Working out better e-health

More and more people are using e-health applications, varying from websites to wearable wristbands. The enormous amount of data gathered with these applications has the potential to not only change individual health and behaviour, but also the way health care in general is organized. Postdoc researcher Liseth Tjin-Kam-Jet – Siemons and PhD-student Floor Sieverink run a project that aims to chart this potential.

‘Take, for instance, one of the wristbands now available’, says Floor Sieverink. ‘It will gather data on your movements, your heartbeat, blood pressure and sleeping pattern. You may have an app that records your eating habits and your weight. That’s a lot of data. You may simply present the results to the user, but it’s also possible to give advice, like: this week you haven’t been exercising as much as usual, shouldn’t you go out for a walk? It may, based on all these factors, even be able to predict that you are at the point of falling ill.’

Chronical conditions

A slightly raised body temperature and heartbeat, for example, may go unnoticed by the wristband’s wearer, whereas they may herald an emergent flu. So, the app may issue a warning about these symptoms, but in what way? Just saying “in two days you’ll be in bed with a fever” will not do, but neither will producing a lot of data and letting the user himself figure out what to do. Ideally, to maximize its benefits to the owner, an app should present its findings in a personalized way.

Such applications may help healthy people to stay in shape, whereas people with chronical conditions such as diabetes may find in it a coach that persuades them to move into the right direction. ‘If you see your doctor four times a year, you must monitor yourself most of the time’, Liseth Tjin-Kam-Jet – Siemons explains. ‘E-health applications are there to support you. However, in practice we find that patients often stop using them after some time. We’re interested in finding out why, because this will give clues how to design applications that perform their coaching role more successfully.’

For their research Tjin-Kam-Jet – Siemons and Sieverink have acquired several datasets. Among those is navigation data from a web portal that supports self-management by patients. Sieverink: ‘This dataset tells us how often people log in, how they navigate through the site, which exercises they take and so on. We use this data, along with real life information from the user’s ailment history, to analyse which behaviour of the portal is most relevant. In turn, our analysis may for instance lead to a redesign of the website or a different navigation of the site for different types of users. If you know at what point users tend to quit an application, you may be able to prevent them coming to this point or, if they do, intervene in order to persuade them to carry on anyway.’

Expert panel

With the benefits come issues as well. That handy wristband, for instance, will send its data to the manufacturer (which users may not realise, as they don’t read the often sketchy terms of service they agree to). This is necessary for the equipment’s future development, because the algorithms will get better if they can be fine-tuned and tested using lots of data.

Other issues may even be more complicated. Can insurers be allowed to reward or penalise their customers depending on the healthiness of their behaviour, as recorded by an app? Given that the training algorithms are often opaque, even to their developers, to what extent is it safe to trust an app’s recommendations?

Taking on such legal and ethical considerations at an early stage is important for a successful e-health application, as any objection may be a reason for users to refrain from using it. To address these issues Tjin-Kam-Jet – Siemons and Sieverink cooperate with an expert panel that feeds them with insights from as many disciplines as possible.

Tjin-Kam-Jet – Siemons: ‘After all, our goal is to acquire information on the way people actually use and experience e-health applications, and combine this with technological, medical, ethical and legal knowledge, with the ultimate goal of developing strategies that lead to better applications.’

Geschreven voor een brochure van de Universiteit Twente