In several years of study, I’ve never been on exchange. Not due to the lack of interest, it just never was the right time. However, the possibility of visiting a new institution/faculty, working with new people, observing problems you’ve spent months or even years working on, from a fresh perspective and, last but not least, doing it in a different country promises an exciting and enriching experience. And although my years of study are slowly but surely coming to an end, I never entirely gave up on the idea of the exchange. Getting involved in the WideHealth project allowed me to finally experience it, even if only for a short time. I visited FBK, a research institute in Trento, Italy. We worked together previously in an H2020 project, WellCo. The primary purpose of the collaboration is to apply findings based on this project to the virtual coach application they developed, the Salute+.
On the final Sunday in August, I’ve packed everything I needed for my 3-week visit into my car and headed towards Italy to Trento. Trento is a small city located in Trentino and surrounded by mountains, so it felt almost like home. The view from the FBK is beyond amazing, and it was an absolute joy to come to work every day for this reason alone. However, as breathtaking as the scenery was, I did not come there for this reason.
I went to Italy for the most fundamental reason ever. For food.
However, nowadays, nutrition has become very complicated. One should not overeat sugar, salt and fat and choose vegetables and fruit instead. However, this same ‘one’ should also eat enough protein, not by consuming too much processed meat, as they are very high in salt and fat. And then there is fibre intake. It seems like an easy one: “Eat cereals”. But not too much. Cereals do not only consist of fibre but also of sugar. And be sure to drink enough. Water. Stay away from juices. You can drink tea. But don’t add sugar.
So how to get all of the above-mentioned information without asking ALL the questions? This is easy. Get the data, try to find patterns and use them to your benefit. But what is the data? And what is a ‘good diet’?
One of the most renowned ‘good diets’ is the one based on the Mediterranean diet – eat enough fruit and vegetables, eat fish, eat white meat, stay away from too many sweets, don’t forget about olive oil… And as Italy is considered the heart of the Mediterranean, it is pretty natural that our FBK partners decided to use a Food Frequency Questionnaire (FFQ) regarding this diet in their application Salute+. Users are supposed to answer questions about their habits daily.
In the WellCo project and as a part of my PhD, I researched how to use Food Frequency Questionnaires (FFQs) to monitor nutrition smartly. Sometimes questionnaires can be very extensive, the one included in the application being one of those, and can become very tiresome to the users if answered frequently. On the other hand, general information about users’ diet quality can be predicted relatively good by using machine learning on a smaller subset of questions. However, diet quality is a very complex thing. Way beyond a simple target.
The FFQ at the Salute+ has 22 questions (e.g. how often do you eat milk/fruit/pasta/rice/fish/meat/ per day/week) that give information about 11 targets (fruit intake, sugar intake, protein intake…). My task was to develop the algorithm that would choose a small subset of the most informative questions for any given combination of desired goals and ask first-time users answers about their nutrition habits. From this, the algorithm decides which of the goals are fulfilled and which are more problematic. For instance, the algorithm sees that the user eats enough fruit but not enough vegetables. Next time the algorithm should use this information and choose the next subset of questions that give more information about the vegetable intake and doesn’t really care about the fruit intake. This would be repeated a few times in a certain period (e.g. two weeks, one month…) and then restarted. After each such period, the diet quality should be evaluated and hopefully, after a few such periods user’s nutrition would improve. The data I’ve worked on in the model training stage was created synthetically by using a much more extensive questionnaire, used in the national survey EUmenu, and narrowed down to the Salute+ questionnaire. As usual, understanding the dataset and preprocessing the data took quite some time. And despite most of the work being performed on the computer and could be done even in Slovenia, it was essential for me to be at the FBK. This helped me really understand what their questionnaire is doing so far, how they implemented it in the application, and how to make it more user-friendly. The option to knock at someone’s door instead of meeting on Zoom or Skype or, let alone, sending an email was of great importance as it allowed me to keep the research and mind flow without having to stop for a few days and wait for an answer.
Of course, not everything could not be done in 3 weeks (or a month), but the foundation for further research was set. Once the algorithm is evaluated on actual data from the application, it would be very nice to test it in person. And although this could be done without revisiting Trento, I firmly believe there would be several significant advantages to being there and testing in person. 😀
For me, a very positive aspect of the exchange was the possibility to focus just on my research work for 3 weeks. No need to think about other less research-related elements of my work. And once there are no bureaucratic disturbances, one can easily sit and work for hours and hours and hours… However, despite long workdays, I was able to get some taste of Italy. I was lucky to meet some really wonderful people, who took me as ‘one of them’ as soon as I could repeat some Italian phrases very convincingly and showed my deep and sincere appreciation for the scenery and the food. Therefore, 3 weeks at the FBK were a short but, at the same time, very memorable experience. I’m very happy and grateful to finally get a chance to do my abroad exchange. It was all I hoped for and more.
The EU-funded WideHealth project aims to conduct research on pervasive eHealth and establish a sustainable network of research and dissemination across Europe.