Search
Close this search box.
Categories
Cybersecurity News

From healthcare data to banking information: Researchers aim to share sensitive data for societal benefit

18 November 2024

From healthcare data to banking information: Researchers aim to share sensitive data for societal benefit  

Denmark risks missing out on significant societal advancements because large amounts of sensitive data cannot be shared across organizations. However, a team of computer science researchers has developed an encryption method that enables data to be shared and analyzed anonymously using machine learning.

Every day, companies, researchers, and institutions gather massive quantities of data that remain unused due to privacy concerns or competitive barriers. But what if this data could be anonymized and shared across organizations – for instance to accelerate breakthroughs in medical innovation or to dismantle criminal networks? Secure data sharing and encryption are crucial to turning this vision into reality.

In the Privacy and Machine Learning project, a team of computer scientists, supported by DIREC, has made significant progress toward a solution that lays the groundwork for more intensive data sharing to benefit society.

“There is great potential in using sensitive data to improve various sectors – for instance by developing more effective medicines or enhancing the detection of money laundering. The key is to develop encryption technologies that allow us to process anonymized data securely,” explains Peter Scholl, Associate Professor of Computer Science at Aarhus University and leader of the project.

Decentralized anonymization is the future

The primary challenge in sharing sensitive data lies in ensuring that privacy is maintained during processing. This is where multiparty computation (MPC) comes into play.

MPC is an encryption technique that allows multiple parties to share encrypted data while keeping their individual inputs confidential. This allows the participants to collectively analyze the data and access aggregated results without revealing the sources of the inputs.

“With MPC, we can process data that remains encrypted for all involved parties. No one can view another party’s data, yet they can still derive insights from the results and use them to create innovative products,” Peter Scholl elaborates.

In the financial sector, MPC could enable banks to collaborate on fraud detection by combining payment data without revealing customer identities or competitively sensitive information. Similarly, in healthcare, hospitals could access larger datasets if anonymized patient scans could be shared nationwide.

Striving for efficiency

MPC is not cost-free. Server capacity, power consumption, and other substantial investments play a significant role. Achieving the right balance between cost, security, and performance is a delicate challenge. The research team is also working to enhance the efficiency of encryption processes.

Hiraku Morita, a postdoc in computer science at the Universities of Copenhagen and Aarhus, is working to solve this challenge. In the paper MAESTRO: Multi-party AES using Lookup Tables, Morita and his research team introduce a new technique employing lookup tables optimize the required calculations and improve algorithmic efficiency.

“Most people can quickly answer questions like “what is five times five?” because they’ve memorized the tables or can recall them visually. Similarly, we can enable the algorithms to “remember” complex functions, reducing the computational and bandwidth demands,” says Hiraku Morita.

He adds that the technology has already sparked interest in the industry.

“Many companies have expressed interest in adopting our methods, but it often takes a long time to develop a protocol ready for commercial deployment. We anticipate seeing practical applications within the next few years,” Hiraku Morita concludes.

Read more about the project here.

Associate Professor Peter Scholl to the left and postdoc Hiraku Morita to the right