AI comes out on top in the category supported by the TIC Salut Social Foundation in the Premis Recerca Jove 2022

Every year, the Agency for Management of University and Research Grants (AGAUR) of the Department of Research and Universities organises the Premis Recerca Jove. The aim of the awards is to promote an interest in science among young people enrolled in their last year of school, i.e. their second year of baccalaureate, before they attend university. The TIC Salut Social Foundation collaborates in the award category related to innovative technology applied to health and social welfare.

In the 2022 edition, Núria González, a student at the Francesc Macià Secondary School in Cornellà de Llobregat, won the ICT Health and Social category for her research project entitled AI applied to the early diagnosis of melanoma. Núria is currently in her first year of the Biomedical Engineering degree at the University of Barcelona and we got the chance to talk to her about her findings, which helped earn her the PRJ 2022 award.

This conversation revealed not only the talent and concerns of the newer generations, but also the fact that the biotech sector is under constant development and is being continually redefined, with the professions of the immediate future being shaped by the demands and needs arising from the new possibilities that technology has to offer. This is what the winner had to say:

  • Why did you choose this topic for your research project?

I wanted to choose a line of work that was related to the university degree that I wanted to do after finishing school, which was Biomedical Engineering and which I am now enrolled in. This degree combines medicine with new technologies that can be applied to medicine. On the other hand, I chose to study cancer because it is one of the most aggressive diseases. I specifically centred on melanoma, which is a form of skin cancer, often caused by solar radiation and with the potential to affect almost anyone.

  • What was your starting hypothesis for this research?

I started from the hypothesis that it was possible to create a computer programme that could use artificial intelligence to diagnose skin cancer with greater efficiency than a dermatologist. I formulated this hypothesis after observing that skin cancer is a very common disease. If it is diagnosed at an early stage, it can very likely be cured. If it is not diagnosed early on, it may become incurable. That is why we need methods that can assess skin lesions as accurately and as quickly as possible.

  • What were your findings?

The experiment was a success and I was able to demonstrate that a system based on Artificial Intelligence has the potential to diagnose with high precision and speed whether a skin lesion is a melanoma or not.

Of all the different methodologies I analysed, the ones that provided the most satisfactory results relating to the essential characteristics for diagnosis were the neural network and the random forest, with a diagnostic accuracy of 94%. This is considerably higher than the average obtained by medical specialists using traditional methods, which portray an 86.6% accuracy rate.

An artificial intelligence software could facilitate early diagnosis, preventing the evolution of the disease. On the other hand, AI could also predict the future course of the disease using predictive algorithms.

This technology could also be applied to any image-based diagnosis, provided that the fundamental characteristics are correctly selected. In any case, in order to corroborate these results, further research is still needed.

  • What career would you like to pursue once you finish your studies?

I’m not sure what I want to do after I finish my degree. There are several different areas of biomedical engineering that I am interested in, such as computer science or the design of diagnostic, monitoring and medical therapy equipment. Right now it’s hard to envision my future because I’m still discovering what I’m really interested in. Plus, as the biotech field is constantly advancing, the career opportunities are also becoming more and more diversified.

  • Out of all the methodologies reviewed, the neural network and random forest were the highest scoring, with a diagnostic accuracy of 94%