Exploring the Impact of AI in Point-of-Care Ultrasound:
A Conversation with Dr. Andre Kumar
Luke Baldwin, VP of Global Marketing at EchoNous, interviewed Dr. Andre Kumar, a distinguished physician, researcher, and educator at Stanford University, who is also a pioneer in POCUS research. The discussion centered around the convergence of artificial intelligence (AI) and point-of-care ultrasound (POCUS) and how this combination is opening doors to groundbreaking possibilities in the ever-evolving healthcare landscape.
[Note: This interview has been edited for clarity and brevity.]
Imagine a future where the power of artificial intelligence seamlessly integrates with the precision of medical imaging, where the boundaries between technology and medicine blur to redefine patient care. Luke Baldwin, VP of Global Marketing at EchoNous, recently sat down with Dr. Andre Kumar, a distinguished physician, researcher, and educator at Stanford University to discuss how close we are to that future.
Background and Journey in Hospital Medicine
Luke Baldwin: Hi Dr. Kumar, it’s great to chat with you again. For those that don’t know you, tell us who you are, and a bit about what you do and how you got your start as a POCUS user.
Andre Kumar, MD: My name is Dr. Andre Kumar, and I am a Clinical Associate Professor specializing in Internal Medicine at Stanford University. In addition to my clinical work, I direct our procedure service and teach point-of-care ultrasound (POCUS). I have the privilege of instructing students locally at our institution’s School of Medicine residency program and nationally through the Society of Hospital Medicine.
I first became interested in point-of-care ultrasound when I was an intern physician. One of my attending physicians brought a point-of-care ultrasound device on rounds, and it completely opened my eyes to the potential impact we can make on our patients. I became passionate about this growing field and decided to focus my career on it. I’m committed to advancing the field through research alongside my clinical and teaching responsibilities. I believe that using new technology to help us diagnose and treat patients is an important aspect of where medicine is going. I am motivated to contribute to sound research in my field.
The Role of Point-of-Care Ultrasound
Luke Baldwin: What was it about hospital medicine that caught your eye as a specialty?
Andre Kumar, MD: Hospital medicine is a fascinating field where you can care for patients with a variety of different diseases, and at the same time, you can connect with them at a time of extreme vulnerability. I also teach at a high level with residents and interns, and the variety of my practice is just wonderful. Additionally, I recognized a growing need for researchers who can pioneer and advance the field of hospital medicine. Becoming a hospitalist requires completing your residency training in internal medicine or pediatric residency.
Luke Baldwin: Can you tell us more about the process of becoming a hospitalist?
Andre Kumar, MD: Hospital medicine as a field only started in the early 1990s when specialists were recognized as needed to care for very sick patients in the hospital. To become a hospitalist, you complete your residency training in internal medicine or pediatric residency. Some fellowships exist in hospital medicine geared more towards individuals who want to make it an exceptional career. Most of us who practice hospital medicine have completed our residency and pursue it through or after residency.
Luke Baldwin: I’m familiar with other global models where the primary physician visits the patient in the hospital. Hospital medicine in the US is newer compared to some other specialties. Is the US leading in this aspect, or are other countries ahead of us?
Andre Kumar, MD: Hospital medicine has grown rapidly in the US over the past few decades. It is the fastest-growing specialty in the US and one of the largest. One of the field’s founders, Dr. Robert Wachter, pioneered hospital medicine in the 1990s. The US has done an exceptional job of pioneering the research, carrying forward the field, and advocating nationally with CMS, Medicare, and Medicaid Services to establish it as a distinct field. We’re just seeing tremendous growth as part of it.
AI Integration in POCUS
Luke Baldwin: It’s great that you mentioned your introduction to point-of-care ultrasound. Can you tell us more about your experience with it and the role you think it can play in hospital medicine? We know that emergency medicine, critical care, and anesthesiology have been using it for some time, but where do you see its impact on a hospital medicine physician?
Andre Kumar, MD: As a hospital medicine physician, I’ve seen how expensive healthcare can be for patients who receive multiple tests and consultations for multiple days. Point-of-care ultrasound has been a game-changer in this regard. We can quickly and cost-effectively answer questions about a patient’s heart, blood vessels, and specific organs. Emergency medicine has used it for over 30 years with excellent results, reducing unnecessary testing, improving diagnosis time, and reducing length of stay. As for hospital medicine, it’s an important cornerstone in our healthcare infrastructure and can significantly reduce costs associated with patient care.
