Federated learning powers more data to support AI processes



The approach has helped in the formulation of predictive models during the pandemic and holds promise for research involving radiological imaging studies.

The use of federated learning as a collaborative approach to conduct research on health data from multiple organizations emerged out of necessity during the COVID-19 pandemic. Now it is seen as a way to expand research, find more widely applicable results, and involve more health organizations.

The ability to access more data for artificial intelligence – while allowing organizations to stay in control of their clinical information and not have to face the hurdles of aggregating it in one place – should have applications in research involving radiological images.

The prospects for using federated learning were discussed in a virtual session at the annual meeting of the Radiological Society of North America (RSNA), held this week in Chicago.

The approach offers a way to better apply artificial intelligence to radiological research. The use of federated learning shows promise in creating larger data pools from which to train algorithms created through artificial intelligence. It is particularly promising to use it to facilitate research on rare diseases or studies involving large x-ray exams that would otherwise be difficult to transfer and consolidate. And it gets around the thorny privacy and security issues that surround data aggregation beyond the walls of an organization.

With federated learning, data stays at the site where it is created, so privacy is always protected and no additional permissions for wider use are needed. Copies of the AI ​​model are sent to each site and AI training is done locally. The approach allows for the use of larger and more diverse datasets that allow AI-based solutions to draw on broader sources than before.

Federated learning first emerged in response to the pandemic, as 20 healthcare organizations around the world joined together to accelerate research, particularly on predicting outcomes in SARS-VOC patients -2. The published to study sought to predict the future oxygen requirements of infected patients based on their vital signs, laboratory data and chest x-rays.

The initiative, called the EXAM study (EMR CXR AI Model), was able to support a rapid data science collaboration that resulted in a clinical decision support algorithm that improved the treatment of COVID-19 patients by predicting a risk score for patients that could predict their likelihood of being admitted and the level of hospital care they would need. In addition, the results of the EXAM initiative could be generalized to larger populations than would have been possible if individual organizations had conducted research based solely on their own data.



Comments are closed.