by the Models and Algorithms for Data & Text Mining Laboratory (MADLab)

The key topics are:

1. 1. Integrating Clinical Expertise with Causal Discovery Techniques. We show how to combine clinical experts' knowledge with causal discovery methods. Participants will learn how to leverage domain expertise to design a causal network that represents the cause-effect relationships.

   2. Assessing the Impact of Specific Cancer Treatments. We focus on leveraging the causal network designed at the previous step to estimate the effects of specific oncological treatments on the cardiovascular outcome. Participants will work with simulated data and link the provided dataset with the causal network structure.

   3. Comparison with Biased Estimates. Finally, we compare unadjusted raw estimates with those derived from the causal network to illustrate the importance of causal inference. Participants will learn techniques to identify and adjust for biases in healthcare data, making the whole tutorial applicable to any real-world setting.

• [08:30 - 09:30] Part I, Alessio Zanga

  • Elements of causal models: introduction to the theoretical aspects of causal models, inference and discovery.

• [09:30 - 10:30] Part II, Alice Bernasconi

  • Cancer in adolescents and young adults: an overview of the case study on oncological treatments and their long-term sequelae.

• [10:30 - 11:30] Coffee break

• [11:00 - 12:00] Part III, Alessio Zanga

  • Implementing a causal network: building a causal network from experts’ knowledge and data.

• [12:00 - 12:30] Part IV, Conclusions & General discussion

  • Wrapping-up the tutorial, fostering the interaction between the experts and participants on specific topics.

This tutorial is designed for researchers in medical informatics and computer science, healthcare professionals, and epidemiologists who are interested in developing, evaluating, or applying causal network models. Participants should have a basic understanding of machine learning and be familiar with fundamental concepts such as probability and statistical modeling.

• Prof. Riccardo Bellazzi, University of Pavia

• Prof. Concha Bielza, Technical University of Madrid

• Prof. Pedro Larrañaga, Technical University of Madrid

• Prof. Peter J.F. Lucas, University of Twente

• Prof. Bojan Mihaljević, Technical University of Madrid

• Prof. Silvana Quaglini, University of Pavia

• Prof. Lucia Sacchi, University of Pavia

• Prof. Marco Scutari, Istituto Dalle Molle di Studi sull’Intelligenza Artificiale

• Prof. Fabio Stella, University of Milano-Bicocca

• Prof. Allan Tucker, Brunel University London

• Prof. Blaz Zupan, University of Ljubljana

• Zanga, A., Ozkirimli, E., & Stella, F. (2022). A survey on causal discovery: Theory and practice. International Journal of Approximate Reasoning, 151, 101-129.

• Bernasconi, A., Zanga, A., Lucas, P. J., Scutari, M., Di Cosimo, S., De Santis, M. C., & Ada Working Group. (2024). From Real-World Data to Causally Interpretable Models: A Bayesian Network to Predict Cardiovascular Diseases in Adolescents and Young Adults with Breast Cancer. Cancers, 16(21), 3643.

• Bernasconi, A., Zanga, A., Lucas, P. J., Scutari, M., Trama, A., & Stella, F. (2024). A causal network model to estimate the cardiotoxic effect of oncological treatments in young breast cancer survivors. Progress in Artificial Intelligence, 1-13.