"Deep learning could determine whether or not a person has COVID-19 with a recording of their cough"

 

Detecting the coronavirus infection in a person by recording their coughing and improving lung cancer diagnoses were just two of the major breakthroughs unveiled at the third edition of the Deep Learning Barcelona symposium in December. This international scientific symposium co-organized by the Universitat Oberta de Catalunya (UOC) brought together researchers from a variety of disciplines related to deep learning, one of the most innovative fields within artificial intelligence (AI). As well as a special session dedicated to life sciences, the convention presented breakthroughs related to new drugs, medical diagnoses, genomics, nutrition, COVID-19 detection, and medical imaging. Carles Ventura, a UOC faculty member and researcher in the Artificial Intelligence for Human Well-being group (AIWell, previously known as SUNAI), affiliated to the University's Faculty of Computer Science, Multimedia and Telecommunications and the eHealth Center, explained some of the highlights of this international congress.

 

Detecting the coronavirus infection in a person by recording their coughing and improving lung cancer diagnoses were just two of the major breakthroughs unveiled at the third edition of the Deep Learning Barcelona symposium in December. This international scientific symposium co-organized by the Universitat Oberta de Catalunya (UOC) brought together researchers from a variety of disciplines related to deep learning, one of the most innovative fields within artificial intelligence (AI). As well as a special session dedicated to life sciences, the convention presented breakthroughs related to new drugs, medical diagnoses, genomics, nutrition, COVID-19 detection, and medical imaging. Carles Ventura, a UOC faculty member and researcher in the Artificial Intelligence for Human Well-being group (AIWell, previously known as SUNAI), affiliated to the University's Faculty of Computer Science, Multimedia and Telecommunications and the eHealth Center, explained some of the highlights of this international congress.

It is an evolution of technology learning models based on neural networks. Its most revolutionary aspects are that it allows us to build more complex models and train technology to learn in a much more efficient manner. While classic machine learning algorithms were based on manually defined characteristics, current models learn directly from data. The models themselves are capable of learning what the most significant characteristics are for tackling a particular problem.

The possibilities for application in our everyday lives are numerous, ranging from cameras with object recognition capabilities installed in cars to reduce traffic accidents to reducing ambient noise in video calls, which is something that has now become a necessity in our working lives.

The researcher Miguel A. González Ballester from Pompeu Fabra University presented a model that makes it possible to re-identify lung cancer nodules in images taken at different times (known as longitudinal data) with no need to register the images. Image registration is a tedious process that is usually necessary because images taken from a patient at different points in time are usually not aligned, because they were not taken in exactly the same place.

Another project in the field of COVID-19 presented by Adrià Mallol-Ragolta from the University of Augsberg consists of a model that determines whether a person is infected with COVID-19 based on a recording of their cough. A model trained with audio data consisting of cough samples from different people and tags indicating whether they are infected with COVID-19 or not provides the user with an infection prediction by simply recording their cough.

AI has an increasingly high profile in the media and in education. But there is a still a long way to go in this regard. I think it is important for AI to be introduced into school curriculums, given that it is already a technology that impacts the whole of society. It is also very important to train the scientists who develop these AI applications in ethical issues.

Certain highly specific tasks for which a specific model has been trained currently outperform the human brain, but there is a long road ahead before we can achieve what is known as general artificial intelligence, which is a system capable of reasoning like a person. For example, chatbots work well in specific domains, such as a company's customer service, but designing a chatbot for an informal conversation much more complicated.

We are in a privileged location in Southern Europe, where more and more companies are setting up shop and the quality of university education is very high. It is a shame that many of our best students go on to conduct their research in other parts of the world, where they are better paid for their efforts. But the congress is an opportunity to lure them back to the city where they trained and to have a local impact on the researchers who have decided to stay and the students we are educating. Let's hope that opportunities for students will increase and improve so they can remain in the city if they wish. And let's hope that those who decided to leave at the start of their research careers have the chance to return. We now have some research groups at major companies, like Amazon, and other companies such as Apple, Meta (formerly Facebook) and Microsoft, are also making headway in this respect.

One of the projects the group presented at this symposium is the prediction of stress situations in drivers based on an automatic analysis of the scene. Our research lines also include affective computing, in which we are embarking on a project to help elderly people who live alone. This will also involve an emotional perspective with a module that is capable of understanding how the person feels and helping them improve their wellbeing.

 

UOC R&I

The UOC's research and innovation (R&I) is helping overcome pressing challenges faced by global societies in the 21st century, by studying interactions between technology and human & social sciences with a specific focus on the network society, e-learning and e-health.

Over 500 researchers and 52 research groups work among the University's seven faculties and two research centres: the Internet Interdisciplinary Institute (IN3) and the eHealth Center (eHC).

The University also cultivates online learning innovations at its eLearning Innovation Center (eLinC), as well as UOC community entrepreneurship and knowledge transfer via the Hubbik platform.

The United Nations' 2030 Agenda for Sustainable Development and open knowledge serve as strategic pillars for the UOC's teaching, research and innovation. More information: research.uoc.edu #UOC25years