Model identification for homeostatic data**

点击翻译按钮获取中文摘要

Various common diseases, such as diabetes and epilepsy, can be managed but not cured. Managing the disease often requires invasive or expensive measurements. The data thus produced can be unreliable and infrequent. A new alternative has presented itself in recent years due to the miniaturization of sensors and their integration with everyday electronics like watches and phones. Homeostatic biomarkers like heart rate, blood pressure and core temperature can be measured continuously and stored in a central repository. The challenge is, to infer from such secondary data early warning signals for the onset of pathological states like hypoglycaemia and seizures. **In this project, we will develop practical algorithms to model and detect anomalies in homeostatic data. These algorithms will combine data-centric techniques like de-noising and curve fitting with elements of modelling of the underlying physiological processes. Ultimately, the purpose of these algorithms is to predict, in real time, the future time course of the relevant biomarkers for individual patients. **The research will take place jointly at the R&D laboratory of Klick Health, the world's largest independent health marketing and commercialization agency, and in the Modelling and Computational Science graduate program at UOIT, a young and STEM-focused university.**The planned research will strengthen the Research as a Service portfolio offered by Klick Health to a variety of clients in the pharmaceutical industry and support Klick Health's goal to enhance their credentials as a top-tier research company. **The student working on the project will be offered a unique opportunity to experience industrial R&D at one of Toronto's fastest growing tech companies, and is expected to continue working towards his or her graduate degree after completing the joint project.******