Simulation-based AI for the Prediction of Stroke

Find out more!

Predicting Stroke

AI for precision medicine in stroke.

Stroke Risk Prediction

Our innovative simulation model enables predictive diagnostics in stroke treatment using digital medical imaging. We process patient-specific MRI and CT images with state-of-the art processing-tools and convolutional neural nets. This information is fed to our unique simulation that allows the estimation of cerebral perfusion changes under different conditions.

The simulation results together with routine clinical data allow predictive modelling for stroke risk and outcome using innovative machine learning techniques. Our approach will lead to personalized simulation of therapy options in stroke enabling patient-centered treatment and personalized risk assessment.


Each year more than 1.2 million people suffer from stroke in Europe alone. 15% of which have a re-stroke within a year.

Prevention strategies and clinical treatment of stroke are based on generalized guidelines and empirical data alone. No personalized diagnostics and treatment strategies are available. Digital opportunities are not exploited sufficiently.

With our AI solution we will save lives of stroke patients by personalizing stroke prevention and treatment utilizing high-precision imaging and unique Machine Learning approaches.

Towards Personalized Therapy

Avoid interventions where possible. Perform minimal therapy when necessary.

High Quality Data

For validation of the product we are using high-quality international, multi-center clinical data.

Patient clinical data, simulation results and treatment information are used to feed state-of-the-art machine learning algorithms to calculate stroke risks, to predict patient outcome and risk of recurring stroke. We study the impact of our clinical AI solution on recommended treatment strategies and examine, if subsequent strokes can be mitigated when our software is used.

Stroke Risk Score & Map

Risk prediction and therapy planning will be dramatically improved using our Stroke Risk Scores.

Treatment options will be provided in transparent reports including risk maps and detailed simulation results. The risk profile of competing treatment options can be quantified and compared. Relevant neurologic features and associated risk factors are visualized using individual patient imaging.

Routine Imaging

Unlike in competitive approaches only routine and standard medical imaging is required for successful simulations and data processing. No additional cost, time or patient involvement is necessary.

For automated image post-processing by convolutional neural nets we use high-precision imaging to allow unprecedented processing speed for real time analysis in the clinical setting.

Empowering Patients

Our innovative approach will help patients receive the best treatment.

Individual Therapy

Treatment decisions are based on patients and their relevant medical indications.

Informed Patient

Reports allow transparent feedback on the risk factors and treatment options and empower the patient for informed consent.


Software-based prediction and simulation are performed without additional procedures.

Trusted & Innovative

Developed at one of the world's top medical research institutions.

Funded by the Federal Ministry of Education and Research as part of GO-Bio 7.

Working Prototype

Our predictive model and simulation component is implemented in a working prototype.

Cost Effective

No requirement on input other than what is routinely gathered in a stroke event.

Funded by

Researched at

Located in

Our Supporters

Our Partners

Dublin Insitute of Technology Dublin Institute of Technology Ireland
Estonian Genome Center Estonian Genome Center Estonia
German Research Center for Artificial Intelligence German Research Center for Artificial Intelligence Germany
Linköping University Linköping University Sweden


Dietmar Frey
Dietmar Frey, M.D. J.D.
Project Lead
Vince Istavan Madai
Vince I. Madai, M.D. M.A.
Scientific Lead
Jasmin Kopetzki
Jasmin Kopetzki
Study Assistant
Heiko Leppin
Heiko Leppin, Dipl.-Inf.
Head of Software Development
Jan Oopkaup
Jan Oopkaup, M.Sc.
Senior Software Developer
Michelle Livne
Michelle Livne, Ph.D.
Head of Machine Learning
Tabea Kossen
Tabea Kossen
Machine Learning Expert
Hannes Karras
Hannes Karras
Research Assistant
Orhun Utku Aydin
Orhun U. Aydin
Doctoral Student
Elâ Marie Akay
Elâ M. Akay
Doctoral Student
Esra Zihni
Esra Zihni
Master Student
Tabea Strunk
Tabea Strunk
Master Student
Drop us an e-mail
or contact us at
Dietmar Frey, M.D. J.D.
Department of Neurosurgery
Charité Universitätsmedizin
Augustenburger Platz 1
D-13353 Berlin
+49 30-450 560398
+49 30-450 560989