A Continuous Authentication Scheme for Digital Therapeutics Using Denoising Autoencoder
Chengling Wang, Yunru Ma, Yuexin Zhang, Ayong Ye, Li Xu · 2023
In healthcare, digital therapeutics provides users with better services in terms of preventing, managing and treating diseases. However, it is a non-trivial task to ensure the security of digital therapeutics. In some scenarios, for example, the wearable biosensors might be used by others or violated by malicious adversaries. When these happen, the patient’s treatment might be affected. Investigation shows that conventional one-time login authentication mechanism cannot solve the above problems. Motivated by these observations, in this paper we present a continuous authentication scheme for digital therapeutics using denoising autoencoder. Specifically, it achieves user-transparent re-authentication when users wear the wearable devices. To improve the re-authentication rate, the denoising autoencoder is employed to learn gait features using three-axis acceleration data. To illustrate the practicality, we evaluate the performance of our scheme using public datasets and compare it with that of relevant schemes. The results show that the re-authentication rate of our scheme reaches 92.8%, and it works well no matter users’ activities (walking, sitting, running and/or jumping, etc).