maandag 18 januari 2021

What is a #healthylifestyle to Instagram users?

What is a #healthylifestyle to Instagram users?

Health is a prominent topic on Instagram and ample research is focused on the opportunities of Instagram for health promotion. In this post, I seek out to find what Instagram users associate a healthy lifestyle with. Using network science, I analyze 5835 Instagram posts that contain the hashtag #healthylifestyle. More particularly, I will draw a network between all hashtags included in these posts. The nodes of the network are the hashtags mentioned in the Instagram posts I collected. The edges between these nodes are co-occurences between the two hashtags in an Instagram post. The majority of the code used for data collection and analysis was borrowed from the splendid post of Sam Ho on Using Network Science to Explore Hashtag Culture. The approach followed here is an excellent way to explore hashtags and their association on any possible topic. 

Long story short, I collected 5835 Instagram posts that contain the hashtag #healthylifestyle, drew a network of the links between the hashtags in the posts and used community detection to group hashtags that frequently occur together. The network looks like this (hover over the nodes to see the hashtag associated with them or select parts of the network you want to check out in more detail):


Four communities of related hashtags can be distinguished on the graph.The more central on the graph a node (~hashtag) is positioned, the more central it is on the network, meaning it has many connections to other nodes in the network. The size of the node represents its frequency of occurrence (the bigger, the more frequent, obviously). 

To explore the four communities of hashtags that are identified in the network, let's look at the following sunburst plot: 


The plot shows 4 communities or clusters of related hashtags associated with #healthylifestyle. Cluster "0" contains tags related to healthy food, healthy eating and related tags. Cluster "1" contains tags related to fitness, gym, exercising & motivation. Cluster "2" consists of tags related to weigh tloss and "slimming" or "slimmingworld", an online weight loss program/community. The last cluster contains tags related to yoga, mediation & mindfulness. 

So in conclusion, 4 clear categories of tags related to health behavior can be discerned in a network around #healthyliving. When Instagram users think of healthy living, they think healthy eating, physical activity, weight loss and a healthy mind.  

As I said, this tool can be used to explore any hashtag and its related network. This is only a first exploration of health related hashtags. I hope to find the time in the coming months to further explore this.

Let me now what you think and feel free to post suggestions to take this research further! Also, if you want more info, let me know!



donderdag 28 mei 2020

Sports app use during the Covid-19 pandemic lockdown in Flanders, Belgium

Last week, an article on the Belgian sports news channel Sporza (here & here) reported on the increase in sports app use in Belgium, using data collected with our MobileDNA app (info &  join the study). More specifically, we looked at the use of the social fitness app 'Strava' and saw that in comparison to the same period in 2019, Strava use increased substantially in our sample. While Belgium is experiencing an unusually dry and warm spring, the increase in Strava use was clearly not solely due people exercising more because of the warmer weather. In follow-up to the study of the Strava data, we decided to look at the use of a few other sports apps as well:
  • Garmin Connect: dedicated app for Garmin devices
  • Polar Flow: dedicated app for Polar devices
  • Start2run: paid running app for beginners
  • Runtastic: exercising app
  • RunKeeper: exercising app
The figure below shows the app use for each of these apps, comparing 2019 with 2020 for February 1st until April 30th (note: figure shows actual datapoints with GAM smoothing spline + 95% confidence intervals). Note that the figure shows the proportion of MobileDNA loggers who used a given application on the dates indicated on the X-axis. This usage can be a number of things: logging a cycling, running,... or other types of workout, but also checking their profile page or timeline (when applicable), checking activities performed by others on their friends list (when applicable), giving likes or comments (when applicable)... Therefore, the data should not be solely interpreted as the specific tracking of exercising activities (although that will be the case to a certain extent), but more generally the use of the app. 


What can we conclude from the plot? First of all, it seems that the daily app use of Strava, RunKeeper, Runtastic and Start2Run increased during the lockdown (note: the number of Runtastic and Start2Run users in our panel is low, so we have to be cautious when interpreting their curves). All of these apps are designed to track exercising activities without need for a dedicated wearable or fitness tracker. The first three don't require subscription for the basic version. Start2Run is a paid service. 

