Advancing our understanding of complex diseases using AI

STFC Hartree® Centre worked with IBM through the Hartree National Centre for Digital Innovation (HNCDI) to develop an AI-enabled multi-omics integration software to enhance understanding of multi-factorial conditions such as inflammatory bowel disease.

Challenge

Inflammatory bowel disease and bowel cancer are classified as multi-factorial diseases due to the numerous complex biochemical processes that can influence their development. To study these diseases and allow disease prediction and prevention, it is essential to collect and analyse data from multiple biological sources, including genes, proteins and metabolites, a process known as multi-omics. However, due to the complexity and resource requirements of multi-omics analysis, both academia and industry often lack the computing power and software necessary to extract valuable insights from their data. Our team aimed to address this challenge by developing an AI-enabled multi-omics integration software.

Approach

Our team created software that integrates open-source multi-omics data for explainable machine learning analysis. The results offer valuable insights into disease prediction and identify explainable omic features linked to disease onset. The software was tested using inflammatory bowel disease and bowel cancer datasets and, after optimisation, achieved a prediction accuracy of 92%. Additionally, the software enabled the identification of key bacteria in the gut and the chemicals they produce that may contribute to inflammatory bowel disease and bowel cancer. This could potentially assist in expanding our understanding of other multi-factorial diseases faster.

Benefits

By identifying key factors that contribute to multi-factorial diseases, we are improving our ability to predict and prevent disease onset. The software offers an automated, high-throughput AI-enabled capability that can accelerate discovery. This tool could drive personalised medicine, improving patient outcomes. Its deployment has the potential to aid disease research by reducing the time from data generation to actionable insights, supporting advancements in life science and healthcare, possibly leading to more effective, targeted treatments.

“This project has provided HNCDI with the capability for rapid AI-enabled multi-omic integration analysis, resulting in novel disease insights for complex disease”

Jennifer Kelly, IBM Research