Enhancing security and healthcare through continuous monitoring on wearable devices

Wearable and mobile devices are becoming deeply embedded in daily life, supporting a growing range of tasks - from communication and entertainment to health monitoring and productivity. As people increasingly rely on these devices, it is essential to ensure both the protection of the sensitive data they store and the ability to derive meaningful insights that enhance users’ well-being. This dissertation investigates how wearable and mobile devices can be leveraged to provide robust and user-friendly solutions through continuous monitoring across two critical domains: security and mental health. The foundation of this dissertation is the development of a security infrastructure, as safeguarding personal and sensitive data is essential for building user trust. Behavioral biometrics are explored as authentication mechanisms for wearable devices due to their unobtrusive nature and ease of collection on these devices, compared to traditional passwords or physiological biometrics. This work defines the swing phase, an interpretable structure in gait cycles derived from accelerometer signals, and proposes a lightweight user authentication model based on features extracted from this structure. It then extends to behavioral biometric-based continuous authentication, which offers ongoing and unobtrusive protection. As single-modality systems often suffer from limited accuracy and vulnerability to adversarial attacks and real-world disturbances, this dissertation introduces SSPRA, a robust multimodal continuous authentication framework. SSPRA models the authentication process as a Markov process and employs a two-level fusion strategy combined with two temporally modeled state transition machines to integrate multiple behavioral biometrics. It demonstrates not only high adversarial attack detection accuracy but also consistent reliability across four simulated real-world scenarios, effectively addressing key challenges such as adversarial attacks, modality dropout, and data interruptions. Building on this security foundation, the dissertation then explores how wearable devices can improve healthcare. As public awareness of mental health’s impact on overall well-being continues to grow, wearable mental health monitoring has become one of the most widely adopted features in daily life. The dissertation first introduces WearStreM, a multimodal stress monitoring framework that adapts SSPRA’s two-level fusion strategy and incorporates a novel multi-state transition mechanism to detect and track episodic stress in real time. A new set of temporal evaluation metrics is also proposed to assess both detection accuracy and responsiveness, supporting more comprehensive and realistic evaluation for practical deployment. Experimental results from both lab-controlled and real-world driving settings validate the model’s ability to detect entire stress episodes and promptly respond to stress changes. To complement stress monitoring and provide a broader perspective on mental health detection and management, the dissertation also explores the recognition of negative emotional states. It presents DSTER, the first transformer-based emotion recognition model using keystroke dynamics. DSTER employs a dual-stream architecture to achieve robust detection of negative emotions and demonstrates strong generalizability across five emotional states and diverse user populations. This approach offers a non-intrusive and privacy-preserving solution to support mental health monitoring and enhance human-computer interaction. By advancing both security and mental health monitoring through continuous, practical, and unobtrusive systems, this dissertation demonstrates the potential of wearable and mobile devices not merely as passive tools, but as intelligent, supportive technologies that promote a more secure, health-aware, and connected way of living.