Using web interaction to monitor parkinson's disease progression through behavioural inferences on the web

Traditional Parkinson's Disease (PD) assessment techniques are inaccurate, sporadic and subjective. Although recent works have used wearable devices to try to overcome these issues, most interfere with people's routines, are uncomfortable to use and unsuitable for long-term assessments. In contrast, my approach aims to monitor PD in a longitudinal, naturalistic, non-disruptive and non-intrusive way. It uses smartphones to log social, environmental and web interaction data about people and their surroundings. This data is complemented with other web data sources and then processed to infer a set of metrics (a latent behavioural variable or LBV) of people's activities and habits. Then, LBVs' trends are quantified and mapped to the progression of the disease. During a first pilot study, I collected a dataset with ≈290 million records that has 34.5x more rows and scanned 4x more data sources than state-of-the-art sets. I used this data to identify six possible PD-related LBVs. This project aims to get a more accurate disease picture and to reduce the physical and psychological burden of traditional and other technology-based assessment methods. Ultimately, the work has the potential to save people's time and improve the efficiency and effectiveness of health services.