patients for new clinical trials [14]. Most of this work focused on the analysis of very large multidimensional longitudinal patient data collected over many years. However, most clinical databases provide low temporal resolution information due to the difficulty in collecting rich long-term time-series data. To bridge this gap, current clinical databases can be enhanced by connecting with mobile health platforms, community centres, or elderly homes such that other information can be incorporated into the system to facilitate clinical descision-making and address unanswered clinical questions. One interesting direction will be to build patient-specific models using data already available in existing clinical databases, and, then, update the model with data that can be collected outside the hospitals. In particular, some chronic diseases are manifested with acute events that are unlikely to be predictable solely by sporadic measurements made within hospitals. Taking thoracic aortic dissection, a relatively rare disease (3–4 per 100 000 people per year), as an example, the disease is typically manifested as a tear in the intimal layer of the aorta, which can later on develop into either type A (involving both ascending and descending aorta) or type B dissection (involving descending aorta only). Type-A patients would require immediate surgical intervention, whereas for type B dissection, it is generally considered as a chronic condition requiring careful long-term control of blood pressure (BP).