For Garmin Connect and Polar Flow, this increase it not as noticeable, even though both companies notice longer and more activities by their users during the lockdown (Polar - Garmin) . The use of the latter two however, is strongly related to the ownership of either a Garmin or Polar wearable device. It could be that our data mainly reflects the app use of those who already owned either a Polar or Garmin wearable at the start of the lockdown (which may not differ substantially from last year, especially when they were already training at a high frequency). Especially for Polar we have to note the small proportion of users in our dataset, which makes it more difficult to draw conclusions. I do assume that both companies (and other wearable manufacturers) may benefit from the increase in the use of sports apps like Strava of RunKeeper. After all, if (hopefully) people enjoy their new experience of running, cycling or other exercising activities, and they decide to keep up their new habit, the purchase of a wearable device is not unlikely. 

And the winner is... Strava? From the plot above, it is clear that Strava experienced a significant increase in use during the lockdown. To me, this comes as no surprise given its focus on creating dedicated social network for (recreational) athletes (see my other blogposts or thesis). Especially during the lockdown, it was a perfect platform to record physical activities and maybe more importantly, to share them with others, with whom our physical contact (including running or cycling together) was limited. Strava, more than any other sports app, has been a leader on this aspect. Given that this element remains free, I expect no major impact of the new paid subscription plan Strava introduced recently (read). 

What's next? In the next months, it will be interesting to see the evolution of the daily usage of the sports apps we looked at here. The most important question is whether this 'lockdown effect' will persist or whether we will see a substantial decline again towards what we saw in previous years. After all, new habits are hard to stick to... New blogpost on that will follow in the next weeks/months. 

Feel free to share your thoughts & comments!

Best,
Jeroen




maandag 11 juni 2018

Our study on Markov Chain analysis of mHealth app data wins ISBPNA Best Conference presentation in e&mHealth!

Our study on Markov Chain analysis of mHealth app data wins ISBPNA Best Conference presentation in e&mHealth!

Last week we attended the annual ISBNPA conference in Hong Kong, a must for physical activity and behavioral nutrition researchers. The program was packed with interesting presentations of recent e&mHealth research. For the full conference abstract book, check here.

Among these studies, our research on assessing how user navigate within mHealth apps, with a case on our own Start2Cycle app, won the Best Conference presentation in e&mHealth award. A great honour and acknowledgement for the steps we are taking towards making more engaging e&mHealth applications!

What did we do? We started from the need in mHealth research to have more profound insights into the engagement of users with mHealth interventions and more specifically, the engagement with the apps, games, websites used in these interventions. These tools produce lots of data, but analysis of these logs often remains limited to descriptive statistics, such as at what points in time an app is used, how often a page is viewed and so on. We wanted to go a step beyond this descriptive analysis evaluate the applicability of Markov Models to assess “how” people navigate within certain applications and whether this is how we intended them to be used. As a first case study, we took the data of our Start2Cycle app, developed in the Imec-ICON project “CONAMO (https://www.imec-int.com/nl/imec-icon/research-portfolio/conamo)”.

The app had a straightforward goal: two teams, one month of cycling, the team with the most km’s wins! With the app, the users could log their rides and check how their teams were doing. Moreover, they could see how they were doing personally within their own team. Per 100 km’s cycled, they could collect trophies (‘badges’) (audiovisual content from TV series provided by our public broadcaster partner VRT). In total, the app contained 8 pages, which are depicted below:



22 users participated and all details of their sessions were logged. On these data, we then performed a Markov Chain analysis to see how they navigated between the 8 pages. This type of stochastic modeling allows you to identify patterns or make predictions based on the data you have. For us, this implies that we wanted to know how our users navigated within the app. The analysis allowed us to answer various questions on the usability and general use of our app. For example: what is the chance that when a user has logged a ride, he/or she will then navigate to the ‘competition page’ to see how this ride aided his/her team in the race with the other team? Or: when a user opens the app, what is the most likely path he will take within the app until exit? From which page do they generally exit the app?

