Sensory seizure detection


Epilepsy is among the most common neurological diseases in the world. Recurring epileptic seizures that put great strain on affected individuals and the people who care for them are a typical symptom. The unpredictability of the seizures as well as impaired consciousness and the loss of control over various bodily functions contribute to the risk that affected individuals may sustain serious injuries, making care more difficult. Detecting seizures in a timely manner can support nursing staff and help them take corresponding steps to ensure the patient’s safety. Aside from early detection, precisely recording the seizures helps with individually tailored therapy.



The challenge

The EPItect project aims to develop a non-invasive sensor system that detects biosignal patterns of relevance for predicting epileptic seizures. Patients will be able to wear the sensor on their ear. Mobile devices provide the data to select people. This means the nursing environment can be included as needed. New seizure detection models are being especially developed in the project. The suitability for daily use is being evaluated in the clinical and home environments. Special emphasis is being placed on taking data protection law provisions into account. Various technical innovations such as alarm services and a mobile attendant solution are being implemented based on the sensor system and networking infrastructure. Not only does this improve care for people with epilepsy, it also enhances their safety, self-determination, and quality of life.


Our contribution

Fraunhofer has assumed responsibility for the requirements analysis under consideration of the user perspective for technical innovations in the project. International standards such as Health Level 7 (HL7), Fast Healthcare Interoperability Resources (FHIR), and Integrating the Healthcare Enterprise (IHE) as well as the applicable legal frameworks were taken into account in the conceptual design of the IT solutions on this basis. The sensor-based networking infrastructure together with the mobile application for patients and the portal developed for patients, caregivers, and professional nursing staff was evaluated in the course of a user study with 40 patients and family members. Another focal point was the development of models for the detection of epileptic seizures based on the sensor data collected within the scope of the project. Models were developed for the detection of tonic-clonic seizures based on acceleration data and the heart rate using sensor data collected in the course of the project. Whether epileptic seizures can be detected using blood pressure fluctuations determined based on ECG and PPG data was also investigated.



With the developed networking infrastructure and the IT applications for networking the participants in the care/treatment process for patients with epilepsy, an initial proof of concept study was conducted for innovative solutions for networking and the detection of seizures. The user study with patients and family members showed that there is great interest in mobile technologies for health management and networking with the attending doctors. Large data volumes for training seizure detection models were collected in the course of a clinical study with 200 test subjects. The scientific results of this study were presented at scientific conferences.



  • Department of Epileptology at the University Hospital of Bonn
  • cosinuss° GmbH, Munich
  • Kiel University Clinic for Neuropediatrics
  • Norddeutsches Epilepsiezentrum in Schwentinental-Raisdorf


Associated partners   

  • University of Health Sciences (hsg Bochum)
  • Epilepsie Bundes-Elternverband e.V. Wuppertal
  • Landesverband für Epilepsie Selbsthilfe Nordrhein-Westfalen e.V.



  • Subsidized by: Federal Ministry of Education and Research (BMBF) under the program “Nursing innovations to support caregivers and professional nursing staff”
  • Project number: 16SV7482
  • Term: March/2016-08/2019




Brochures and reports for download

Information sheet

Sensorische Anfallsdetektion