Virtual Human Twin Models for Brain Tumour Progression Modelling & Personalised Treatment Response HORIZON-MISS-2026-02-CANCER-01

1) EXCELLENCE

Vision: Create dynamic, multiscale Virtual Human Twins (VHTs) for hard-to-treat brain tumours (e.g., high-grade gliomas / early-onset and refractory cases) that combine mechanistic modelling + multimodal AI, updated longitudinally to simulate disease evolution and compare treatment strategies in silico, and made accessible via UNCAN.eu and the Advanced VHT Platform.

Specific objectives (science & tech)

• O1. Multiscale VHT core: tumour–host models spanning molecular → cellular → tissue/organ levels, capturing infiltration, angiogenesis, oedema, therapy response and resistance.

• O2. Patient-specific parameterisation: infer VHT parameters from MRI (multi-sequence), pathology-derived phenotypes, (when available) omics/immune profiles, and real-world clinical data; represent uncertainty explicitly.

• O3. Longitudinal updating: “twin refresh” after each follow-up scan/clinical event using Bayesian/data-assimilation methods to track progression under therapy.

• O4. Clinically meaningful validation: demonstrate predictive/decision utility for progression and treatment response, with usability evidence for clinical uptake.

• O5. Open, interoperable delivery: publish reusable model/data assets with metadata in line with EU requirements and platform integration.

Core innovation (what is new)

• Mechanism + AI coupling: AI provides robust patient-specific states (tumour compartments, uncertainty, phenotypes); mechanistic simulators provide time-evolving causal dynamics and counterfactual treatment testing.

• Explainable VHT “drivers”: attention/XAI outputs are aligned to mechanistic variables (e.g., infiltration rate, radiosensitivity proxies) to produce actionable explanations, not just heatmaps.

• Interoperability by design: assets packaged as containers/APIs for integration into EU platforms and imaging ecosystems. +

2) IMPACT

Contribution to expected outcomes

• Advanced multiscale VHTs used by multidisciplinary researchers to understand onset/progression mechanisms (tumour–host–immune interactions; therapy resistance).

• VHT-based solutions for personalised treatment (simulated therapy pathways with uncertainty) enabling improved treatment stratification.

• Access via UNCAN.eu + Advanced VHT Platform with open-science assets and reusability across Europe.

Key exploitable results

• VHT-Brain Engine (hybrid mechanistic + AI inference, HPC-ready)

• Longitudinal “twin refresh” toolkit (calibrated updating from follow-up imaging)

• Clinician-facing decision dashboard (uncertainty-aware scenario comparison; explainability)

• Open benchmarks & synthetic twin generator (privacy-preserving validation and stress tests)

3) IMPLEMENTATION

Consortium partners:

• Clinical neuro-oncology partners: cohorts, endpoints, tumour board workflow integration, prospective observational validation.

• Your AI group (key competence): transformer segmentation, explainable attention, multimodal fusion, optimisation, survival/outcome modelling—feeding patient-specific states and uncertainty into the VHT.

• Mechanistic modelling/HPC partners: tumour growth/invasion solvers, therapy PK/PD modules, scalable computing.

• Data/standards & platform partners: FAIRification, metadata pipelines, UNCAN.eu and VHT platform asset packaging and interoperability.

Work plan (WPs)

• WP1 Coordination, ethics, patient involvement (sex/gender considerations; governance)

• WP2 FAIR multimodal data pipeline (imaging + clinical + optional omics/immune; harmonisation; metadata)

• WP3 AI perception layer (segmentation/classification/phenotypes; XAI/attention; uncertainty)

• WP4 Multiscale VHT engine (mechanistic models + AI coupling; twin refresh)

• WP5 Verification, validation & clinical usability (retrospective + observational prospective; decision studies)

• WP6 UNCAN.eu & Advanced VHT Platform integration + open science (assets, APIs, documentation)

• WP7 Mission cluster networking (budgeted participation in the Cancer Mission “Understanding” cluster)