put these machines in the hands of a large number of new physicians, not all of whom have received adequate training. This bottom-up movement demands an appropriate response from leaders in healthcare to ensure that enough money and resources are devoted to creating and maintaining programs that train all POCUS users to adequate standards. In doing so, institutions can be at the leading edge of the POCUS wave that has already begun and harness it to positively impact patients seeking care at a wide variety of destinations.
The ability to bring high-quality imaging to any patient anywhere is the most exciting aspect of the future of POCUS and will be its long-lasting legacy.
Kat Jercich is senior editor of Healthcare IT News.
Healthcare IT News is a HIMSS Media publication.
Discover Past Posts