CC BY 4.0Wichum, FelixFelixWichumLazzari, Nico DeNico DeLazzariGötte, MiriamMiriamGötteDavid, CorinnaCorinnaDavidWiede, ChristianChristianWiedeSeidl, KarstenKarstenSeidlTewes, MitraMitraTewes2023-08-152023-08-152022-09-02https://doi.org/10.24406/publica-383https://publica.fraunhofer.de/handle/publica/427375https://doi.org/10.24406/publica-38310.1515/cdbme-2022-104410.24406/publica-383Exercise therapy is able to reduce symptom burden in advanced cancer patients (ACP). However, ACP daily form differs between days e.g. tumor and therapy induced. We included five ACP to classify individual exercise capacity due to cardiovascular parameters. Features are extracted from the electrocardiogram and then processed with a neural network after feature selection. Results indicate a high classification quality with an F1 score up to 0.95 ± 0.05. Including neuronal networks for training control can potentially help to manage exercise intensity ideal.enexercise therapyadvanced incurable cancerartificial intelligencemachine learningtraining controlexercise classificationheart rate variabilityrespiration rateheart ratejournal article