As of January 2023, over 520 AI algorithms have gotten the green light from the FDA for distribution in the U.S., with the majority designed for use in medical imaging.

While the number of algorithms in other specialties are not even in the triple digits, the FDA recorded 396 in radiology, according to Health Exec. AI’s high accuracy in pattern recognition has boosted diagnostic capabilities, leading to a concentration of market approvals in medical imaging.

Most are approved for imaging across specific subspecialties, such as the brain, breast, cardiac, lung and stroke. The first was cleared by the FDA in 1995, with fewer than 50 more approved over the next 18 years. But in the last decade, approvals have skyrocketed, with more than half entering the market between 2019 and 2022, which is more than 300 apps in just four years.
"Even though all these FDA-approved algorithms are using AI, it is not at all what most people think of, where it is making a diagnosis or finding a lesion. There is a big variety in the ways to classify AI,” Dr. Keith J. Dreyer, vice chairman of radiology at Massachusetts General Hospital and American College of Radiology (ACR) Data Science Institute chief science officer, told Health Exec.

In imaging, AI is being used for identifying critical findings; automating time-consuming functions such as quantification; data mining; workflow improvement and automation; clinical decision support; iso-centering patients, choosing protocols and speeding up MR exam times; image reconstruction; automatic anatomical identification and contouring of organs and tissues; and guides for capturing quality scans.

The FDA cleared 178 AI and machine learning solutions in October 2022 and expects that number to increase rapidly. Because some types of AI can learn on their own, the FDA is looking into modifying its clearance process for these solutions, adopting measures that will allow it to make adjustments in real-world settings, while maintaining the safety and effectiveness of the software as a medical device. This system being considered is being developed with feedback from a 2019 AI stakeholder’s meeting.

Additionally, more algorithms are being developed for nonclinical work, including population health, health tracking apps, health equity gaps, revenue cycle management streamlining, hospitalwide monitoring, data analytics for key performance indicators, and better patient wellness and preventive care.

Back to HCB News

X-Ray Homepage