Supervisors; Dr Nicola Perra, Dr Li Zhou, Dr Luis E Rocha, Dr Vittoria Colizza
The project aims to characterise behavioural changes induced by seasonal influenza on mobility patterns and physiological indicators. The project will also study well-being and healthy ageing from digital indicators.
We will use a unique anonymous dataset collected by Nokia from the activity trackers (e.g. fitbit) of 150K people in 20 cities of 12 countries over 3 years. The high-quality data provide, the number of steps, heart beats and sleeping patterns at daily resolution for each individual. Furthermore, the data provides demographic information such as age and gender.
Hypotheses H1: Seasonal influenza affects mobility patterns and physiological indicators at individual and population levels.
H2: Activity trackers can be used to infer behavioural changes and improve models of seasonal influenza.
H3: Well-being and healthy ageing can be inferred and monitored from the activity trackers.
Methodology and Innovations We will use machine learning methods to identify periods of significant changes in the mobility behaviours of individuals via trackers data. The large temporal and geographical coverage of the dataset will allow us to study the extent to which such changes can be linked to the flu progression. We will consider physiological indicators to infer the level of stress and immune responses against infections. Furthermore, demographic information will be used to further characterise the changes in behaviours.
While local variations might take place throughout the year for a variety of reasons, major events such as the annual epidemic season of influenza likely trigger correlated dynamics visible at the population level. The project will first identify and characterise such dynamics.
The project will leverage these outcomes and develop innovative data-driven compartmental epidemic models to account for behavioural changes. We will propose different mechanisms (models) to connect variations in mobility and in physiological indicators with epidemiologically-relevant dynamics, i.e. changes in contact rates.
Each model will be empirically parameterised using tracker and flu data. Model selection methods will identify the best model/mechanism(s) which will be adopted to forecast the progression of the influenza season and to establish a causal link between behavioural change and epidemic spreading.
Using machine learning approaches we will also study the link between physiological and demographic indicators to identify signs of well-being and healthy aging across time and the various countries/cities we have access to. In doing so, the project will capture and investigate the digital bio-rhythms of very different populations, at scale.
Applications Applicants must apply using the online form on the University Alliance website at https://unialliance.ac.uk/dta/cofund/how-to-apply/. Full details of the programme, eligibility details and a list of available research projects can be seen at https://unialliance.ac.uk/dta/cofund/
The final deadline for application is 12 April 2019.
DTA3/COFUND participants will be employed for 36 months with a minimum salary of (approximately) £20,989 per annum. Tuition fees will waived for DTA3/COFUND participants who will also be able to access an annual DTA elective bursary to enable attendance at DTA training events and interact with colleagues across the Doctoral Training Alliance(s).
This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 801604.