FDA on track to clear 1,000+ clinical AI algorithms by end of the year
The number of clinical artificial intelligence (AI) algorithms cleared by the U.S. Food and Drug Administration (FDA) is on track to surpass an all-time total of 1,000—and that figure only includes AI used in patient care. As of writing, the agency has approved 950 different AI algorithms since beginning the process in 1996, according to its internal data last updated July 2024.
In 2024 alone, the FDA has already approved 107 algorithms.
In 1996 when the FDA began approving AI for clinical use, only 33 models were cleared over the 20 years that followed. The numbers of AI approvals exploded in 2016, then began growing exponentially each year after.
Medical imaging dominates
Radiology has been the clear leader, well ahead of all other medical specialities, in commercialized AI and deep-learning algorithms. The radiology field has seen a total of 723 cleared AI algorithms, meaning the specialty accounts for 76% of all AI cleared by the FDA to date. And when broken down by sub-specialty—orthopedics, neurology, dental, surgery and pathology—the number could be closer 90%.
Cardiology and neurology in second and third place
The field of cardiology now has 154 FDA cleared AI algorithms at its disposal, which includes algorithms listed under the primary category of radiology but are specific to cardiovascular imaging.In total, that accounts for 16.2% of cleared AI models.
Neurology is a distant third, with 34 algorithms, or about 3.5%.
All other medical specialities make up the remaining 10% percentage of so of AI clearances. The breakdown of FDA-cleared AI by speciality is as follows:
• Radiology 723
• Cardiology 98, (or 154 if including cardiac specific AI listed under radiology)
• Neurology 34
• Hematology 18
• Gastroenterology and urology 14
• Ophthalmic 10
• Anesthesiology 9
• Clinical chemistry 8
• Pathology 8
• General and plastic surgery 6
• Orthopedic 5
• Microbiology 5
• General Hospital 4
• Dental 3
• Ear, nose and throat 2
• Physical Medicine 1
• Immunotherapy 1
• OB/GYN 1
Future of nonclinical commercial AI may face FDA review
AI has been integrated into the backend of many healthcare IT systems for more than a decade, but if it does not have a direct patient-facing impact on care, it has not been subject to FDA review.
Hundreds, if not thousands, of AI applications help to manage clinical data and fall into this category, which includes AI to streamline analytics, data mining, population health, data flow and connectivity in and between electronic medical records (EMR) and nearly all other IT systems in healthcare facilities.
Over time, the lines between clinical and nonclinical system have become increasing blurred. Examples include AI that decides what prior exams, procedures and medical history is relevant to show a physician to streamline the amount of data human doctors otherwise need to sift through, or which proper exams would be helpful to show a radiologist. This includes clinical decision support programs and AI models that analyze medical images—both of which are sometimes subjected to FDA clearance.
The FDA is considering regulating a greater number of software system AI algorithms that impact patient care, asking for public comment on how to proceed earlier this year. Last year, the agency made steps to further
In April 2023, the FDA issued a draft guidance on marketing submission recommendations for a predetermined change control plan for AI/ML enabled device software. The FDA is focusing on asking vendors to bring the AI products forward that match the definition of a SaMD.
In October 2023, the FDA issued the document Predetermined Change Control Plans for Machine Learning-Enabled Medical Devices: Guiding Principles. This documents identified certain changes to MLMDs, such as changes to a model or algorithm, may have a significant impact on patient care and may require regulatory oversight, including additional premarket review.
AI is also increasingly being used in drug development, which is another area the FDA is monitoring. Since 1995, the FDA has received over 300 submissions for drugs and biological products with AI components, explained FDA Commissioner Robert M. Califf, MD, in an article he wrote from March 2024.
"Submissions have included aspects related to drug discovery and repurposing, enhancing clinical trial design elements, dose optimization, endpoint/biomarker assessment, and postmarket surveillance. These submissions also cover a growing diversity of medical devices that leverage AI to improve clinical workflows and patient experiences or outcomes in addition to sophisticated prediction algorithms," Califf said.
FDA looking at adopting its own AI to streamline workflow
Califf also said the FDA is also exploring its own use of AI technologies to facilitate internal operations and regulatory processes.
"At its most basic, AI can strengthen our operational systems and bring increased productivity, opportunity, and efficiency to our work, helping us process and analyze complex data faster, including data from medical imaging or digital health technologies," Califf wrote. "We can free up staff by automating repetitive administrative functions and enable them to focus on more complex meaningful activities to weigh the evidence and arrive at better decisions."
He added that this could benefit both agency experts and the public by streamlining workflows to facilitate faster approval for AI going forward.
Dave Fornell has covered healthcare for more than 17 years, with a focus in cardiology and radiology. Fornell is a 5-time winner of a Jesse H. Neal Award, the most prestigious editorial honors in the field of specialized journalism. The wins included best technical content, best use of social media and best COVID-19 coverage. Fornell was also a three-time Neal finalist for best range of work by a single author. He produces more than 100 editorial videos each year, most of them interviews with key opinion leaders in medicine. He also writes technical articles, covers key trends, conducts video hospital site visits, and is very involved with social media. E-mail: dfornell@innovatehealthcare.com