Skip to main content


IDSAI Seminar: Novel Machine Learning Methods for Cancer Research

Open to University of Exeter staff and students

IDSAI Seminar given by Dr Colin Campbell, Reader in Mathematics for Information Technology at the Intelligent Systems Laboratory, University of Bristol.

Event details


The talk will have two parts, both illustrating the potential promise of using innovative machine 
learning methods in application to the large and diverse omics datasets now being derived within cancer research. In the first part we consider novel methods, based on machine learning, for predicting the pathogenic impact of variants in the human genome (disease-driver or neutral) in the cancer genome (e.g., in non-cancer contexts (, for 
predicting the functional impact of indels ( and haploinsufficiency.

Applied to the genomes of common solid tumours, and excepting phenomena such as hypermutation, these methods indicate that core disease-driver variant sets are typically small in size and the drivers partially identifiable, though driver set sizes are very variable by cancer type. Towards the end of the first part we will also briefly review several further projects, not cancer-related, but oriented towards using machine learning applied to general biomedical research data. 

An algorithmic framework previously proposed and called Latent Process Decomposition (LPD), 
has been used by Prof. Colin Cooper's team at UEA to efficiently separate aggressive from benign prostate cancer, a very promising result given the 12.0% UK male liftetime risk of diagnosis with the disease, and the clinical importanceof correctly targeting treatment. In the second part we give a brief overview of LPD and a general outline of recent work in this area.

Dr Colin Campbell gained a First Class Honours degree in Physics from Imperial College, London and a Doctorate from the Department of Mathematics, King's College, London, under the supervision of Professor Peter West FRS.  His research interests are in artificial intelligence, specifically machine learning. Within this discipline, interests include probabilistic graphical models and kernel-based methods, algorithm design and the applications of machine learning techniques in bioinformatics, particularly medical bioinformatics.



Laver Building LT6