A recent study has revealed that patterns of mobile phone use can highlight intriguing aspects of human behaviour – even down to the spread of diseases. The implications of this discovery are potentially enormous, but is society ready for such high levels of surveillance?
The ability to detect outbreaks of illness as they emerge is crucial in the battle to combat the spread of disease. But, as the swine flu pandemic made clear, tracking infectious illnesses can be a nightmare for health authorities.
Conventional methods of monitoring the spread of disease depend on reports from doctors and swab tests by public health laboratories. But, by the time reports are filed and test results are collated, it’s usually too late. It can take days from the first snivel to a firm flu diagnosis – days in which disease can spread.
What’s always been clear is that reasonable predictions, rather than firm diagnoses, are far more useful in the quest to monitor and control the spread of disease. But making those predictions has always been a matter of informed guesswork – until now.
New research reveals that smartphones, rather than stethoscopes, could provide a faster and more accurate indication of who is ill and how illnesses spread. The findings are contained in a study that used “socially aware” mobile phones to monitor students’ behaviour at the Massachusetts Institute of Technology (MIT). “This technology could be an early-warning system to enable us to spot outbreaks of influenza,” says Anmol Madan, a human dynamics specialist at MIT and author of the study.
Madan and his team studied behavioural changes that could be “sensed” and then reported by mobile phones. These included the extent to which people moved around and how often they used their phones.
The study was conducted as part of a much wider investigation into the way that behaviour and opinions spread in social groups. In the research, 70 residents of an undergraduate residence hall at MIT were given phones containing software that gave the researchers data on personal movements, as well as communication patterns, including calls and text messages.
Students who became ill had a tendency to move around less, making fewer calls late at night and early in the morning. The software used to monitor these variables was able to identify flu sufferers correctly nine times out of ten in daily checks.
It’s long been recognised that illnesses such as influenza have a distinct behavioural and geographical fingerprint. But, until now, measuring these changes reliably has been impossible.
Mobile phones provide a unique window on the world of individual and mass behaviour in a way that no other technology currently can. Collecting and analysing data of this type with the aim of predicting behaviour is described as “reality mining”. But this facet of the mobile revolution is still in its infancy.
Localisation technologies, which include the hardware, software and back-office systems needed to pinpoint an individual or monitor large groups, are the key to gathering fine-grained data needed to create what Madan calls a community microscope.
“Our ability to understand social interactions will also improve with advancements in sensing hardware,” explains Madan. “For example, GPS and assisted cellular triangulation are better localisation technologies than the 802.11 WLAN APs used in our study. The resolution of face-to-face proximity measurements will improve if infra-red sensors become common on smartphones.”
But, even without pinpoint localisation technologies, mobile phones still have the potential to reveal huge amounts about who we are, what we do, where we do it and when.
There are about 5.6 billion mobile phones in use worldwide, with handsets handling voice, text and e-mail, as well as serving as address books, diaries and more. The incorporation of everything from cameras to accelerometers means that today’s smartphones also have the potential to act as real-world sensors.
Crucially, the technology needed to monitor and potentially capture all of this data is no longer a limiting factor. It’s already possible to build an app that can log and share your phone calls, e-mails, texts and internet browsing history, couple it with GPS and then display it on Google Maps, along with streaming video and audio from your smartphone, for the entire world to see. Somebody might even pay you to run such an app on your phone.
But disclosure on this scale is unlikely to prove attractive to phone users. What the MIT project reveals, though, is that even relatively small amounts of individual data, such as movement and call volumes, can prove remarkably revealing.
Studies in the US and Switzerland have focused on the potential of mobile phones to identify the repeating structures that underlie the principal components of an individual’s behaviour. These are known as Eigen behaviours, which can include activities such as “sleeping in late”, “working at home” or “going out on the town”. Critically, research shows that only six Eigen behaviours are needed to approximate subjects’ behaviour with 90 per cent accuracy.
