Related Experiment Video
Updated: Sep 17, 2025
13:12
Translational Brain Mapping at the University of Rochester Medical Center: Preserving the Mind Through Personalized Brain Mapping
Published on: August 12, 2019
45.8K
Predicting cognitive function 3 months after surgery in patients with a glioma
Sander Martijn Boelders1,2, Bruno Nicenboim1, Elke Butterbrod3,2
1Department of Cognitive Sciences and AI, Tilburg University, Tilburg, The Netherlands.
Neuro-Oncology Advances
|June 27, 2025
Summary
Predicting post-operative cognitive function in glioma patients is challenging. Current models, relying heavily on pre-operative scores, do not reliably forecast individual outcomes, highlighting the need for better data.
Area of Science:
- Neuroscience
- Oncology
- Medical Statistics
Background:
- Glioma patients frequently experience cognitive impairments before and after treatment.
- Accurate prediction of post-operative cognitive function is crucial for personalized treatment planning.
- Optimizing the onco-functional balance requires reliable prognostic tools.
Purpose of the Study:
- To predict cognitive functioning 3 months after surgery in 317 glioma patients.
- To evaluate machine learning models using pre-operative neuropsychological and clinical data.
- To assess the reliability of predictions for individual patient outcomes.
Main Methods:
- Employed nine multivariate Bayesian regression models with a machine-learning approach.
- Utilized pre-operative neuropsychological test scores and clinical predictors.
- Compared model performance using expected log pointwise predictive density (ELPD) and R-squared (R²).
Main Results:
- The best model achieved a median R² of 34.20%, indicating moderate predictive power.
- Pre-operative cognitive functioning was the most significant predictor.
- Models incorporating clinical predictors showed similar performance to those using only pre-operative functioning.
Conclusions:
- Post-operative cognitive functioning could not be reliably predicted using pre-operative data.
- Predictions were highly dependent on pre-operative cognitive status.
- Larger, multi-center, multimodal datasets are needed for more accurate individual predictions.
