[1L4-OS-18a] 生体信号を活用した医療・ヘルスケアAI

[1L4-OS-18a-03] A Differential Privacy Method to Balance User Privacy Preservation and Usefulness in Services Using Time-Series Heartbeat Data

〇Ryusei Fujimoto¹, Yugo Nakamura¹, Yutaka Arakawa¹ (1. Kyushu University)

Keywords:Time-series heart rate data, Differential privacy, Healthcare, Privacy preservation, Abnormal heartbeat detection

In recent years, with the spread of wearable devices and their built-in sensors, the analysis of time-series heartbeat data has been promoted in research in the healthcare field. Because time-series heartbeat data can be used to detect abnormal heartbeats and estimate psychological stress, there are many medical and healthcare services that use heartbeat data. However, time-series heartbeat data contains a lot of privacy information, which requires careful consideration of privacy protection for users of these services. In this study, we proposed and evaluated a privacy-preservation mechanism using ``differential privacy'' for a service that detects abnormal heartbeats based on time-series heartbeat data. The evaluation results show that the proposed method is more effective than conventional differential privacy-based mechanisms in balancing service usefulness while realizing user privacy protection.