Projects


Clinican Experiences, Interpretability and Trust

Digitalization is profoundly changing the healthcare ecosystem, with an enormous potential to provide improvements to healthcare professionals, patients and other healthcare stakeholders but is also affiliated with potential challenges and risks. Often clinicians encounter frustrating experiences, especially when digital solutions are developed and deployed with insufficient consideration and understanding of their needs and workflow requirements.

Our Project CEDI – Clinician Experience with Digital Tools aims to study the experiences of clinicians, including physicians and nurses, and their use of digital tools that are an integrated part of their daily work. The objective is to explore how the transformational processes of digitalization impact the work, the use of current tools and workflows, and the professional identities of clinicians, in order to better understand the human experience of healthcare professionals on digital transformation in the clinic. We are interested in notable successes as well as particular pain points and daily frustrations, across all medical specialties, clinical roles, and types of digital tools, including electronic health records, decision support systems, and workflow automation. We performed interviews with clinicians in Switzerland to learn about their experience of using digital tools during their work in the clinic, and subsequently, we will analyze our findings aiming to understand how these tools affect their daily work life, workflow, and professional identity.


Health Artificial Intelligence

We develop and evaluate artificial intelligence approaches for health, biomedical discovery, and clinical applications. We are working in particular on knowledge-informed learning algorithms based on deep neural networks, with a focus on transfer learning and re-use of large-scale open-source generative models (language, images) for clinical applications. We also look at interpretability and implementation considerations for artificial intelligence applications in clinical contexts. 

We work on AI methods to support the interpretation of clinical metabolomics data. For example, we created Chebifier, a tool that uses deep learning to automatically classify metabolites.

We are investigating novel generative AI methods for multiple applications in health and medicine. A particular interest is models that are able to generate images based on text prompts, as these have many potential applications in medical education and in synthetic data generation. However, these models also have potential societal side effects. Our project UnRealBody explores and aims to mitigate the potential negative effects of hyper-idealisation in such models of human body images.

We are collaborators in the Human Behaviour-Change Project, funded by the Wellcome Trust and led by UCL’s Center for Behaviour Change. This project is developing a «knowledge system» based on annotated intervention evidence reports, semantics in the form of an ontology, and an artificial intelligence-based system that is able to predict the outcomes of hypothetical behavior change interventions, applied to an initial case study of smoking cessation behaviors.

We are collaborators in the Sinergia project «Connecting properties of the micro- and macrovasculature from multimodal imaging through genetics and deep learning to better understand vascular pathomechanisms and predict disease risk» (VascX) led by the Computational Biology group at UNIL. This project will explore the genetic basis of vasculatures in multimodal imaging datasets, and our group will explore the semantics of features learned by deep-learning approaches.


Evidence Synthesis and Knowledge Management

We are interested in approaches to semi-automate evidence synthesis in the clinical domain and accelerate the translation of evidence into guidelines and practice. We work on ontologies and semantic approaches to support integration, aggregation, and summarization of evidence. We are also interested in public understanding of medical knowledge and how this can be improved through better digital tools. 

We are partners on the Wellcome-funded GALENOS project that is providing a global resource of living evidence for early phase research in anxiety, depression and psychosis. This ground-breaking project aims to catalyse innovation and new treatment directions for these challenging mental health conditions.

We are deeply invested in the field of clinical decision support, particularly in the context of treatment planning for oncology, with a specific focus on prostate cancer. By leveraging our expertise and collaborative efforts with the Kantonsspital St. Gallen (KSSG) we aim to contribute valuable insights and support to enhance the planning and implementation of effective treatments for prostate cancer patients. Our project ONCO-TP aims to evaluate the potential for automating ranking evidence and information extraction to support decision-making for treatment planning with a case study in mCRPC.