PHARMOPRO-KI is a collaborative project between DFKI and the Department of Anaesthesiology at Saarland University Hospital, commissioned by the Saarland State Research Fund. As part of the project, the project partners will investigate whether AI-based pharmacometric models can provide more accurate results than the current standard models in the case of Propofol.
Pharmacometric models (PM) are used by physicians to predict the drug concentration in a patient's body over time and to determine a dose optimised for their physical characteristics. In the case of the intravenous anaesthetic Propofol, which is used for general anaesthesia and sedation in intensive care units, pharmacometric models are particularly relevant. Propofol is one of the few drugs for which PM is used directly in special syringe pumps. However, the mathematical and statistical methods used have remained unchanged since the 1980s and the Propofol concentrations calculated using these standard models are subject to considerable error.
DFKI will pursue two different approaches. Firstly, a purely data-based AI model based on neural networks and random forest algorithms will be evaluated. In addition, a hybrid model will be developed that is based on the standard model but optimises its classically calculated parameters using AI methods.
Partners
Lehrstuhl für Anästhesiologie, Fakultät M, Universität des Saarlandes (UdS), Universitätsklinikum des Saarlandes (UKS) / Chair of Anaesthesiology, Faculty M, Saarland University (UdS), Saarland University Hospital (UKS)