The development of medical autonomous systems in general and physiological closed-loop systems in particular requires an increased level of safety in a highly uncertain and volatile environment. To achieve optimal therapeutic outcomes, adaptation to the individual patient is required. Due to the varying clinical environment and the limited information available during therapy, adaptation is often not easy to achieve.
At Fraunhofer IMTE, algorithms are being researched that allow physiological parameters to be influenced while ensuring safe operation for patients and providing additional information to clinical staff. This is increasingly achieved through the use of predictive and model-based controller architectures, which can adapt to the clinical context on a patient-specific basis and integrate machine learning methods. The challenges of device integration and verification are addressed early in the design process and addressed through appropriate architecture selection and the development of test systems.
Research into these aspects offers the potential to increase patient safety, relieve the burden on clinical staff and more effectively utilise the increasing interconnectivity between devices to improve therapy outcomes.