Artificial intelligence (AI) algorithms are increasingly being used by cardiologists and other cardiovascular professionals. In fact, of the more than 500 clinical AI algorithms approved by the U.S. Food and Drug Administration (FDA), cardiology has many specialties except one. There are over 400 in radiology and over 60 in cardiology.
These numbers continue to grow, suggesting that this is just the beginning of the relationship between cardiology and this evolving technology.
In addition to FDA-reviewed clinical and patient-focused AI algorithms, there are hundreds of non-clinical algorithms on the backend to speed workflows, improve efficiency, complete time-consuming tasks, analyze data, and more. Embedded in healthcare IT systems.
Below is a breakdown of the number of FDA-approved algorithms across specialties as of the FDA’s last update in January 2023.
• Radiology 396
• Cardiology 58
• Hematology 14
• Neurology 10
• Clinical Chemistry 7
• ophthalmic 7
• Gastroenterology and Urology 5
• General and Plastic Surgery 5
• Pathology 4
• Microbiology 4
• Anesthesiology 4
• General Hospital 3
• Orthopedics 1
• Dentistry 1
A complete list of FDA-cleared algorithms can be found here
In addition, the American College of Radiology (ACR) also frequently updates and maintains a database of FDA approvals for medical imaging AI. An additional 35 medical imaging AI algorithms have been cleared from January to March 30, 2023, according to the ACR database. It includes 8 new cardiac-related algorithms, bringing the cardiology total to 66. The total number of radiology algorithms is currently 431.
The first AI algorithm was approved by the FDA in 1995, and less than 50 algorithms were approved over the next 18 years. But that number has grown rapidly over the past decade, with more than half of the US market’s algorithms cleared between 2019 and his 2022, and over 300 apps cleared in just four years. Last October, the FDA approved 178 new AI and machine learning (ML) systems. The FDA said that number is expected to grow rapidly in the future.
Get AI buy-in from cardiologists
Ami Butt, M.D., chief innovation officer at the American College of Cardiology (ACC) and cardiologist for adult congenital heart disease at Massachusetts General Hospital, says artificial intelligence is actually a collaboration between human doctors. , said that it really should be renamed “collaborative intelligence.” A machine that harnesses the best that both have to offer to improve efficiency in patient care and healthcare. The ACC has been a strong advocate for bringing more cardiology AI to market to help augment the number of cardiologists in the US as the shortage grows.
“It’s just that the human eye can’t necessarily take all the data and process it like a computer. are trying to work at their best.” And sometimes our clinical acumen is superseded by computer suggestions, which is fine and should be. Because then we re-educate it. It’s a way of learning with computers,” Batt explained.
Batt also emphasized that this does not mean that all doctors need to be experts in AI.
”[Physicians just] You have to understand that there is an AI here that has a computational side and its purpose is to: Beyond that, we need to leverage clinical acumen,” she said.
Bhatt et al. also point out that for AI to be usable by clinicians, it must be seamlessly integrated into workflows, similar to medical devices and reporting software.
What cardiologists need to know about AI
“We really need to increase cardiologists’ understanding of this technology. We live in a digital world and medical care tends to be pretty conservative, but people are embracing AI and they have to embrace it. In the clinical world, we have to be at the forefront,” said Consultant Cardiologist and Emeritus Senior Clinical Lecturer at King’s College London, Cardiovascular Computed Tomography. Dr. Ed Nicol, president-elect of the Society of Radiographers and Telegraphers (SCCT) explained.
The academic debate on AI and its future applications is almost over. What is currently being discussed is the actual FDA-approved product that is on the market and in clinical use.
What makes this all the more realistic is the fact that the first cardiac AI algorithms are now included in clinical practice guidelines in both Europe and the US. The ACC/AHA 2021 Chest Pain Guidelines include the recommended use of AI-driven preliminary blood flow ratiometry derived from non-invasive CT imaging (FFR-CT).
