ESRA 2019 Draft Programme at a Glance
Novel uses of wearables, sensors and mobile apps to collect health and environmental data in survey research.
|Session Organisers|| Dr Heidi Guyer (University of Michigan Survey Research Center)
Ms Carli Lessof (National Centre for Research Methods, University of South Hampton)
|Time||Friday 19th July, 09:00 - 10:30|
Survey research is rapidly changing in part due to the proliferation of new and emerging technologies. This session will discuss the growing interest in incorporating wearables, sensors and mobile apps to collect in-depth, repeated and objective measures with a specific focus on health and environmental measures.
The use of mobile apps, sensors and wearables to collect data allows for measures that often cannot be gathered using traditional data collection methods or involve high levels of cost or burden. Wearables, sensors and apps offer the possibility of collecting data in the moment, including ecological momentary assessment, as well as offering both objective and subjective measures collected at one point in time, at several points in time, or continuously over a defined period of time. Data collected using new technology can represent various measurement levels including the individual, household, neighborhood and the general environment. Additionally, interventions can be targeted based on data derived from sensors and wearables, thereby both collecting data and providing targeted prompts over time.
Examples of wearables used to collect health data include actigraphy, fitness trackers, sleep trackers, heart rate and blood pressure monitors. Sensors are available to measure individual behaviors as well as areas of daily living including air temperature and quality, environmental noises and sound quality, and intelligent in-home equipment such as smart refrigerators which track the availability or shortage of food items. Mobile apps often take advantage of the technical capabilities of smartphones and tablets, for example cameras and GPS, in combination with the ability to deliver online surveys or trackers of health behaviors.
We invite submissions addressing the following topics:
• Novel uses of wearables, sensors and/or mobile apps
• Participation rates and response bias; compliance rates over time
• Considerations regarding providing standardized equipment versus requesting consent to link to data from devices owned by the participant (including burden, compliance, cost).
• How do objective measures from wearables and sensors compare with standard or subjective reports?
• Measurement error resulting from the type of sensor or wearable used, its location, timing or frequency of measurement or other factors
• Costs and practicalities of developing and incorporating wearables, sensors and apps in surveys
• Data issues including accessing commercial datasets, ensuring data security, managing data sharing, data management challenges arising from the complex data structures and large volumes of data
• Ethical considerations related to collecting data using wearables, sensors and mobile apps for research purposes.
Keywords: sensors, wearables, mobile apps, measurement, data collection
Integration of Surveys, Wearables and Mobile Devices
Dr Arie Kapteyn (USC) - Presenting Author
Ms Jill Darling (USC)
The USC Center for Economic and Social Research (CESR) is conducting a series of pilot studies aiming to develop procedures and best practices for collecting data on the daily lives of older Americans in unprecedented detail and frequency, making use of opportunities offered by advances in technology.
Project participants are members of CESR’s Understanding America Study (UAS, https://uasdata.usc.edu/index.php). The UAS is a probability based internet panel of approximately 7000 respondents. Respondents are interviewed about once or twice a month about a variety of topics, resulting in a very large set of background characteristics for all panel members, including demographics (e.g. age, gender, race), financial situation (e.g. income, wealth, retirement savings), health (e.g. self-assessed health, physical measures, health behaviors), personality traits (the Big Five) and cognition measures (e.g., number series, propositional analogies, picture vocabulary).
The current project will maintain a subsample of the UAS of approximately 1000 individuals, who will respond to burst designs that will include weeklong EMAs, wearable devices, and recordings of daily narratives. To understand how respondents experience events and how they perceive their potential for coping, participants will consent to wear devices that measure physical activity and biometrics, and respond several times per day over a one-week period about their current mood and social interactions. Respondents also record short narratives on their smartphone at the end of the day. The narratives are transcribed and analyzed using Linguistic Inquiry and Word Count [LIWC, Pennebaker, J.W., Boyd, R.L., Jordan, K., & Blackburn, K. (2015)].
