Development of digital health techniques to quantify the impact of personal exposure to air pollution on chronic respiratory diseases

Chronic respiratory diseases (CRDs), such as asthma and chronic obstructive pulmonary disease (COPD), are major global health challenges, causing significant morbidity and healthcare costs. Exacerbations, or acute worsening of symptoms, are driven by factors like air pollution and respiratory infections, leading to lung function decline and increased risk of future events. Innovative solutions to manage and prevent exacerbations are urgently needed. This thesis explores advanced digital health solutions for CRD management, emphasizing environmental risk factors like air pollution exposure. It focuses on two key objectives: (i) developing and validating the AirPredict platform, a modular digital health solution integrating wearable sensors, a mobile app, and a web interface for monitoring asthma patients and their environment, and (ii) creating a machine learning (ML) framework to predict COPD exacerbations using environmental, physiological, and lifestyle data, with a focus on personalized and population-level models. The six chapters cover: an introduction to CRDs and digital health tools (Chapter 1); AirPredict's architecture and implementation (Chapter 2); platform evaluation and data analysis (Chapter 3); development of ML-based predictive models and patient subtyping (Chapter 4); enhancements through incremental learning and adaptive thresholding (Chapter 5); and synthesis of findings, implications, and future directions (Chapter 6). The research demonstrates the feasibility of combining wearable technology, digital platforms, and ML for personalized care. The AirPredict platform showed high user satisfaction and clinical utility, while predictive models highlighted the role of environmental exposure and individual variability in disease progression. These advancements pave the way for integrating such solutions into clinical decision support systems, offering proactive and personalized approaches to CRD management. This work contributes to digital health and personalized medicine by advancing tools to monitor air pollution's impact on health, improving asthma and COPD outcomes, and laying a foundation for integrating predictive models into healthcare systems.