Imageomics Speaker Series: Janna Hastings, University of Zurich

All dates for this event occur in the past.

smiling woman with blue eyes and light skin. She has her brown hair pulled back into a low braid and is wearing a dark blue shirt.

Ontology pre-training as a strategy to improve data-driven predictions:

Recent breakthroughs in artificial intelligence have resulted from the combination of big data with large-scale neural networks. However, neural networks face several challenges: they are opaque and cannot offer explanations, they are vulnerable to adversarial challenges, they can be brittle and lack generalisability, and their operation cannot be controlled in semantic terms accessible to domain experts. With an application to prediction of chemical properties from chemical structures, I will describe an approach to enhance the training of neural network-based predictive systems with ontology pre-training. With this strategy, the meaningful classifications from a domain ontology are taught to the network first in a dedicated training phase before the network is fine-tuned for the specific property prediction task. We have been able to show that this strategy improves prediction performance, reduces training time and improves the interpretability of the associated model. 

Speaker Bio: 

Janna Hastings was born in Cape Town, South Africa where she completed her undergraduate studies in Mathematics and Computer Science. Thereafter, she moved to Cambridge, UK to join the Cheminformatics and Metabolism group at the European Bioinformatics Institute (2006-2015) where she led the development of the ChEBI molecular ontology and metabolism knowledgebase. She completed part-time Master's degrees in Computer Science (University of South Africa, 2011) and Philosophy (Open University, 2012). She obtained her PhD in Computational Biology from the University of Cambridge (2015-2019), where she studied the role of metabolism in healthy ageing using multi-omics data and a time-series modelling approach.

She completed postdoctoral studies at the Otto-von-Guericke University Magdeburg (2019-2022) working at the intersection between knowledge-based systems and artificial intelligence, at the EPFL (2020-2022) in neurodegenerative disease bioinformatics, and with the Human Behaviour-Change Project at University College London (2017-2022) where she helped to create an artificial intelligence-driven knowledge system for evidence about human behavior change to support decision-making, policy and practice.

Since August 2022 she has been an Assistant Professor of Medical Knowledge and Decision Support at the Institute for Implementation Science in Health Care, Faculty of Medicine, University of Zurich, and Vice-Director of the School of Medicine at the University of St. Gallen. The focus of her research is on digitalization in the clinic and how new digital technologies can be harnessed to accelerate biomedical discovery and improve the working lives of clinicians.