APC Microbiome Ireland, a world-leading SFI research centre based at UCC (University College Cork), has joined forces with BowelScreen and the Mater Misericordiae University Hospital, as part of 8 partner organisations from 5 European countries involved in the new Microb-AI-ome project launched in May of this year. Led by Prof. Dr. Jan Baumbach, an Artificial Intelligence (AI) expert from the University of Hamburg, the project aims to reduce the high false-positive rates associated with traditional CRC (colorectal cancer) screening and develop a more targeted and cost effective screening programme.
According to the WHO (World Health Organisation), CRC is the 2nd most common cancer globally. Early identification of CRC can dramatically improve patient prognosis and outcomes. Traditional CRC screening is based on a faecal immunological test (FIT) to detect colorectal lesions associated with CRC, but typically these tests have a false-positive rate of over 37%, which can often result in unnecessary invasive and disruptive patient colonoscopy treatments. Previous work in Prof. O’Toole’s APC laboratory has shown that changes in the gut microbiome occur in people with either polyps (early-stage colon growths) or colon cancer, suggesting they could be used as sensitive diagnostics.
Funded by the Horizon Europe programme, the new Microb-AI-ome project will leverage AI to conduct a more detailed and thorough analysis of stool microbiome data and is targeting a significant reduction in the false-positive rate, and well as enhancing the sensitivity for early detection of colon cancer. This innovative approach could significantly reduce the volume of colonoscopy procedures required as a screening tool with significant cost savings and harm reduction.
A core element of the Microb-AI-ome approach involves creating a world-first ‘privacy preserving federated’ network of CRC database repositories spread across Europe; and utilizing AI and Machine Learning (ML) to harness and process the big-data. AI is enabling an increasingly personalized patient diagnostic and treatment capability. However, the ability to utilize AI to predict colonoscopy intervention on a targeted patient basis is hampered by privacy regulations and a widely dispersed and unconnected international database infrastructure. The privacy-preserving federated machine learning approach circumvents these issues and combines the power of AI in Hamburg with the established national bowel screen programmes in Ireland (Bowelscreen) and France, and the microbiome expertise of APC.
Prof. Paul O’ Toole of APC notes:
“The Microb-AI-ome project will respect privacy regulations by integrating isolated, national databases into one international and privacy-preserving federated database network, as opposed to being cloud-based and only anonymous AI model parameters are exchanged during the model learning process within this database network. Crucially this ensures that all personal data remains safely on the local databases without infringing privacy regulations; and at the same time will be able to feed and train the AI models. The project will integrate microbiome data collected during its runtime for over 4,000 individuals from Ireland and France. The federated database network will be ready for integration of around 100,000 datasets from synergistic programmes by 2026.”
Prof. Padraic MacMathuna, Clinical Director of Bowelscreen and alumnus of the Mater Misericordiae University Hospital added:
“This ground-breaking project combining international expertise in CRC, AI and microbiomics is a significant development in medical research and leverages the established scientific and clinical expertise available in Ireland. Outcomes from this Horizon Europe project have the potential to significantly enhance CRC early detection with consequent increased survival.”