Pipeline Aplenty: FDA Greenlighting Medical AI Apps

When it comes to AI and machine learning, the regulatory trail has been blazed and the approval gates through open. The FDA has approved a couple dozen apps over the last year and a half—and the momentum is clearly building with Scott Gottlieb at the agency’s helm and recent moves to ramp up staffing to meet the demand.  

A review of standout approvals over the past 16 months suggests the technology is building a serious head of steam. Here’s a sample timeline:

January 2017. Arterys wins 510(k) clearance for its Cardio DLTM application. The first FDA-cleared clinical product to use cloud computing and deep learning, it precisely segments and assesses ventricular function from MRI scans of the heart.

July 2017. Quantitative Insights receives de novo clearance to sell its QuantX Advanced platform, a CAD system that taps machine learning to evaluate potential breast cancers.

November 2017. Arterys picks up another nod, this time for a web-based analytic tool called MICA that uses AI to provide real-time insights for a variety of medical imaging modalities.  

January 2018. FDA grants its “expedited access pathway” designation to MedyMatch Technology that uses deep learning to analyze CT images of the head for signs of intracranial hemorrhage—and when it detects bleeding in or around the brain, it automatically alerts the treating physician.

February 2018. Viz.ai gets the go-ahead on its Contact software for helping determine if a patient is having a stroke. That same month, Arterys continues its roll with Oncology AI suite, which aids in diagnosing cancers of the liver and lungs.

April 2018. Another first and this time it’s AI software that detects diabetic retinopathy (DR) without a doctor’s help.

So what’s coming down the pike? 

Researchers from the Medical University of South Carolina and Siemens Healthcare created a machine learning algorithm that, in a trial, outperformed coronary CT angiography and quantitative coronary angiography in identifying heart blockages.

Using digital pathology images, a team led by investigators at Stony Brook University have used deep learning to map cancerous immune-cell patterns, their hope being the pattern recognitions may guide new cancer therapies.

A research team from Cleveland Clinic confirmed the efficaciousness of an AI-based EKG app for the Apple Watch at detecting atrial fibrillation. The FDA had already cleared the device for the market, but the researchers extolled the merits of its AI component in particular, possibly heralding the approach of many more consumer-ready algorithms to come.