The PRIVASA project, led by the University of Turku, has progressed well during its first year. Understanding of the nature of the data being processed has increased steadily, and the infrastructure required for the project has also been completed.
The aim of the project is to develop artificial intelligence methods that can replace sensitive data from real patients with secure de-identified synthetic data based on the real patients. The end result is material that can be processed without compromising the privacy of individuals.
“The aim is to generate original, synthetic data from sensitive health data that cannot be returned to its source”, Jussi Salmi, adjunct professor at BCB Medical, clarifies the main goal of the project.
Synthetic data helps to tackle data protection challenges
There are two years left in the PRIVASA project, and the goals of the project are clear. “Our goal is to create services for companies and researchers that can be used to produce high-quality and easily accessible data. When the information is synthetic, the processing does not require data protection permits”, Salmi says.
BCB Medical’s goal is to finalize the process into a production pipe that can be used to produce the required data smoothly and to ensure its quality. To this end, the company is developing metrics to ensure that the data is sufficiently different from its source, but still of sufficient quality to match the original.
The University of Turku agrees with the objectives of the project: “The use of synthetic data solves companies’ problems with data collection, storage and disclosure. At some point, it may be the case that synthetic data can replace genuine data”, says Antti Airola, Assistant Professor of Health Technology at the University of Turku.
Versatile cooperation strengthens the project
BCB Medical’s strength in the project is the number and quality of registers it develops.
“We have several registers with large amount of usable data. The registers serve as a good starting point for producing synthetic data, which is of interest to both pharmaceutical companies and academic researchers”, says Salmi.
At the University of Turku, an interdisciplinary research team has studied widely the nature of the data including from a legal and ethical point of view.
“Data protection only applies to real individuals, so one of the research questions is how to ensure that the material being processed is artificial and not a copy from a genuine source. It must be shown that the method can be used to produce useful, but at the same time anonymous data”, says Airola.
Airola says that health data analytics is one of the key research questions at the Department of Computing at the University of Turku. In addition to researchers from the Department of Computer Science, researchers from the Faculty of Law and the School of Economics are participating in the research. The research team focuses on developing methods for producing high-quality data and protecting data security.
The PRIVASA project as a whole is the first of its kind and the Nordic countries are a scientific pioneer in this field.