Earlier this summer I looked at the Moov Now wearable device that aims to take the growing trend for health and fitness related apps to a new level. In addition to monitoring what you do, whether on the road, in the pool or on your bike, it also comes with an AI ‘coach’ to both suggest improvements and chivvy you along.
The app is part of a growing realization that whilst there are many out there who are sufficiently motivated to keep up with their exercise regime, there are many who fall off the wagon. Finding ways to maintain motivation is therefore a major challenge for app developers.
Nudging People to Train
A recent Cornell based app is taking its cue from behavioral economics. The app, called MyBehavior, offers users a range of food and exercise suggestions each day, together with the outcomes they’d achieve (in terms of calories) if they achieved their target.
The app aims to provide sensible suggestions that blend with that persons lifestyle and habits rather than bombarding them with information. It also learns as it goes so will adapt as it learns your peccadilloes.
When the app was tested out on a small sample over a 14 week period, it emerged that users tended to eat less and exercise more each day when using the app. The numbers seem quite small, but when added up over the week resulted in 1,365 calories extra that were either burned off or not consumed.
Cheered on by the findings, the developers hope to roll out publicly in September, with a paper also due for publication that will document their pilot study.
In terms of its tracking capabilities, it is fairly run of the mill and tracks your runs, walks and even sitting time. Its algorithms will then generate various suggestions for improvements to your lifestyle.
For instance, if you largely have a sedentary lifestyle, it might suggest you get up and walk for even a few minutes every hour, before telling you how many calories you’d burn from doing that.
Users can log their food intake by snapping their meal with the camera on their phone and uploading it to the app. A group of Mechanical Turk users then label it and make a guesstimation for how many calories it contains.
This data is used to make suggestions for other (healthier) foods you might enjoy. There are often severe flaws in the calorie estimations from many wearable devices, with an over-estimation of calories burnt quite common, so it’s crucial that this is accurately achieved.
Whether taking a human based estimate will do this I’m really not sure about, so I have reservations about the success of this particular app.
What is interesting however is the growing trend for health related apps to learn our lifestyle and make small suggestions for how it can be improved. This seems a trend that will continue in the coming years.