The analysis revealed two most likely paths in the app: a tracking session and a gamification sessions. The two main purposes of the app clearly stood out in the visualized Markov Chain. Moreover, by applying a sequence clustering algorithm, this insight was confirmed in an automatic fashion. Great to see! Of course, the app in itself is quite straightforward, but we would like to do more of these and other analyses on complex log data of e&mHealth apps and games. This was just the first (baby)step of what we want to do in the future. So more coming up!

Data!
One more thing! For those interested in some hands-on experience, we are currently developing a tool to generate a visualization of your data using Markov Chain analysis. Check out: https://ibcnservices.github.io/MCAppAnalysis/


Sample data (your data should be in the right format) of our app, as well as the python code for the Markov Chain analysis can be found on https://github.com/IBCNServices/MCAppAnalysis. Using your own data works as well, as long as it complies to the pre-defined format (CSV file with either 2 columns (from & to) or 4 columns (from, to, session_id and timestamp)). 

Let us know how it went or tell us if we can help!

The ISBNPA 2018 presentation and abstract can be found here and here. Stay also tuned for our journal paper on this research coming up!

You can always contact us for more information:


Acknowledgement
This study originates from the CONAMO project, funded by the Imec ICON research program with project support of Vlaio. Companies and organizations involved in the project are IMEC, imec-IDLab, imec-MICT-Gent University, imec-SMIT-VUB, imec-Living Labs, UGent-Victoris, VRT, Energylab & Rombit. Special gratitude goes out to public broadcaster and project partner VRT and its team for the development of the app, especially Matthias De Vriendt for outlining the raw data structure and format. Gilles Vandewiele is funded by a PhD SB fellow scholarship of FWO (1S31417N).

donderdag 26 oktober 2017

About Speedo On...

I recently came across the Speedo ON platform and I liked it. I decided to dedicate a few words on it on this blog.

I look at these platforms from the perspective of the theoretical framework I developed in my PhD (see figure). In short, the framework contains the essential elements for online data analysis platforms to be 'engaging', which implies that the user enjoys the platform and finds it worth it to keep using it. This engagement has a strong link to self-determined motivation and implies that for a data analysis platform to be engaging, it should nurture the user's sense of autonomy, competence and relatedness. For a person to be motivated to perform certain behavior, he or she wants to feel in control of the behavior (autonomy), wants to feel competent and wants to do it together with others or have some connection to others performing the same behavior (relatedness).

Platforms can address these basic psychological needs through various types of platform features, such as leaderboards or the ability to connect to other athletes. Through these features, a platform has certain 'affordances' to assist in the user's motivation, also termed 'motivational affordances'.

                            

In order to address all 3 basic needs (autonomy, competence and relatedness), a platform should have self-regulatory features, social interaction features and gamification. In my PhD, I largely illustrated the model by the example of Strava, a platform that is strongly used by cyclists and runners (and multisports athletes). Strava has elaborate self-regulatory features such as progress monitoring, goal-setting and data analysis and visualisation. In fact, most platforms like Strava (e.g. Garmin Connect, Endomondo, RunKeeper) have these self-regulatory features. Where Strava excels is the social interaction and gamification (see my other blogpost on this: https://jeroenstragier.blogspot.be/2017/10/what-can-we-learn-from-strava.html). These features allow competition, interaction, connecting between athletes and afford the user to have the sense of being part of an 'online sports team', which is nice to have for people who practice sports that are very often performed individually.

Now, I was going to talk about Speedo ON. Why do I believe that it has the potential to become the go-to platform for swimmers?

First and foremost: it focuses on swimmers. I have not yet seen a platform focusing solely on swimmers. Yes, of course, there's lots of devices that can track in and- outdoor swims and you an upload them to the platforms of these devices (like Garmin Connect or Polar Flow), and you can then sync them on Strava, there's lots of swims on Strava right? Yes. But are these swim from swimmers? Or multisports athletes? I think it is largely the latter. Strava is above all frequented by cyclists, runners and multisports athletes. Speedo ON is the first platform that focuses on swimmers only.

Second, upload from various devices: quantification of behavior through wearables is the baseline of my model. Being able to only see your data on the platform that goes with your wearable implies that when you switch brands, you have to use another platform. I know, Strava and others already allow this, I just want to point out that Speedo chose the right path by not wanting to develop their own tracker and keep the data to 'Speedo only platform'. Right now, there's only Garmin and Misfit that can be connected, but I expect others to team up quickly.