Role of POCUS in Reducing Hospital Stays and Improving Patient Care
Luke Baldwin: What is the alternative to POCUS for hospitals that don’t use it?
Andre Kumar, MD: Resource availability at many hospitals is very scarce, which leads to delays in patient diagnosis and care. For instance, a patient may need to wait for a week to get an echocardiogram or consult a cardiologist, which is prevalent across the country. In these resource-limited settings, patients may not receive the study they need because it’s unavailable or they don’t want to wait. Technologies like POCUS and AI-augmented POCUS can reduce these delays and make healthcare more accessible to patients. This technology can be especially impactful in rural cardiology offices where patients may have to wait months to get an echo.
Luke Baldwin: I heard you talk about the potential impact of point-of-care ultrasound on the length of stay in the hospital. Do you think the resource constraints and delays in diagnostic test availability contribute to longer hospital stays, or is it more anecdotal?
Andre Kumar, MD: Research shows that diagnostic test availability and specialist consultation are the most significant factors contributing to delays in patient care and extended hospital stays. While point-of-care ultrasound has been around for several decades, little research has been done on its impact until the last five to ten years. We’re now seeing more studies on how it can improve important metrics for hospitals and patients. Outside of POCUS, we see a similar focus on length of stay reduction in cardiology, gastroenterology, and procedure delays.
Challenges and Opportunities in POCUS Adoption
Luke Baldwin: What challenges are associated with adopting point-of-care ultrasound (POCUS)?
Andre Kumar, MD: One of the biggest challenges we face in learning POCUS is the feedback loop. If I have an attending or a mentor, show me how to scan on a patient, and I feel confident, and then I go off on my own and start scanning, that mentor may no longer be there, and I am, in many ways, scanning blindly or without substantial guidance. That is, I think, one of the most challenging things. Anyone can take a weekend seminar and learn how to scan or learn what a particular disease looks like on an ultrasound. That’s very common. We are missing a critical component: providing real-time and consistent feedback for learners. To take our skills to the next level, that’s where technologies like AI can bridge that gap.
Luke Baldwin: Your recent research looking at how AI-enabled ultrasound (using Kosmos) can impact novice users is fascinating. What inspired you to study this?
Andre Kumar, MD: AI integration into medical technology can overcome many pain points in healthcare, including point-of-care ultrasound. I was motivated to see how a device with AI integration could improve learner skills over time. As an educator, I noticed that learners don’t receive directed feedback once they leave the classroom. Using a device that provides guidance for scanning and image quality can make learners more efficient and augment my teaching ability. This is where the field should be going, and I’m excited to be part of the research that explores this possibility.
Luke Baldwin: Is it challenging to provide feedback to all the medical students and residents who want to learn point-of-care ultrasound (POCUS)? Do you feel pressure to be available for everyone?
Andre Kumar, MD: Yes, providing feedback to all the learners is challenging, especially when there are so many students and residents to teach. I wish I could be there for every moment of their scanning, but there is only so much time in a day. This is where technology like AI can be extremely helpful in guiding learners and promoting expertise. In settings with no teacher, such as limited resources or rural settings, AI can be particularly useful in guiding learners with limited access to expert guidance.
Research Findings on AI-Augmented POCUS for Novice Learners
Luke Baldwin: Can you tell us about your study on using artificial intelligence (AI) for limited echocardiograms and its impact on educational outcomes?
Andre Kumar, MD: Our study aimed to answer a simple question: if you give learners an ultrasound augmented with AI and have them use it for self-directed learning, are they any better at scanning compared to learners who don’t have access to AI or feedback?
We found that learners who had access to AI-augmented devices were faster at acquiring echocardiogram images and acquired higher-quality images. They were also more likely to recognize pathology on actual diseased hearts than those without AI access. These findings highlight the potential of AI in augmenting learning, especially in point-of-care ultrasound.
Luke Baldwin: What do you think the implications of your findings are, particularly for novice users?
