IKTUS: Cloud-based Explainable & Causal AI for Stroke (and Cardiovascular) Risk Detection & Prognosis

1. Challenge / Need

Stroke remains a leading cause of death and long-term disability. Clinicians often need fast, interpretable decision support for risk stratification and prognosis, using data that is available early (and not always “perfect”).

2. Our Solution (What ICTUS is)

ICTUS is a cloud AI service for detection and prognosis of stroke risk factors, producing clinician-readable, explainable reports from easy-to-access data (e.g., routine clinical variables, basic tests, patient-reported data when relevant).
It also enables “what-if” scenario simulation to explore the impact of modifiable risk factors and treatment pathways.

3. Key Innovations / Technologies

• Explainable AI using Tsetlin Machines (transparent rule-based patterns rather than black-box outputs)

• Causal AI to support counterfactual / scenario simulation (“what-if”)

• Designed for interpretability and traceability to support clinical uptake and trust

• Background: research output includes a paper presented/published in an explainability venue (Explainability 2025)

4. Value / Expected Impact

• Actionable traceability of stroke-related risk factors

• Faster clinical response through immediate decision support

• Easy-to-read outputs for medical staff (not only data scientists)

• Generalizable approach across settings and potentially broader cardiovascular risks
  This aligns with Cluster 1 priorities around AI-enabled health solutions and work on cardiovascular risk prediction / early detection.

5. Fit with 2026 Cluster 1 Health (indicative)

We see strong fit with 2026 directions related to non-communicable diseases and cardiovascular health, including work that targets risk prediction, early detection, and digital approaches.
(If helpful for the platform, you can reference the 2026 call structure/opening info in the Work Programme.)

6. What We Offer to the Consortium

• Corporate-grade R&D + delivery: end-to-end product engineering (cloud architecture, MLOps, cybersecurity-by-design, deployment)

• AI methods & implementation: explainable + causal modelling, robustness testing, bias assessment

• Clinical decision-support outputs: interpretable reporting, clinician UX flows, validation-ready documentation

• Data strategy: privacy-preserving pipelines, governance, interoperability planning (FHIR-ready approach if requested)

• Pilot execution: technical pilots with KPI framework, integration with partner environments

7. Partners We Are Looking For

We want to build a balanced consortium including:

• Clinical partners: hospitals / stroke units / cardiovascular prevention clinics (access to cohorts + clinical pathways)

• Research organisations: epidemiology, neurology, cardiovascular risk, causal inference, biostatistics

• Data holders / infrastructures: registries, biobanks, RIs/ERICs, secure data platforms

• Industry: medtech / digital therapeutics / clinical workflow tools (integration & exploitation)

• Ethics / legal / HTA: GDPR, medical device pathway (MDR/IVDR where relevant), health economics, adoption

• Patient & public involvement: patient associations for co-design and acceptability

8. Role We Can Take

• WP leader for AI platform development, integration, and validation tooling

• Technical lead for explainable/causal AI work package

• Partner (or co-lead) for pilot deployment and scalability plan
  (Open to coordinator role if the topic and consortium are mature.)

9. Expected TRL / Maturity (editable)

• Current maturity: prototype / pre-commercial demonstrator (estimate TRL 4–6 depending on your current state)

• Target by project end: validated pilot(s) in real clinical settings + exploitation-ready roadmap

10. Keywords

Stroke, cardiovascular risk, non-communicable diseases, clinical decision support, explainable AI, causal AI, Tsetlin Machines, what-if simulation, risk prediction, digital health, cloud platform, MLOps, GDPR, MDR.