Third: social interaction. ON affords users to connect to other users (follow and be followed), comment each other's activities. A crucial element from the perspective of a user's need for relatedness. Swimming is a very individual sport and the social interaction on ON can generate a sense of community much appreciated by swimmers.

What I don't see is elements of competition on the platform. I don't think that this is crucial, but it has its merits in terms of motivation and engagement, depending on the type of swimmers that will frequent the platform. I realize that a concept of 'segments' like it is used on Strava can be copied to  swimming, but I think there are opportunities.

It's (to a certain extent) all there. Now, all it needs is critical mass: I'm curious to see how the swimming community reacts. Do they seek these kinds of communities? Is there a need for a platform like this? We'll see I guess. Anyway, I think it's a nice platform and nice move by Speedo. It will generate lots of useful data for them.

Greetings!
J.

*Note: I have no connection to Speedo whatsoever :- ).

woensdag 18 oktober 2017

Strava launches new ‘Posts’ feature

Strava is an online community bringing (recreational) athletes together around their (mostly) running and cycling activities. As I’ve written before, the strength of Strava is their focus on social and gamification features alongside of self-monitoring features (http://www.victoris.be/what-can-we-learn-from-strava/). It has made Strava the most attractive online community for recreational runners and cyclist. Originally, their user base consisted of considerably avid athletes, mostly cyclists with a competitive orientation who are perfectly served by Strava’s segment based leaderboards. The social interaction around their activities perfectly fits within the needs of this niche group.  

To date, however, their user base is growing and broadening. I see it every day when I get notifications of whom of my friends has joined Strava. They cannot be classified into this original ‘Strava niche population’ of the early days. They are less competition-oriented and care much less about leaderboards in which they’re  #3452. They do like to share their activities with their network of friends on Strava, they like to receive kudos and comments and hand them out frequently as well. 

The core of Strava is posting of activities, logged with e.g. wearable or gps devices. As of today, Strava has launched a new feature, allowing their user to add ‘posts’ to their timeline that are not directly based on activities they performed. They can share pictures of their favorite recipes, ask questions to their network, share links to relevant websites… All with the idea of giving the user a feature that allows them to share these non-activity based post with their network, without having to write gigantic titles for their activities, or put it into a comment under one of these activities. A good ‘relief’ to an existing ‘pain’ it seems.

A lot of Strava athletes however, seem worried about this update, judged from their reactions on the Strava Facebook page (https://www.facebook.com/Strava/). Summarizing them in once sentence would go something like ‘don’t become another Facebook or Instagram’. Yet, keeping the types of posts that now go on to these social networks on Strava, is what inspired this new feature.


I think it’s a logical step for Strava to introduce this kind of feature (apart from the fact that the former Instagram vice-president is now amongst their ranks), especially in the light of their broadening user base. But I also understand that the ‘original’ Strava user is scared of the thought of having his/her timeline filled with redundant posts ('how long until pictures of peoples dogs come up on my timeline’). The question is whether Strava wants to stick exclusively with this original user base? Is there much growth left among this group? The launch of this new feature should be framed in the light of this question I suppose. 

 To be continued…

What can we learn from Strava?