“If you told me 15 years ago that we would put some sort of computational fluid dynamics tool into the U.S. chest pain guidelines, you would have laughed. Everyone said you were crazy. “Based on the evidence, we now see FFR-CT listed in international guidelines, and this will likely be the first of many,” Nicol said.
He said the biggest thing cardiologists need to understand is how the AI they are evaluating or using works. This will help you understand if your AI-generated data is flawed and how it should be validated.
“We clinicians really need to own this,” Nichol explained. “And you can own and challenge this only if you understand the strengths and weaknesses of AI and AI. We are turning the entire radiology/cardiology community into programmers.” No, even if they don’t understand every string of computer code.”
“One of the things that amazes me is the rapid progress being made in trying to standardize analysis using non-invasive imaging and artificial intelligence,” said Cardiovascular Research Foundation (CRF) president and CEO, CEO), Dr. Juan Granada, M.D., explained the key points. The technology trends he sees in cardiology.
AI in cardiac imaging enables faster lab readings by automating quantification, increasing measurement consistency, and eliminating the normal variability between radiologists and cardiologists. Granada believes AI will soon play a major role in cardiac care across all subspecialty areas, including interventional cardiology.
Although FDA-cleared AI algorithms exist to automatically evaluate ECG and wearable heart monitor data, most cardiac AI is focused on image processing. Moreover, many radiology-approved algorithms are actually specialized for cardiovascular, peripheral, and neurovascular imaging. These include uses in CT, MRI, nuclear imaging, and echocardiography.
AI in echocardiography
“Many areas of echocardiography are incorporating AI. AI is exploding and very exciting,” explains echocardiography expert Patricia A. Perica, M.D. .she is the editor-in-chief of American Journal of Echocardiography, Mayo Clinic Ultrasound Research Center Director, Mayo Clinic Cardiology Consultant. “One of those areas is using AI to improve ultrasound acquisition of images by teaching inexperienced users how to acquire images. It’s about applying AI to and re-measuring things, or applying AI to all the data’ measurements that have already been taken to detect disease. ”
During his research at the University of Mayo, Pelika has been involved in developing AI that directly detects disease by examining ultrasound images. This algorithm can detect radiological features of disease in images that may not be apparent to the human eye.
“I think there’s a lot of potential there,” Pericca said.
Over the last few years, there has been a significant increase in the use of point-of-care ultrasound (POCUS) systems in a variety of settings, including clinics, clinics, emergency departments, and ICUs. These echo exams are performed by much less experienced sonographers than in hospital echo labs. Several AI vendors have developed FDA-approved algorithms that show POCUS users how to move the probe into the correct position and explain how to acquire each standard echo view. AI also tells the operator that they are in the correct position and determines the quality of the image they are getting. Many echo experts say this will greatly improve the diagnostic quality of the test, reduce retesting, and allow patients to be diagnosed more quickly and appropriately.
“This extends the reach of cardiovascular ultrasound to places and times when there are no experienced echocardiographers available to perform the imaging. I think this has very interesting potential.” explained Perika.
AI is also being used to expedite patient assessment by automating echo strain, ejection fraction, and other measurements. Pellikka said several vendors are currently developing FDA-cleared algorithms in these areas. This automation eliminates the need for the sonographer to manually contour the ventricle or use calipers to manually perform various measurements. She said sonographers can still edit these AI-generated contours if they feel they’re wrong. Overall, she said, this will significantly speed up testing and post-test processing, increasing patient throughput.
Importantly, Pericca and other echo experts say this kind of AI automation can also reduce measurement variability between different sonographers. AI automatically selects the same landmark locations to perform measurements, resulting in more consistency. This is especially important for long-term patient monitoring.
“I think this is just the tip of the iceberg. I think many other measurements will also be automated, such as automated assessment of valvular heart disease. “It will promote standardization.” There is more standardization between centers that do echoes and other centers than there is now,” Pericka said.
