skip to primary navigationskip to content
 

Smartphones, Big Data, and Psychiatry

last modified May 20, 2015 05:43 PM
Conor Farrington, School of Clinical Medicine

Smartphones, Big Data, and Psychiatry

Dr Neal Lathia, Computer Laboratory, University of Cambridge

Dr Conor Farrington cjtf2@cam.ac.uk, School of Clinical Medicine, University of Cambridge

Abstract

The advent of smartphones and sophisticated wearable sensors, and the associated capacity to generate Big Data on individuals’ movements and behaviours, has significant implications for psychiatry. Clinicians can now supplement established sources of data relevant to diagnosis and intervention – including psychiatric assessments and patient self-reporting – with a wide range of additional data generated by smartphones, wearable sensors, and ecological momentary assessments (near real-time prompted patient self-reports). By providing clinicians (and patients) with new insights into patterns of mental illness and the links between these patterns and patients’ wider social behaviours, Big Data approaches have the potential to add breadth and depth to psychiatric practice while helping to overcome some of the challenges faced by established data sources.

We illustrate this potential through a discussion of Emotion Sense, a smartphone app designed to collect data about the associations between subjective wellbeing and various facets of behaviour. The app quantifies mood, a key aspect of subjective wellbeing, by asking users to indicate their current feelings by tapping on a two-dimensional affect grid measuring valence (from negative to positive) and arousal (from sleepy to alert); positive and negative feelings were also rated in greater detail (e.g. enthusiastic/calm for positive, anxious/sad for negative). On this basis, and also on the basis of a broader measure of life satisfaction provided elsewhere in the app, Emotion Sense computes a happiness score for each user. Physical activity is quantified by asking users to indicate, at randomly selected times, which active or inactive behaviours they have engaged in the past 15 minutes. In addition to self-reported activity, the app regularly samples from the smartphone accelerometer to gauge physical activity. In a study with 2,971 users who provided self-reports of activity in addition to responses for all happiness measures, we found that activity, which has been linked with poor physical health outcomes, also correlates with subjective wellbeing, and that sedentary behaviour is linked with lower subjective feelings of happiness. The information gleaned from apps such as Emotion Sense can be used to detect patterns that precipitate certain psychological states (e.g. cigarette cravings, depression) and then deliver interventions designed to prevent symptom onset.

While this study demonstrates the feasibility of using smartphones to conduct large-scale measurements, it is also important to recognise the multiple complexities that are likely to confront Big Data solutions in psychiatry. The presentation will conclude by considering some of these complexities, including: variable levels of smartphone usage among different population groups (and populations); resistance from stakeholders with vested interest in the status quo (including healthcare professionals); user concerns about surveillance, privacy, and data security; the potential impact of stigmas surrounding mental illness on the acceptability of mental health apps; and user concerns about technology replacing human contact in psychiatric care. In contrast to a technological determinist approach, which holds that the usage of new technologies can be predicted from their design, we suggest the need for a more sophisticated, sociologically-informed approach to the complexities of technology usage as it occurs in concrete social contexts. By taking account of such complexities, Big Data interventions in psychiatry will be better able to realise their significant potential to augment traditional models of care.