We have conducted several pilots (with sample sizes on the order of 50-100), including the development of a mobile app that can be loaded on participants own mobile devices for this purpose. We will report on the results of three full pilots of the burst process, including survey and mobile app
Accelerometer measured versus self-reported physical activity: Evidence from two population based birth cohorts
Dr Lisa Calderwood (UCL Centre for Longitidinal Studies, University College London)
Mr Robert Jones (University of Nottingham)
Dr Emily Gilbert (UCL Centre for Longitudinal Studies, University College London) - Presenting Author
Mr Matt Brown (UCL Centre for Longitudinal Studies)
Professor Emla Fitzsimons (UCL Centre for Longitudinal Studies)
Professor Alice Sullivan (UCL Centre for Longitudinal Studies, University College London)
Professor George Ploubidis (UCL Centre for Longitudinal Studies, University College London)
Professor Mark Hamer (Loughborough University)
Inactivity is known to be associated with increased risk for chronic diseases and premature mortality. Data collection at the population level usually involves self-reported measures of Physical Activity (PA) but increasingly objective PA measures are used. Self-reported measures are likely to be biased from measurement error due to issues of recall and response bias or some process that is driven by the respondent’s personality and circumstances. Objective measures may also be affected by device specific errors, but are generally believed to offer more precise estimates of PA and remove many of the issues of recall and response bias. However, they generally have more missing data due to non-compliance and this selection may lead to bias.
We use data from two UK based birth cohort studies that have recently collected accelerometer data. The Millennium Cohort Study (MCS), that collected data with the wrist worn GeneActiv at age 14 (n = 3652), and the 1970 British Cohort Study where the thigh worn ActivPAL was employed at age 46 (n = 5252). In MCS, we use both self-report questionnaire data as well as contemporaneously collected time-use diary data from the same days the accelerometers were worn. We examine whether the association between a wide range of socio-economic and demographic factors and PA is dependent on measurement method, whether these factors are associated with non-compliance as well as the extent to which they explain differences between self-reported PA and accelerometer measures. Preliminary results show that a worryingly low proportion of 14-year olds in the UK are not meeting current PA recommendations and that females and children from a Southern Asian background are typically more inactive than their peers.
What Really Makes You Move? Identifying Relationships between Physical Activity and Health through Applying Machine Learning Techniques on High Frequency Accelerometer and Survey Data
Mr Joris Mulder (CentERdata - Tilburg University) - Presenting Author
Dr Natalia Kieruj (CentERdata - Tilburg University)
Dr Seyit Höcük (CentERdata - Tilburg University)
Mr Pradeep Kumar (CentERdata - Tilburg University)
Physical activity is an important indicator of health, but an objective measurement is needed to gain insight and understanding of what drives differences in physical activity and how this influences health. Existing studies are generally based on self-report surveys, but there are limitations to their use, e.g. varying perception of physical activity, social desirable answers, and incomplete recall of activity. Wearable accelerometers as a measurement device provide a more complete and objective picture of physical activity and opens up new ways to study the relationship between physical activity and health. In addition, the influence of socioeconomics, -demographics and personality traits can be taken into account when studying these relationships.
To study these relationships in detail, an experiment using accelerometers was conducted in the Dutch LISS panel. 1.000 panel members were asked to wear an accelerometer for 8 days, measuring their physical activity level day and night. During this period, respondents filled out surveys indicating what specific activities they conducted during the day. Furthermore, they provided details about their sedentary behavior, perceived health, social context, and their associated mood.
First analyses of the objective accelerometer data and the subjective self-reported data showed how people tend to over-report physical activity compared to the objective measurements from the accelerometers (Kapteyn et al., 2018). In a follow-up study we take the analyses of these high frequency data a step further by applying data science methods. Using machine learning techniques, such as deep (convolutional) neural networks for pattern recognition, we are able to identify specific physical activity patterns, i.e. walking, running, cycling, sitting, sleeping. Combining this with longitudinal survey data from the LISS panel we obtain relationships between physical activity and health on a detailed level. In essence, we gain insight in the relationship between the specific identified activities and personality traits, health, and socioeconomic and demographic status.
Approaches to research health inequalities with Big Data
Mrs Zora Hocke-Bolte (PH Schwaebisch Gmuend) - Presenting Author
There is a large amount of data, which can be used to research about health inequalities. This includes, for example, health data from governmental Open Data-Tools, Social Media, data from sensors or self-collected data by affected persons.
This presentation intends to provide an overview about the potential approaches to analyze the data treasures found in big data, to facilitate a deeper understanding of health inequalities and to improve health promotion and prevention. The primary focus will be on large and unstructured data which couldn’t be analyzed by previous methods. This will be supported by case studies on the results of big data research projects about the combination of different data sources like social media or environmental data.
The connection between social and health inequalities will be the key aspect. Furthermore, the purpose is to fill research gaps in the field around the digital behaviour of persons with chronic illnesses and their eHealth Literacy.