Andre Kumar, MD: This was the first randomized study to examine point-of-care ultrasound AI and learning outcomes, which was very promising. The implication is that AI may continue to augment physicians’ skills when using handheld devices. However, there’s a lot more work to be done. Eventually, I would like to see the technology put in the hands of non-medical individuals, such as home health aides or nurses, who can acquire images to help direct a patient’s care. This could increase the accessibility of care, particularly for patients in rural areas who have difficulty receiving follow-up care.
AI's Impact on Experienced Ultrasound Users
Luke Baldwin: Can AI-powered POCUS lead to time savings and increased efficiency in healthcare?
Andre Kumar, MD: Absolutely. For instance, in a benchmark study, we observed that it takes 2-3 minutes to get an ejection fraction through The Simpsons method on more advanced cardiac systems, while our machine with AI assistance took only about 15-20 seconds. Speed can be of the essence when you have a critically ill patient in the emergency department or the intensive care unit or limited time to make clinical decisions in a primary care clinic. Being able to make accurate decisions quickly can have a significant impact on patient care. Hence, I believe that AI-powered POCUS can bring considerable benefits to healthcare.
Future Research and the Role of AI in Healthcare
Luke Baldwin: What’s next for you in terms of research, and what questions do you want to have answered?
Andre Kumar, MD: Taking what we’ve done with AI and POCUS and expanding it to more formalized echocardiograms or multiple views and demonstrating the benefits across multiple centers or sites is a key use case for it. We’re potentially working on a grant to achieve this. The future is also about putting this technology in the hands of nurses and home health aides at skilled nursing facilities or patient homes and demonstrating that they can acquire the images and impact patient care in meaningful ways. Those are the directions we want to go next.
Luke Baldwin: How much research is needed until people are convinced that AI will help them or their students?
Andre Kumar, MD: Studies like this are extremely important for laying the foundation for future studies at larger centers or more schools. But even before then, there’s already a very compelling argument for AI to augment learning. More attention should be paid to devices that offer AI functionality because it will only grow. It will be a $188 billion industry by 2030. AI can potentially improve student learning, and it’s a key aspect of point-of-care ultrasound in the future.
Luke Baldwin: Is all AI created equal? How do you decide what AI is going to be helpful?
Andre Kumar, MD: AI is a broad term, and there are different subcategories to it. AI will be used in medicine for deep learning and large language models. Deep learning involves teaching a computer to learn from a cache of images and make predictions about future images. As for large language models, they include things like chat and the Open AI platform. These two approaches will be game-changers in medicine. However, I don’t think that AI across the board is a great thing for medicine, especially when it comes to sensitive patient information and making medical decisions. It’s important that physicians participate in the research and partner with companies to guide the technology in a safe and effective direction for healthcare organizations and patients.
Luke Baldwin: Can you tell us more about the role of point-of-care ultrasound in hospital medicine?
Andre Kumar, MD: As a hospital medicine physician, I’ve seen how expensive healthcare can be for patients who receive multiple tests and consultations for multiple days. Point-of-care ultrasound has been a game-changer in this regard. It allows us to make real-time diagnoses and treatment decisions that can significantly reduce the length of patient stays and associated costs.
Luke Baldwin: In your recent study, the group with AI-powered devices used them about twice as much as the non-AI group. Was that surprising to you?
Andre Kumar, MD: Yes, it was. However, it speaks to the fact that when you have technology that can potentially augment performance and make you better, you may be more interested in using it, or it’s just something novel and new, and you want to explore it further.
However, it’s important to note that just because they did twice as many scans doesn’t necessarily mean they were much better because they had more practice scanning. They’re still at a level where the skill will not necessarily take off exponentially. Nonetheless, if you implement these technologies into the devices, they may use them more often, and there’s still something inherent about AI that can improve performance.
Luke Baldwin: Is there an age preference for AI?
Andre Kumar, MD: Yes, the younger group is very fascinated by this technology, very interested in it, and all very engaged. Older or more senior physicians find it more challenging to engage in point-of-care ultrasound due to the perception that the technology is not accessible, or they just don’t have a lot of feedback. However, AI can bridge this gap for them. A sub-study can be done in the future to explore this further.