But some experts warn that there are risks in drawing too many inferences from mobile phone data. Studies of mobile use following cholera outbreaks in Rwanda revealed a clear reduction in population movement. This could have been caused by illness or by the floods that triggered the outbreak in the first place. Correctly identifying the cause of this reduced mobility was impossible.
Risks aside, the prospect of being able to determine what people are doing – and what they’re likely to do next – is hugely attractive. Disease prevention and public health are obvious applications, but the implications for business are equally significant.
Sense Networks is one of a new generation of businesses that are capitalising on the commercial potential of reality mining. Founded by computer scientists from MIT and Columbia University in New York, Sense Networks harvests movement data in real time from devices with GPS or Wi-Fi positioning technology to index and rank the popularity of various locations. The company’s CabSense application for smartphones, launched in 2010, addresses the perennial problem of never being able to find a taxi when you want one. The application directs users to the best places for finding available cabs and provides a location rating from zero to five stars.
Developments of this sort highlight the fact that the ability to deliver value-added services lies not in smartphone handsets or vehicle-tracking devices, but in intelligent back-office systems supported by increasingly powerful algorithms. Location-based services are a case in point. These tap into the potential of GPS-enabled handsets to know where they are. That information can be used to alert other users to your whereabouts. And it makes it possible to access data linked to that location.
“Location-based services identify where a mobile phone is geographically located and use that information for some purpose, such as finding a restaurant or checking the weather,” says Juan-José Juan, head of innovation at Vodafone. “These services will revolutionise the way people interact with urban spaces.”
It also allows service providers to push location-relevant information (specifically, advertising) to phone users. Many of these developments are being driven by the migration of social networking from conventional computers to mobile devices.
Facebook is already seeking to capitalise on this trend with Places. This works by getting smartphone users to tap a check-in button to see a list of places near where they are and then to choose a place that matches their precise location. Users can also automatically let people know where they are by tagging other Facebook friends, but only if they check-in to the same place too.
Location-based services are still in their infancy, but interest is growing fast, thanks to the prospect of targeted advertising and promotions based on proximity. According to a study by Borrell Associates, a US-based media consultancy, services of this sort could generate up to $4.1bn in advertisement sales annually by 2015.
Getting access to the personal data needed to make predictions or provide advertising raises some fundamental questions. Far from conjuring up visions of a utopia where epidemics are nipped in the bud, or where a taxi is never far away, some people fear the emergence of a surveillance society where the individual’s every movement is logged.
This was an issue that Madan’s team at MIT encountered early on in their research. Non-participants were given the opportunity to voice their concerns about the possible privacy implications of the study. Not all of them were happy about this. “I personally have problems with people who don’t live here leaving things in the dorm,” wrote one respondent. “Especially on a long-term basis, especially without permission, especially if they’re trying to ‘study’ us.”
The idea of “them” studying “us” is an emotive one and makes some people uneasy, no matter how beneficial the promised outcomes. And it’s something that all players in the communications business – academics, telecoms providers, businesses and governments alike – are going to have to face.
It’s a concern that Facebook is already addressing with its “Places” applications. Users can control how much information they share through the service and can remove themselves from it if someone tags them. And, if they lose interest, they can disable the application.
Vodafone’s Juan has some further words of warning: “Location apps are fun to use, and it’s nice to be able to see where your friends are, but announcing to the world that you’re in a particular place can serve as an invitation to stalkers or a tip-off to burglars. It’s important to be sensible with new technologies.”
The question of sharing personal data remains a challenging one. According to a recent survey, 79 per cent of Americans want to keep the files they store on their computers private. About 50 per cent of those surveyed said they would be embarrassed about friends or family seeing certain files on their computer or smartphone. And then there’s the question of the “creepy line” raised by Google’s CEO, Eric Schmidt, and reported widely in the press.
His comments refer to the perceived borderline between openness and intrusion in the matter of personal data.
Social acceptance is a moving target, though. Technologies that were once considered “creepy” – automatic number-plate recognition, for example – are now ubiquitous and generally accepted by the public. As new social norms evolve, it’s becoming clear that what’s considered creepy today could be standard fare five years from now.