As humans, we all strive to nurture our basic needs of competence, autonomy and relatedness through our behavior. We want to feel a sense of accomplishment through what we do, we want to be in control of it and equally important: we want to experience it together with others. Exercising is a good example of behavior through which we want to nurture these needs. To keep ourselves motivated, we set goals, e.g. enrolling in a 5K running event 6 weeks from now, we team up with others for that extra bit of peer pressure that hopefully will prevent us from backing out… But sometimes, others back out before us… and then we follow… goals are not completed and our running shoes will soon collect a fresh new layer of dust.
Technology to the rescue? Maybe! Everybody knows somebody who owns a fitness tracker or an even more advanced GPS-enabled sports watch. They are developed to collect large amounts of data on our daily physical activity and/or exercise behavior. They count our steps, tell us how many calories we’ve burned, how far and for how long we’ve been running and many more. These data can be consulted on an online data analysis platform. Most of these platforms do a great job on telling us all about how well (or bad?) we are doing and how we are progressing towards achieving our physical activity goals. Feedback on our behavior, that’s the key to keep us motivated, it has been proven so many times… at least in the short term.
It appears that we abandon our fitness trackers almost as quickly as we’ve adopted them. How can this be? They give us this precious information on how well we’re doing and help us to achieve our self-set goals, fulfilling our much desired need for competence and autonomy… but we still can’t keep it up? Wait, what was that third thing we needed? Right, relatednessWhat if we could share our achievements with others online? 
I´m sure you already heard about Strava?! It’s a great example of an online fitness community that brings it all together, with success. On Strava, you can do everything we mentioned before, like uploading your running and cycling activities, set goals, plan your training, etc. But in addition to that, Strava successfully created a community around all of these data. Besides offering us features to ‘self-regulate’ our exercise behavior, Strava implemented two other key elements, which in our opinion, are crucial in its success: social interaction and gamification.
Social interaction features allow the users to communicate around their activities and most importantly: support each other. You can endorse your friends by e.g. giving ‘kudos’ or comments to their activity. Social support has been designated as a crucial factor for keeping up healthy behavior. When we don’t receive it offline, we can now get it online! Hurray for social interaction features, designed to fuel our sense of competence and relatedness!
What about gamification? ‘Isn’t that like giving virtual badges and stuff? That’s interesting… for a little while…’ Yes and no. Gamification on Strava is largely social. You can still get trophies for besting your 10k time, but through so-called leaderboards and other comparison tools, you can also see how you’re doing compared to people you know (or don’t know).
‘Interesting’, you say, ‘but I’m sure I’ve seen this on many other platforms?!’. Right you are, but interestingly, Strava has by far exceeded their competition in creating a social and gamified platform.
Let me show you a graph about this, based on data we collected from 332 runners who use platforms such as RunKeeper, Endomondo, Strava and Garmin connect. We asked them to indicate (on a seven-point scale) how often they use self-regulatory features (like planning a workout, setting a goal and monitor their exercise behaviour), social features (‘liking’ and commenting activities, connecting to other runners) and gamification features (collecting badges and comparing their achievements on leaderboards). While it is clear from the graph that these platform are primarily used to self-monitor exercise behavior (that’s what they’re made for), it is also pretty clear that social interaction and gamification features are a lot more intensively used on Strava compared to other platforms.

strava-graph
‘Why is that?’, you may be thinking. First of all: open profiles by default, which makes it fairly easy to follow other athletes on Strava. You can close your profile, but what is there to hide? Don’t you want to show off with your new KOM (King-of-the-Mountain)? Various other platforms expect you to file a formal ‘friend request’ to see each other’s activities in your feed. Strava’s Twitter-like approach is just more appealing and social. Second, granted, on average, Strava users are more competitive and therefore more susceptible to social interaction, comparison and gamification. This will have a significant impact on the use of gamification and social interaction features, for sure. Nevertheless, to some extent, everybody seeks recognition for their achievements. Finishing your first 5K race and having no one to congratulate you, or share your sense of accomplishment with, is just very unfulfilling. Lastly, while other platforms are increasingly offering social interaction features as well, Strava has done a particularly great job on integrating their social gamification features such as leaderboards and segments. The dynamics created by these features are fueling the interaction in the community.
What can we learn from this?
While Strava is indeed for a large part populated by intrinsically motivated athletes, there are things we can learn from their successful community approach and translate these ‘good practices’ to those people who are less eager to jump on a bicycle. If we want to motivate people to move more, sit less, eat healthier,… we should not limit ourselves to designing interventions and platforms that are too strictly focused on quantifying our behavior into numbers and then use these numbers for goal setting purposes (people constantly evaluating themselves in an isolated environment). It works, no doubt, but also in the longer term? If we want to achieve long-term behavior change, shouldn’t we enable people to achieve this is a more connected environment? The example of Strava shows that integrating the right features can result in a social and engaging environment, which may lead to a  more sustainable behavior change.
To be continued…