The amount of data being digitally collected and stored is vast and expanding exponentially. As a result, the science of data management and analysis is also advancing, so much so that now data is not just analysed in retrospect but used and collected to become predictive. Computer scientists have invented the term big data to describe this evolving technology.
Big data has been successfully used by search engines to customise your searches; by political analysts to understand voting patterns of different demographics in political cycles; and especially by pioneers in healthcare. Google Flu Trends aggregated google search queries and made predictions about influenza outbreaks and activities in more than 25 countries (the service is no longer analysing the data itself but now providing data to institutions specialising in infectious disease research). Twitter has also been used for population-level influenza surveillance, along with understanding public sentiment towards vaccination, and investigating prescription drug abuse to state a few.
GV (formerly known as Google Ventures) is investing 36% of its sizable $2 billion portfolio in life sciences start-ups. The same trend is now starting to manifest in mental health as well thanks to wide-spread application and use of big data which along with the maturation of Natural Language Processing and Machine Learning technologies is offering exciting possibilities for the improvement of both population level and individual level mental health.
Pioneering work by Munmun De Choudhury and colleagues have applied computational methods to monitor population health and identify risk factors for individuals for a number of mental health issues using various social media platforms, including Twitter, Facebook and Reddit. For the population health domain, a crowdsourced data set of tweets derived from Twitter-users with depression-indicative CES-D (Center for Epidemiological Studies – Depression) scores was collected and then used to train a statistical Machine Learning algorithm capable of identifying depression-indicative tweets. These were then applied to geocoded Twitter data derived from 50 US states and the results correlated well with the Centers for Disease Control depression data.
For the identification of risk factors for individual domain , Twitter data was used again to investigate experiences of postpartum depression in new mothers. Birth announcements from public Twitter data using phrases such as ‘it’s a boy/girl’ were automatically identified and then the pre and post-birth Twitter feed of the new mothers’ was analysed. It was found that using Machine Learning techniques along with analysis of pre-birth behaviour patterns could predict postnatal emotional and behavioural changes with 71% accuracy.
The World Well Being Project is another example of the fascinating ways in which big data can be used to augment population level mental healthcare. One of the studies conducted used 148 million tweets along with the Centers for Disease Control mortality data and found a high correlation between words characteristic of negative emotions and heart disease mortality figures—more highly correlated than official socio-economic, demographic, and health statistics.
Big data is also being used to augment mental healthcare at a more individual level. Researchers at the University of Chicago are developing an app which monitors sleep and activity patterns to combat depression in university students. When the app picks up on behaviours that match certain symptoms of depression—such as irregular movement and physical activity, disturbed or abnormal sleep patterns, social isolation, drop in class attendance—it not only gives real-time suggestions and activities such as a counsellor would, but the data is also analysed and transmitted so that the university counsellor can keep a better track of more students and extend services to those in need.
In the world of individual therapy as well, big data is making an entry. Feedback informed treatments or FIT are psychotherapy metrics drawing on historical data to predict when clients are at a risk of deterioration. These metrics are surveys that clients fill in as a part of their therapy, detailing their progression through the weeks of therapy. The algorithm then predicts which clients are at a high risk of drop out, relapse in the case of substance abuse or ‘deterioration’ as otherwise specified. These metrics are said to work in two ways. One, it provides an element of blunt feedback that therapists often lack from their patients. Two, risk alerts allow therapists to adjust treatments and can compensate for clinical blind spots or overconfidence.
Despite the advancements in the field of mental health and big data, it is far from perfect. While there is an abundance of mental healthcare apps (honestly just type in depression into your app store and a list of hundred programs will pop up—and that's just for depression), they have come under scathing indictment from the American Psychiatric Association’s (APA’s) Smartphone App Evaluation Task Force . Studies by the APA and University of Liverpool have found that, despite their ubiquity, many of the smartphone apps have a lack of an “underlying evidence base, a lack of scientific credibility, and limited clinical effectiveness.”
Despite so, this burgeoning industry meets an important need of getting treatment to those with limited or no access to it. Given the pervasiveness of smartphones, these apps might serve as a portable therapist—particularly in rural and low-income regions. Public health organisations seem to be buying the concept and in its Mental Health Action Plan 2013-2020, the World Health Organization (WHO) recommended “the promotion of self-care, for instance, through the use of electronic and mobile health technologies” as a part of its agenda.
Given the personal and private nature of mental healthcare, it is difficult to imagine something as disconnected as big data anywhere near it. This is probably why the pace of acceptance of big data into mental health is glacial, especially by mental health professionals. A particular issue of serious concern raised is that of privacy. There is no real framework in place to ensure that privacy and confidentiality of records remain intact; nor are there any clear boundaries for ethical data collection.
A question of ethics is also raised when we think of the possibility of computer algorithms making clinical diagnoses or recommendations. The technology seems to be moving at a faster pace than law makers can keep up with. Despite the errors and risks, the big data analytics boom holds significant promise for understanding and improving mental health at both the individual and population level.
This article first appeared in Live Mint on Monday on 29 May, 2017, (http://bit.ly/2qGZp7W)