Talk by Ewan Birney

Title

AI modelling of Healthcare trajectories & Danish and UK history with genetics

 

Abstract

Part 1: AI modelling of Healthcare trajectories

In the first part I will briefly outline work on modelling healthcare trajectories using the UK BioBank data, and comprehensively tested using Danish data. The modelling technique is a repositioning of generative AI techniques, namely attention based generative pretrained transformers, but adapted for the heterogeneous time data present in healthcare. Our method does as well as if not better than many existing epidemiological methods for specific diseases, but is trained – and predicts – all diseases (as defined by ICD10 codes). We are also able to sample future health trajectories of patients with well calibrated probabilities. We tested the model in a variety of ways, with our most comprehensive test being a zero-shot (ie, no parameter change) of the UK model in a 1.9 million cohort from Denmark. 

I will describe the method briefly, outline why we are confident we have a generalised solution and briefly discuss extensions for other features such as prescriptions, genetics or environmental measures. 

 

Part 2: Danish and UK history with genetics

In the second part of my talk I will discuss an interdisciplinary project between geneticists and historians. Our DNA carries an accurate record of our recent family structure, such that nearly every individual related by a 7th generation grandparent or younger, and many of the people related by a 20th generation grandparent share a region of the genome which will be “identical by descent” (IBD). When individuals are genotyped (a cheap form of determining their genome) we can recognise these regions which are identical by descent. The length of the region (accounting variation in recombination frequency, so called genetic distance or centi-Morgans) is expected to change depending on the number of generations separating the two individuals of interest, meaning that the length of the segment is related to age of the common ancestor of two individuals. 

Denmark and United Kingdom are two countries with extensive genotype cohorts available for responsible research. Having gathered appropriate approvals to access this data, we have generated the largest ever inherited by descent comparison, involving over 800,000 individuals and over 13 billion comparisons. From this we can link individuals in these cohorts which are up to 10th cousins apart, with the estimated time of the oldest common ancestor being between 500AD to 1100AD. The presence of the extensive pedigree in Denmark has allowed us to verify the consistency of the IBD map with known pedigree information. This comprehensive IBD map therefore informs our understanding of population movements from this point in history to the present day. Our analysis both supports and sometimes challenges historical evidence about population movements.

In the talk I will outline this method and discuss how the IBD sharing within Denmark, within the UK and between the two countries integrate with our current views of Danish, UK and North Sea history.

 

Speaker

Ewan Birney is the Executive Director of the European Molecular Biology Laboratory (EMBL). Together with Peer Bork, who is the Interim Director General of EMBL, Ewan leads EMBL’s activities, continuing to deliver on the vision and priorities set out in the ‘Molecules to Ecosystems’ programme. Ewan also runs a small research group working on intraspecies variation in medaka fish and humans. Ewan completed his PhD at the Wellcome Sanger Institute with Richard Durbin. In 2000, he became Head of Nucleotide data at EMBL-EBI and in 2012 he took on the role of Associate Director at the institute. He was Director of EMBL-EBI between 2015 and 2025. In 2020, Ewan also became the Deputy Director General of EMBL, and in 2025, he was named Executive Director of EMBL.

Ewan led the analysis of the Human Genome gene set, mouse and chicken genomes and the ENCODE project, focusing on non-coding elements of the human genome. Ewan’s main areas of research include functional genomics, DNA algorithms, statistical methods to analyse genomic information (in particular information associated with individual differences in humans and Medaka fish) and use of images for chromatin structure. Ewan is a non-executive Director of Genomics England, and a consultant and advisor to a number of companies, including Oxford Nanopore Technologies and Cantata Bio. Ewan was elected an EMBO member in 2012, a Fellow of the Royal Society in 2014, and a Fellow of the Academy of Medical Sciences in 2015. Ewan was appointed Commander of the Order of the British Empire (CBE) in the 2019 New Year Honours. He was also elected as a member of The Royal Danish Academy of Sciences and Letters in 2024. In 2019, Ewan became a Board Member of the Biotechnology and Biological Sciences Research Council (BBSRC). Ewan serves on a number of Scientific Advisory Boards (SABs), including the DKFZ, ICR, OICR, Institut Pasteur and Riken institute.

He has received a number of awards including the 2003 Francis Crick Award from the Royal Society, the 2005 Overton Prize from the International Society for Computational Biology and the 2005 Benjamin Franklin Award for contributions in Open Source Bioinformatics. Ewan holds honorary doctorates from the University of Tartu, Estonia and Birkbeck College at the University of London, UK.