Imagine your body is like a complex machine where some parts are naturally weaker due to your DNA, and others are worn down by habits like drinking. This work figures out exactly which genetic 'weak spots' make some people get liver cancer from alcohol while others don't. By using AI to spot these patterns early, doctors can find the disease when it is still treatable.
Alcohol-related liver cancer is often detected too late, leading to a low 15% survival rate. There is a critical lack of tools to identify which alcohol-consuming individuals are genetically predisposed to develop tumors.
A genome-wide association study dataset of over 11,000 individuals and a Polygenic Risk Score (PRS) to predict cancer risk. AI models are being developed to integrate imaging and genetic data into a minimal viable product.
If you are a drug discovery firm dealing with a lack of specific targets for liver cancer — this project identified a new signal at EPHA2 and other risk loci. This provides a concrete biological target to develop new chemopreventive drugs.
If you are a diagnostic test developer dealing with late-stage cancer diagnoses where survival is only 15% — this project developed a polygenic risk score (PRS). This tool can identify high-risk patients with a two-fold increased risk, enabling earlier intervention.
If you are an AI health-tech startup dealing with fragmented patient data — this project built AI models that integrate genetic information and digital imaging. This creates a minimal viable product for cost-effective risk stratification in at-risk individuals.
Based on available project data, the specific pricing is not mentioned, but the project aims to develop 'cost-effective strategies' for prevention and detection.
The project uses a massive dataset of 4,418 cases and 6,655 controls across 20 cohorts, suggesting the underlying data models are designed for large-scale population application.
Based on available project data, specific licensing terms are not provided, but the project has identified a novel genome-wide significant signal at EPHA2 which may be subject to intellectual property claims.
The AI models are designed to integrate genetic and non-genetic information, including digital imaging, to be used within HCC surveillance programs.
The project runs from 2023-01-01 to 2027-12-31, with AI models reaching the minimal viable product stage by the end of the project.
The consortium is research-heavy with 15 partners across 7 countries, featuring a strong academic core (5 universities, 4 research institutes). However, there is a modest industrial presence (2 industry partners, including 1 SME), representing a 13% industry ratio. This suggests the project is currently in the high-value discovery phase, with a lean bridge to commercialization via the AI and diagnostic partners.
Contact UNIVERSITE LIBRE DE BRUXELLES regarding the AI MVP and EPHA2 findings.
Contact us to connect with the GENIAL consortium for licensing the Polygenic Risk Score.
