Fresenius to develop AI models for dialysis care with new database of over 540,000 patients

This audio is auto-generated. Please let us know if you have feedback.

Dive Brief:

  • Fresenius Medical Care has gathered anonymized data on more than 540,000 dialysis patients, providing a launchpad for its artificial intelligence aspirations.
  • The company, which runs dialysis centers around the world, has aggregated data on patients in 40 countries across six continents. The data covers more than 350 patient treatment parameters, 140 million dialysis treatments and 34 million laboratory assessments.
  • Fresenius sees the resource supporting improvements in care quality and patient outcomes, including by providing high-quality, relevant and timely data for use in the development of AI models.

Dive Insight:

Fresenius’ status as a manufacturer of dialysis equipment and operator of dialysis centers gives it opportunities to collect data. However, the company had to work through challenges to establish a viable database, said Len Usvyat, head of clinical advanced analytics for Fresenius.  

“Dialysis care generates a large amount of data that can be used for secondary purposes, but multinational datasets are scarce due to the fundamental need for adherence to varying complex data protection regulations around the world, as well as the challenges in harmonization of data from different clinical systems,” Usvyat said in a statement. 

Fresenius harmonized and aggregated data from its dialysis centers, resulting in a resource that Usvyat said “increases the speed and robustness of the company’s analytical capabilities and provides greater consistency in generating data-driven clinical insights.” The dataset is fully anonymized and meets privacy requirements, including HIPAA in the U.S. and GDPR in the European Union, according to Fresenius.

Details of the impact of the dataset are yet to emerge. The project will be featured in multiple of Fresenius’ publications at the upcoming American Society of Nephrology Kidney Week conference.

Fresenius said the database is supporting more than 15 clinical improvement projects, including work to assess the feasibility of using an AI model to optimize the use of erythropoietin stimulating agents and iron therapies in dialysis patients.