MICRO-PREDICT AI

Project concept – MICRO-PREDICT AI

MICRO-PREDICT AI is a research and innovation project concept aligned with the EU Cancer Mission, aiming to advance cancer prevention and early detection through the development of AI-driven microbiome-based prediction tools capable of identifying cancer risk before the onset of disease, ideally up to two years in advance.

Cancer often develops silently over long periods and is still frequently diagnosed at advanced stages, when therapeutic options are limited and outcomes are poor. Early, minimally invasive and reliable biomarkers are therefore critical to improve survival and reduce the overall cancer burden. Growing evidence shows that alterations of the human microbiome and the transition towards dysbiosis are strongly correlated with oncogenesis and tumour progression. Despite the availability of large microbiome biobanks and registries at national and international level, this predictive potential remains largely underexploited in real prevention strategies.

The core idea of MICRO-PREDICT AI is to translate longitudinal microbiome knowledge into actionable, personalised cancer risk prediction, integrating microbiome profiles with clinical, phenotypic, behavioural and patient-reported data. The project will move beyond static analyses and focus on longitudinal risk trajectories, capturing how microbiome dynamics evolve over time and interact with host factors to anticipate cancer development.

The project will develop and validate AI-based risk modelling approaches, combining:

• large-scale longitudinal microbiome datasets from existing cohorts and registries;

• clinical and demographic data;

• lifestyle and behavioural factors (e.g. diet, physical activity);

• early, non-specific signals reported by individuals (fatigue, pain, weight loss, body changes).

Advanced AI techniques will be used to identify predictive patterns, stratify populations at risk and estimate individual cancer risk profiles. These models will be complemented by organ- and body-level simulations inspired by the Virtual Human Twin paradigm, enabling the exploration of host–microbiome interactions and the progression from dysbiosis to oncogenesis across different cancer types, where feasible.

A key element of the project is the comparison of microbiome-based prediction tools with other minimally invasive approaches, such as liquid biopsy tests, in terms of predictive power, simplicity, cost-benefit ratio and potential for large-scale implementation. Validation will be performed using independent cohorts to ensure robustness, reproducibility and clinical relevance.

Beyond prediction, MICRO-PREDICT AI will support preventive action. The project will contribute to the development of evidence-based guidelines to manage cancer risk through lifestyle, nutritional and behavioural interventions, empowering citizens—especially people at higher risk—to act early. Citizen and patient engagement activities may support awareness, data and sample collection, and education on cancer prevention.

The project is designed to deliver impact across multiple levels:

• Citizens and people at risk will benefit from earlier awareness and actionable insights to reduce cancer risk before disease onset;

• Clinicians will be enabled to initiate treatments earlier, with higher chances of success;

• Policymakers and public health authorities will gain robust evidence to design advanced, data-driven cancer prevention programmes.

From a technological perspective, the project will adopt FAIR-by-design data architectures, ensure interoperability with European research and health data infrastructures, and align with resources provided by the Knowledge Centre on Cancer to foster EU-level coordination and uptake.

Role of Acube

Acube will participate as a core AI and advanced modelling partner, contributing expertise in:

• AI/ML development for microbiome-based cancer risk prediction

• multimodal data integration and longitudinal analytics

• explainable AI and bias analysis (age, sex, gender)

• advanced risk modelling and simulation approaches inspired by Virtual Human Twin concepts.

In addition, Acube can involve an external organisation with proven expertise in Virtual Human Twin, strengthening the project’s capacity in computational modelling, simulation frameworks and scalable infrastructures.

Overall, MICRO-PREDICT AI positions itself as a high-impact, multidisciplinary and translational initiative, ready to leverage existing European assets and contribute concretely to the EU Cancer Mission objective of preventing cancer and detecting it earlier, before it becomes a disease.