Understanding how those different levels connect is the challenge at the heart of Research Program 3 (RP3), one of three research programs within CLARA — the Center for Artificial Intelligence and Quantum Computing in System Brain Research.
CLARA is a European center of excellence funded by the European Union's Horizon Europe program and co-founded by the OP JAK complementary funding, coordinated by INDRC together with CIIRC CTU, ICRC, VSB-TUO, LRZ, and the Paris Brain Institute.
RP3 focuses on developing multiscale and cross-modal computational models of Alzheimer’s disease and related processes using artificial intelligence, high-performance computing, and emerging hybrid quantum-classical approaches.
The RP3 team is co-led by Ara Khachaturian and Jean-Marie Bouteiller, with Zaven Khachaturian, Mona Heidari, and Lukáš Drahník contributing complementary expertise across neuroscience, computational modeling, AI, and advanced computing.
Here is a closer look at what the team is building and where the work is headed
FROM MOLECULES TO THE CLINIC: BUILDING ONE CONNECTED MODEL
RP3 focuses on multiscale brain modeling: developing computational approaches that connect processes at the molecular and cellular levels with larger biological systems and, ultimately, observations from real patients.
Today, researchers often study these levels separately.
Molecular biology may describe proteins and signaling pathways.
Cellular neuroscience examines neurons and supporting cells.
Imaging and physiological measurements describe larger networks and systems.
Clinical studies characterize cognition, function, symptoms, and disease progression.
RP3 asks a harder question: can these different levels be connected computationally so that information observed at one scale helps explain, predict, or test what happens at another?
The long-term objective is not simply to create a larger database. It is to develop an in-silico framework in which biological and clinical information from different scales can be represented, modeled, and tested together.
Doing this requires understanding the relationships between scales: how molecular events influence cells, how cellular changes affect networks and biological systems, and how those processes may eventually become visible in measurable patient-level outcomes.
CLARA is also exploring how different computational approaches can contribute to this problem. Classical high-performance computing provides the foundation; AI and generative modeling can help identify complex relationships and generate testable models; and hybrid quantum-classical methods are being investigated for problems where emerging computing architectures may eventually offer advantages.
THE SCIENCE BEHIND THE VISION
The strategic research agenda supporting RP3 is organized around four objectives.
- First, CLARA is developing simulation frameworks capable of representing biological processes across multiple spatial and temporal scales.
- Second, the program is applying generative and other AI-based modeling approaches to the study of resilience, adaptation, dysfunction, and decline.
- Third, these computational models must be tested against real evidence. The validation strategy therefore connects in vitro, in vivo, and clinical data so that increasingly complex models can be evaluated against biological observations rather than remaining purely theoretical.
- Fourth, the work is being designed for long-term scientific value through European and international partnerships, shared infrastructure, and reusable research tools.
These objectives are implemented through three Research and Innovation Priorities, or “RIPs”.
RIP1 focuses on the foundational modeling methods and computational architecture.
RIP2 examines disease mechanisms and multiscale simulation, including the biological interactions that influence resilience, adaptation, and degeneration.
RIP3 focuses on validation, translation, and innovation: determining whether the models perform as intended, whether they can be reproduced across settings, and where they may eventually support useful scientific or clinical applications.
RESPONSIBLE BY DESIGN
Responsible research and innovation is built into the CLARA research strategy rather than added after the technology has been developed.
The program incorporates FAIR and open-science principles, ethical design, diversity and inclusion, and a human-centric approach to artificial intelligence emphasizing transparency, interpretability, and meaningful human oversight.
Just as importantly, CLARA is treating its models, methods, and computational components as reusable scientific infrastructure. The goal is not simply to produce individual publications, but to create tools and frameworks that other researchers can test, challenge, extend, and improve.
TURNING MODELS INTO TOOLS THAT MATTER
The RP3 strategy also identifies several initial areas in which this infrastructure could eventually support practical scientific services.
One is AI-assisted patient stratification for clinical research, helping investigators identify meaningful patient subgroups and potentially improve trial design and monitoring.
A second is a retrieval-augmented knowledge system capable of connecting information across the scientific literature to support evidence synthesis, knowledge mapping, and hypothesis generation.
A third is federated machine learning, which allows institutions to collaborate computationally while sensitive health data remain within the institutions responsible for them. This could become increasingly important as neuroscience research depends on larger, more diverse, and geographically distributed clinical datasets.
These services are not separate from the multiscale modeling effort. They represent different ways in which the underlying scientific and computational infrastructure could ultimately be translated into usable research capabilities.
THE WORK SO FAR — AND WHAT COMES NEXT
During the first 18 months of CLARA, RP3 has concentrated on establishing the conceptual and computational architecture required for multiscale modeling and on beginning an integrated pilot connecting several elements of the research program.
The roadmap extends through 2030 and is deliberately phased.
During 2025–2026, the emphasis is on architecture, framework design, and an initial pilot demonstrating that different components of the program can be connected in a coherent workflow.
From 2026–2028, the focus shifts toward model expansion and validation. The aim is to incorporate richer biological and clinical information and test whether the models reproduce, explain, or predict observations across in vitro, in vivo, and patient-derived data.
From 2028–2030, the program is expected to move toward scale-up and translation, including greater use of European high-performance computing infrastructure and deeper connections with clinical and translational neuroscience networks.
Importantly, this is not simply a calendar of activities. The roadmap includes decision points at which technical performance, biological validity, translational relevance, and sustainability must be assessed before moving to the next stage.
For INDRC, RP3 represents one of the most ambitious parts of the CLARA research program: an attempt to connect levels of brain biology that have traditionally been studied separately and to build computational models capable of moving between them.
The ultimate ambition is to understand not only how the brain deteriorates, but also how it resists injury, adapts to change, and maintains function — and to determine whether those insights can eventually improve the way neurodegenerative disease is studied, predicted, and treated.
It is early-stage work, and the scientific challenges are substantial. But solving problems across scales is precisely what will be required if computational neuroscience is to move from increasingly sophisticated representations of the brain toward models that can be tested against biology and, eventually, used to improve real-world decisions.
Image source: INDRC, AI