Machine Learning for Personalized Medicine

Marie-Curie Action: "Initial Training Networks"

Discovering Subgroups of Disease with Model-based Machine Learning

by Iain Buchan

“Personalised Medicine”, “Precision Medicine” and “Stratified Medicine” are overlapping, widely-cited challenges to better target healthcare interventions to the needs of individual patients. Before this can happen, however, crude diagnostic labels such as asthma need to be broken down into sub-diseases with distinct biological mechanisms that convey different risks of developing conditions or responding to treatments. In his talk, Professor Buchan will present studies where machine learning methods have been used to generate new hypotheses about sub-diseases by exposing structure in complex longitudinal datasets. This will be compared with regression-based latent class analysis that can over-average multi-factor associations, masking subgroups. Professor Buchan will expand on the general methodology of generating and following hypotheses in this way to ‘disrupt’ discovery science with electronic health record data. He will also explore machine learning to improve software-driven interventions in human health, and put the case for a different approach to experimental medicine in learning health systems.

***************

Iain Buchan is Professor in Public Health Informatics and leads the Centre for Health Informatics at the University of Manchester. Nationally, he is Director of the MRC Health eResearch Centre of the UK’s Farr Institute for Health Informatics Research. He has over twenty years’ experience in Health Informatics, with backgrounds in Clinical Medicine, Pharmacology, Biostatistics and Public Health. His research is focused on harnessing large-scale linked health data to build usefully complex models of health and care. He leads a multi-disciplinary research team that is deep in statistical and software engineering methodology. He writes software (e.g. www.statsdirect.com) as well as working at higher levels of abstraction and leadership in advancing reproducible, larger scale health science. He is a Fellow of the American College of Medical Informatics, the top honour in the field.

>back to "Talks and Speakers"