Tensor-based machine learning for personalised medicine

This project proposes to solve personalised medicine problems using advance machine learning tools. Nowadays, a large amount of genetic data are being generated from patients with different diseases, with a large number of treatment options available to each patient. This presents great opportunities for personalised medicine, but also poses great challenges. Machine learning is a core technology in artificial intelligence that can transform medicine and healthcare by extracting useful information from medical data for accurate predictions. This project aims to improve the success rate of finding personalised treatments for complex disorders like motor neurone disease and pulmonary hypertension by learning the relationships between the genomes of patients and their clinical outcomes. Such data, i.e., genetic and clinical information, will be organised as multi-dimensional data sets, known as "tensors". We will develop a new tensor-based machine learning approach that can deal with the challenges of a high number of features but low number of samples in genomic datasets. We will investigate whether relationships between gene mutations and disease traits could be used to predict whether a treatment is effective in a specific patient. We will examine large datasets from nationally run genomic studies with the objective of reliable predictions on the outcomes of patients. This project will not only advance the field of personalised medicine but also lead to breakthroughs in artificial intelligence for the healthcare field.