Date: Wednesday 23 March
Time: 1.00pm – 2.00pm
Speaker: Professor Maia Angelova
Title: Time series in healthcare: challenges and open problems
Abstract: Time series datasets, such as electronic health records (EHR), electrocardiograms (ECG), electroencephalograms (EEG), sleep records, monitoring vital signs, COVID-19 spread, are sources of information that can capture the onset and spread of disease, lifestyle risks, the results and efficiency of treatment. The transformation of healthcare through advanced analytics and machine learning is based on the time series to model the trajectories for health and disease. With the advances of technology, data collections come in different forms, volume, time periods. Furthermore, wearable devices facilitate the production of vast amount of complex data at low cost. Data science and artificial intelligence provide powerful tools for data discovery and automatic decision making based on the patterns hidden in data. Time series are the skeleton of patient-centred healthcare. The challenges and open problems in time series analysis include representation, fusion, forecasting, visualization, explainability and interpretability of the modelling outcomes. Clustering, dynamic forecasting, selection of machine learning models, early diagnosis, prediction of healthcare cost, automatic decision making, these are some of the current challenges the solution of which is dramatically changing the healthcare. In this talk I will discuss the current developments in my research program to assist decision making in healthcare. I will focus on research in sleep, diabetes, exercise, and ageing.
Bio: Maia Angelova is a data scientist and applied mathematician. She joined Deakin as a Professor of Data Analytics and Machine Learning in 2017. She was the founding director of Data to Intelligence research centre and now leads Data Analytics research lab at Deakin University. Prior to that, she was a Professor of Mathematical Physics at Northumbria University and Lecturer in Physics at Somerville College, University of Oxford, United Kingdom. Maia’s research is in data-driven modelling, time series analysis, machine learning, dimensionality reduction, dynamical systems, complex networks, automatic decision making, early warning signals. She develops hybrid models combining the dynamical properties of the complex system with the patterns hidden in the data. The focus of her research is precision medicine and decision making in healthcare and she is very interested in research translation. Her research was funded by several projects from Australian Defence, European Framework Programs, EPSRC, MRC, AMS. She has worked on applications in the domains of health, sports, defence and wireless communications.