Generative AI to promote open science in register based research

In the empirical sciences, replicability and validation of research findings are critical for the integrity of scientific advancements.

The open science movement aims to foster a replicable and robust scientific environment through transparency and collaboration, advocating for the sharing of research data and code.

However, studies involving sensitive data, such as Swedish linked registers, face challenges in adhering to these principles due to privacy concerns.

This dilemma is addressed in this project by proposing the creation ofsharable synthetic data through the development of generative artificial intelligence methods.

The project goals include creating sharable synthetic data that accurately represents original register data, ensuresreplicability of register studies, and provides privacy guarantees.

Additionally, it seeks to establish protocols for creating and using synthetic data in register-based research, facilitating data sharing within the scientific community.The research team combines expertise in register dataset design, register based observational studies, and advanced machine learning techniques, including deep neural network architectures.

This unique blend of skills is pivotal for the project success in generating relevant synthetic data with privacy guarantees.

By bridging the gap between openness and confidentiality, the project aims to enhance the value of register based research and promote scientific advancement.