Back to the previous page

Jointly organized by TheSense (HES-SO Valais/WallisCHUVUnil) and IDIAP, the 8th Valais/Wallis Workshop brought together 70 actors in the field of artificial intelligence at the Swiss Digital Center in Sierre. The objective of the workshops is to bring together engineers and researchers active in the field of pattern analysis and artificial intelligence with applications related to security, health or energy.

An 8th edition dedicated to health

Co-organizer of this workshop, Henning Müller, professor and researcher at the Institute of Informatics, wanted to propose exchanges on the theme of health. Indeed, the use of AI in clinical routine represents both opportunities and challenges. Two invited lectures underlined the central theme of this morning: machine learning (ML) and deep-learning models are now achieving performances close to those of humans in the classification and analysis of imagery. However, their adoption in clinical settings remains a challenge due to the lack of transparency in the interpretation of data and algorithmic decisions.

The HES-SO Valais-Wallis proposes concrete solutions

Among the research work presented, the HES-SO Valais/Wallis presented applied solutions. Niccolo Marini spoke about reducing the need for manual annotations when extracting multimodal data, Valentin Oreiller detailed the development of support systems for clinical decision making contributing to optimal treatment decisions for breast and prostate cancer survivors. Anjani Dhrangadhariya's work presented a study on weak supervision for fuzzy clinical extent detection and Gaetano Manzo discussed breast cancer survival analysis for clinical decision support.

Opportunities, challenges and confidence in AI

Partner discussions focused on confidence in the results produced by artificial intelligence and their interpretability. Many questions remain unanswered, including infrastructure costs, data storage, the evolution of European and American regulations, the definition of acceptance and responsibility criteria (who of the technology or the user is responsible for what?), but also the support towards the adoption of tools and new processes.