Tech

Want AI-powered doctors to free patients? Try sticking a sticker that looks like a food label, says Uncle Sam. • The Register

The U.S. Food and Drug Administration has given manufacturers special labels on AI-powered medical devices to help patients understand the technology and test it with real-world data for wild performance beyond the clinical environment. I want to confirm.

Watchdog publishes five action plans for regulating standalone devices that implement machine learning algorithms. AI in healthcare is a huge market that is projected to reach a whopping $ 6.6 billion in the United States in 2021.

Companies that want to get slices of pie need to ensure FDA approval if they want to bring AI-based products to market. However, it is difficult for regulators to develop this process because of the rapid evolution of technology.

“This action plan outlines the FDA’s next steps to facilitate AI / ML-based SaMD monitoring,” said the Medical Device and Radiation Health Center, which is responsible for the approval of premarket medical devices. Bakuru Patel, director of the FDA Digital Health Center of Excellence, said. It’s software as a medical device – SaMD – in case you’re wondering.

“This plan aims to promote the great potential that these technologies must improve patient care, while providing safe and effective software features that improve the quality of care that patients receive. It outlines an overall approach based on monitoring the entire life cycle. To stay up-to-date, address patient safety, and improve access to these promising technologies, this action plan over time. I expect it to continue to evolve. “

Doctor

Leading Doctors Blame Google for Not Supporting Incredible Claims for Superhuman Cancer Detection AI

read more

First, the action plan outlines the need to update the current regulatory framework to scrutinize AI software that changes over time. The algorithm is improved and updated by training new data. Despite these changes, it is important to ensure that your device remains safe for medical use.

The manufacturer must provide a “predetermined change control plan”. This is a document that outlines what has changed exactly and explains how the algorithm learns those changes. The updated regulatory framework will be detailed in another report published later this year.

The second point of the plan is working on what regulators call “Good Machine Learning Practice (GMLP for short).” We need to establish a set of standards to guide companies that develop and manufacture AI medical devices.

To ensure uniformity, the FDA said it has partnered with other communities working on similar benchmarking issues, such as the British Standards Association and the Institute of Electrical and Electronics Engineers (IEEE).

Machine learning software bias is especially important in health care, where the quality of care can vary by people’s age, gender, wealth and ethnicity. The FDA warned that if the device was trained with historical data that reflected these biases, the software could run into problems and slow performance in different groups of people.

“The Agency emphasizes the crucial importance of medical devices adapting well to racially and ethnically diverse patient populations and the need for improved methodologies for identifying and improving machine learning algorithms. We are aware, “said the action plan.

We work closely with scholars to devise algorithms that may even eliminate biases to identify biases and investigate future software.

Who reads the label?

If your food is labeled with ingredients and nutritional value, why not use a medical device? The FDA wants to provide a clear label that describes what the manufacturer goes in and out of the algorithm.

eye

Research: AI designed to detect diabetic eye disease blinks in the real world increases the work of doctors

read more

The label should describe the data used to train the software, what the device takes as input to measure or produce results, and overall performance with respect to accuracy. We argue that increased transparency means that patients and users are more likely to trust and adopt the technology. The FDA will hold a public workshop later this year to see how to implement such labels.

Finally, the agency requires companies to enroll in a pilot study focused on testing AI medical devices using real-world data. However, this study is voluntary and requires the general public to be willing to become a guinea pig.

“The assessments performed as part of these efforts are the most important metrics for AI / ML-based SaMD RWP, including those that can be used to proactively address safety and ease of use concerns. It can be used to determine thresholds and performance ratings, to get feedback from end users, “says Watchdog.

“We recognize that AI / ML-based SaMD is a rapidly evolving area, and this action plan has evolved as we pursue these activities and strive to provide more clarity in this area. We expect to continue, “concluded the FDA. The agency spokesperson couldn’t immediately make any further comments. ®

Want AI-powered doctors to free patients? Try sticking a sticker that looks like a food label, says Uncle Sam. • The Register

Source link Want AI-powered doctors to free patients? Try sticking a sticker that looks like a food label, says Uncle Sam. • The Register

Back to top button
Close