Objective: The present study aims to evaluate a multivariable model to predict verbal working memory (WM) performance following a targeted cognitive training intervention, WM training (WMT) in patients with Parkinson’s disease (PD) using a supervised machine learning approach. Furthermore, we compare the predictiveness of single data domains, namely cognitive, clinical, and demographic variables from baseline alongside learning parameters extracted from the training performance itself.

Background: WMT is a promising intervention approach against cognitive decline in patients with PD. However, heterogeneity in WM improvement suggests that WMT may not be equally efficient for all patients.

Method: 37 patients with PD (age: 64.09±8.56, 48.6% female, 94.7% Hoehn & Yahr stage 2) participated in a 5-week WMT with maximally 25 training sessions. Four random forest regression models were built to predict immediate and 3-month follow-up WM performance: The ‘cog’ model consists of variables assessing baseline performance in different cognitive domains. The ‘learning’ model includes slope and intercept parameters for each composite WMT score. The ‘cog/learning’ model combines the ‘cog’ and ‘learning’ variables and the ‘all’ model additionally includes clinical and demographic information. Model performance was quantified by root mean square error (RMSE), compared using pairwise permutation tests.

Results: The ‘all’ model containing the entire set of variables predicted verbal WM with the lowest RMSE compared to the other models using partial number of features, at both immediate (RMSE 0.184; 95%-CI=[0.184;0.185]) and 3-month follow-up (RMSE 0.216; 95%-CI=[0.215;0.217]). Cognitive baseline parameters were among the most important predictors, followed by learning parameters. The ‘cog/learning’ model significantly outperformed the ‘cog’ model.

Conclusion: Demographic and cognitive variables, that are commonly used in WMT research, provide robust prediction of response to WMT. Nonetheless, consideration of training-inherent learning parameters further boosts precision of prediction models that may maximize training benefits following cognitive interventions in patients with PD.

« Back to 2022 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/cognitive-performance-and-learning-parameters-predict-response-to-working-memory-training-in-parkinsons-disease/