Revolutionary Data Fusion Technique Safeguards Privacy and Boosts Recognition in Medical Research

Ensuring unbiased data collection is crucial for enhancing the performance of artificial intelligence (AI) analysis. However, it is essential to maintain confidentiality and protect personal information while integrating data from multiple institutions. To address this challenge, a research team at the Center for Artificial Intelligence Research (C-AIR) has developed a secure AI technology called "non-readily identifiable data collaboration analysis. This approach enables the integrated analysis of personal information while sharing only abstract data that cannot be readily identified with the original data.

Image Credit: Gorodenkoff / Shutterstock

The team has established mathematical definitions for readily identifiable data and proposed an integrated analysis algorithm that shares abstracted data instead. By implementing this technology, a broader range of data, including personal information, can be used in AI analysis, leading to significantly improved accuracy.

Practical applications of this technology include disease prediction by analyzing test and medication data from multiple medical institutions to estimate risk factors. It can also enhance educational effectiveness by integrating student data from various educational institutions. Overall, this innovation paves the way for a new platform that gathers high-quality personal information from diverse sources, safeguards original data, and leverages AI for comprehensive analysis.

In the future, the researchers plan to compare the performance of their method with other approaches, including federated learning, and further refine the algorithm and software, such as by incorporating vertical data partitioning. Additionally, they intend to collaborate with multiple medical institutions to conduct multi-source data fusion using real medical data.

This work was supported in part by the New Energy and Industrial Technology Development Organization (NEDO), Japan, Japan Science and Technology Agency (JST) (No. JPMJPF2017), and the Japan Society for the Promotion of Science (JSPS), Japan, Grants-in-Aid for Scientific Research (Nos. JP19KK0255, JP21H03451, JP22H00895, JP22K19767).