What your phone could tell you about your health
By Gemma Milne
There's a whole world of untapped data sitting on our mobile devices. Could it be used to make us healthier?
It’s no secret most of us feel we spend too much time on our phones.
Scrolling through Facebook, double-tapping down Instagram, retweeting on Twitter – we create so much data on our likes, our lives and our livelihood that advertisers, insurance companies and retailers can read us like a book, by analysing our digital behaviour.
From tracking our emotions and mood through what’s called ‘sentiment analysis’, to simply knowing our physical movements through our GPS records, companies can sell us more relevant goods, and assess what we’re really like as individuals.
But what if our time online could be put to better use? What if our phones ceased to only be used for communication, and instead also became our very own medical device?
Paul Dagum is Founder and CEO of Mindstrong Health, a US company working on what’s called ‘digital phenotyping’.
The idea behind their work, is that the behaviour we exhibit while we use our phones day to day – from the speed at which we type and how fast our responses are, to which apps we use and when – can build a mood profile for each individual and track a person’s cognitive function.
They focus on three key areas. Firstly, your voice- tracking sentiment, mood and how coherent you are.
Secondly, human-computer interactions, meaning your swiping and scrolling activity and how often you tap and, finally, the phone’s in-built sensors eg tracking your location or how social you are based on messages in versus messages out.
By combining all these different factors, Mindstrong can paint a broad picture of you, and look for signs of negative mental health.
“We can look for changes in cognition, anxiety and depression,” Dagum notes, “and we can track day after day, at home, as opposed to in a clinic.”
Mindstrong are focusing on building a way for doctors to work with patients with chronic mental health disorders, to spot relapse and allow for earlier intervention with care.
Mothers who might be prone to postpartum depression, and people dispatched from trauma centres – who have a high risk of developing PTSD – are two groups they are looking to target first.
The way we swipe on our phones is one approach to building a digital health profile – another is looking at the actual content we post on social media.
One such example is a recent study looking into photos posted on Instagram, and whether a machine can diagnose depression from the filter chosen, the number of people present in photos and even the frequency of updates.
The study showed promising results, which begs the question of if and when this information should be passed on from the Instagram company computers, to a user’s doctor.
Facebook have had a long-standing relationship with the Samaritans, acknowledging that their users will at times post suicidal tendencies and other such worrying posts on their platform.
Together, they have created a way for friends to ‘report’ an individual’s post if they are concerned about their health, so a trained team gets in touch to offer help and support.
Of course, this requires a human to flag a post, but the efficient capability of the service to reach out to those in need suggests that, if we could prove what sort of phrases flags suicidal behaviour, a machine could surely do this unaided.
Dr Taha Yasseri of the Oxford Internet Institute, however, spoke of the limits of using social media alone to conclude how someone is feeling.
“You can’t get full passive data with social media – for Twitter, you would need people to be online and tweeting often enough to get full results.” There’s just as much, if not more, that we don’t post online, as we do.
Maxine Mackintosh, a data science doctoral researcher at University College London is working on another such example. She is investigating the link between data generated as a by-product of daily activities, for instance the number of touch points with your doctor over a 20 year period, and dementia.
“Something can be physiologically nonsensical, but also highly predictable,”she explains.
“How often you text, for instance, is not biologically linked to your dopamine levels, but it can be a marker or a representation of change.
It’s exactly this concept which drives researchers like Maxine to seek out ways of spotting concerning health issues using the data from our phones.
With such huge amounts of data already being collected by phone companies, social media platforms and internet providers, you would be forgiven for thinking we already have enough information to work out what patterns predict which health issues.
The difficulty comes down to ethics – who has the right to the data, who can properly analyse it, and do we even want these companies to know what’s going on in our bodies in the first place?
Recently, Strava released all the location data of their runners online, and it didn't take long for people to work out where secret US Army bases were all over the world, as a result of the open nature of the data.
One example of why we might want to keep our health data private, is that insurance companies and future employers might – deliberately or unconsciously – make judgements on your abilities based on your past health record.
But when you consider how much time and energy we spend with our phones by our side, putting our digital data to better use seems like an inevitable no-brainer.
And when you start to include your fitness tracker data, your existing doctor’s health records and any other kind of data you record about your health, you start to see how strong a picture of a person we can create with arguably little effort – as long as we’re careful about how we manage that information.
After all, if advertising companies are already using this data to better sell us stuff, we might as well find ways to use our own information to better our wellbeing.