IoTMindCare: Safe and Personalised IoT-Based Depression Management

Abstract

Depression is a complex and highly individual mental health condition that requires timely, personalised, and context-aware support. Traditional depression management often relies on periodic clinical assessments and self-reported symptoms, which can limit the ability to detect changes over time, respond to individual needs, and provide continuous support in everyday environments. Recent advances in Internet of Things (IoT)-enabled health technologies have created new opportunities for continuous and personalised depression management. However, existing IoT-based mental health systems remain fragmented, with many focusing on isolated sensing, symptom prediction, or prototype-level functionality rather than integrated, safety-aware, personalised, and implementation-informed solutions. This thesis proposes IoTMindCare, an implementation-informed reference architecture for safe, personalised, and context-aware depression monitoring and management. The thesis addresses these gaps by integrating evidence from existing IoT-based depression systems with data-driven monitoring, engagement-focused design, architectural modelling, and expert-informed refinement to inform safer, more personalised, and more implementation-aware depression management. First, a systematic review establishes the current state of IoT-based solutions for depression and emotional well-being, identifying key gaps in sensing approaches, data analysis, user engagement, privacy, security, interoperability, and system integration. Second, a data-driven IoT approach examines how environmental factors, particularly temperature and humidity, may provide an additional contextual layer for interpreting mental health-related patterns. Third, a gamified IoT-based study explores how micro-games, adaptive surveys, sensor data, and personalised feedback can support user engagement, self-awareness, and ongoing self-management. Fourth, the IoTMindCare reference architecture integrates these insights into a modular framework that combines multimodal sensing, hybrid edge-cloud processing, personalised analytics, anomaly detection, alert mechanisms, and stakeholder-centred design. Finally, an expert-informed refinement study extends IoTMindCare by incorporating technical and clinical perspectives on feasibility, safety, personalisation, governance, interoperability, workflow integration, and deployment constraints. The main contribution of this thesis is a coherent architectural foundation for moving IoT-based depression management beyond fragmented prototypes toward safer and more personalised digital mental health systems. The architecture is evaluated through design-level analysis and expert-informed refinement; its integrated operational performance, clinical effectiveness, and deployment feasibility remain to be established through future empirical studies. Rather than treating depression monitoring as a purely predictive or technology-driven task, the thesis argues that future IoT-based depression management requires contextual interpretation, personalised calibration, human-in-the-loop oversight, alert governance, privacy-aware design, and alignment with clinical and real-world implementation requirements. IoTMindCare provides a foundation for future prototyping, simulation, pilot studies, and empirical validation of IoT-enabled systems for depression monitoring and management.