This presentation offers an overview of recent advancements in data warehouse-based predictive modeling relevant for the stroke unit setting and with a particular focus on predicting pneumonia, patient outcomes, and atrial fibrillation. We have integrated multimodal data routinely collected during standard stroke workup, particularly focusing on monitoring data that affords automated analyses with high feasibility. Our research has resulted in the creation of robust algorithms to anticipate outcomes and complications post-stroke. Heart rate variability was identified as a key driver of our models‘ performance in all use cases. These machine learning approaches demonstrate improved performance when benchmarked against established clinical scores and highlight a path forward to real-time, prospective validation and intervention in a next step.

Dr. Maximilian Schöls is a physician in training at the Department of Neurology at the Charité - Universitätsmedizin Berlin and scientifically active in the Computational Neurology working group of PD Dr. Meisel. He is particularly interested in scalable approaches to optimize stroke care.

Alexander Nelde, M. Sc. is a physicist with a master’s degree from TU Berlin, specialized in complex systems and nonlinear dynamics.  Within the working group Computational Neurology, his scientific focus is on the development of multimodal prediction models and signal analysis.

Join the talk online here!