News

We have recently finished the curation process of the mobile big dataset obtained in our first study (around 18 million records) so that it can be used both by the POSTCOVID-AI platform and by any other interested researcher or entity. As we have already indicated in previous posts, this is a one-month study where, by using a mobile application developed for the project, heterogeneous data have been collected from 110 strategically selected individuals. This dataset is available at osf.io/6wrv3 where you can find 10 files with cleaned and preprocessed data as well as a detailed description of them. Specifically, in this first curated dataset generated by the POSTCOVID-AI project one can find:

- Data continuously collected through the participants' smartphone sensors: physical activity (resting, walking, driving, cycling, etc.), name of detected WiFi networks, connectivity type (WiFi, mobile, none), ambient light, ambient noise, and status of the device screen (on, off, locked, unlocked).

- Data corresponding to an initial questionnaire, with information related to demographic data, symptoms and COVID vaccination, average hours of physical activity, and answers to a series of questions to measure mental health, many of them taken from internationally recognised scales (PANAS, PHQ, GAD, BRS and AAQ).

- Data corresponding to daily questionnaires, where variables related to mood (valence, activation, energy and emotional events) are measured.

- Data corresponding to weekly questionnaires, where information on work situation, symptoms and COVID vaccination, hours of physical activity per week, questions related to physical and mental health, etc. are requested.

We are currently working on obtaining new higher-level or statistical indicators from the curated dataset that can be directly interpreted by POSTCOVID-AI stakeholders.

Following the collection of data from the first study of the POSTCOVID-AI project, the team has embarked on a meticulous and necessary task: data curation. The data collected is heterogeneous in nature, i.e. diverse. They range from sensor data, such as continuous numerical records of brightness or environmental noise measurements, to categorical values of physical activity or textual responses to some of the surveys. All these records are susceptible to reflecting values outside the permitted ranges, duplicate records or even the absence or loss of some of them. This is why it is necessary to curate or harmonise them in order to generate a clean and functional database.

Although there are several curation techniques, most of them coming from the so-called Data Science field, the main methods used in our project are labelling, coding, deletion of non-relevant fields, elimination of outliers, and imputation of missing values. Labelling and coding allow the contextualisation of some of the recorded values so that the user of the data can clearly understand what they refer to (for example, indicating the physical activity performed by each user at each instant of time recorded). Field deletion allows the deletion of records that are technically necessary but do not add value to the dataset (e.g. the time at which the data was sent from the mobile application to the server). The elimination of outliers refers to the use of statistical techniques that detect numerical values that escape the normal distribution of the values (e.g. an absurdly high environmental noise value). Finally, imputation also makes use of statistical techniques such as interpolation to fill in some of the gaps generated by the missing data record.

In the last post, we described the main features of the first study of the POSTCOVID-AI project. After an intensive month of data collection, we can now say that the study has been successfully completed! 

The data are currently being "curated" by the project research team. In the next post, we will tell you in more detail what this process consists of and what the resulting product of this process will be. As a preview, we can say that the curation process will allow cleaning and preparing the big set of recorded data so that they can be used both by the POSTCOVID-AI platform and by any other interested researcher or entity.

After several months of work and preparations, the first study within the POSTCOVID-AI project was launched on 15 November, which will allow us to obtain our first results on the impact of the pandemic on the well-being of the Spanish population.

The prerequisites to be able to participate in this study were to be of legal age and to have a mobile phone with an Android operating system. A very important aspect to consider in the selection of participants was that they should faithfully represent the general population. To this end, rigorous stratification criteria were followed based on certain socio-demographic characteristics. The final result was a group of 110 people (50% women) aged between 18 and 70. Another important characteristic taken into account was representativeness in relation to geographical distribution, for which the Nielsen criteria for the division of areas were followed. Finally, a representative distribution of the participants according to their gross annual income, has been sought, considering five ranges from less than 12,450€ to more than 60,000€.

During the month of the study, participants will have to keep the project application installed on their mobile phones and answer the questionnaires received throughout the day

The use of mobile applications to carry out observational studies based on scientific evidence is becoming increasingly common. Such apps allow the collection of data from users through a variety of sensors available on mobile devices, which can measure behavioural aspects such as activity or sleep patterns. In addition, these applications allow for quick surveys to be conducted anytime, anywhere. This, in addition to the high availability of these devices among different segments of the population, makes mobile phones an unparalleled data collection tool. Thanks to this novel way of collecting data, participation becomes more convenient and realistic, as it is carried out with minimum effort by the participants and in their most natural environment.

At POSTCOVID-AI we have developed a mobile app specially designed to collect real-time data on the social, behavioural and emotional behaviour of the population. The app will fulfil a dual function on a daily basis during the time the application is installed on the user's mobile phone. On the one hand, it will passively record data from sensors on the participant's mobile phone that will provide information about the weather, physical activity or environmental noise, among others. On the other hand, the app will facilitate the self-recording of participants' emotional experiences. To do this, the app is programmed to send notifications throughout the day, with a survey consisting of two questions: "How do you feel now" and "Since your last register, have you experienced any remarkable emotional situation?". All this data* will be combined and processed in the future with sophisticated Artificial Intelligence techniques that will allow us to infer valuable information about the impact of the pandemic on the well-being of the Spanish population.

*The data processed through the POSTCOVID-AI app are stored on a secure server of the University of Granada and comply with the Spanish Organic Law 3/2018, of 5 December, on the Protection of Personal Data and Guarantee of Digital Rights and the General Data Protection Regulation (EU) 2016/679 of the European Parliament and of the European Council, of 27 April 2016.