# Exploring Factors Influencing Driving Simulator Performance in Patients With Acquired Brain Injury Using Hierarchical Clustering Analysis of Principal Components

## TL;DR

This study identifies cognitive factors affecting driving simulator performance in brain injury patients and suggests personalized rehabilitation could improve outcomes.

## Contribution

The study introduces a novel clustering approach to classify ABI patients based on simulator performance and cognitive profiles.

## Key findings

- Three distinct patient clusters were identified based on driving simulator performance and cognitive profiles.
- Cluster 1 showed superior cognitive abilities and consistent high performance compared to other clusters.
- Tailoring rehabilitation to cognitive profiles may improve driving outcomes in ABI patients.

## Abstract

Background

Driving simulator training is widely recognized as an effective tool for driving rehabilitation. However, the key factors influencing simulator performance and the extent of training-related improvements remain insufficiently explored. This study aimed to identify the demographic, motor, and cognitive factors associated with driving simulator performance and post-training improvements in patients with acquired brain injury (ABI) using clustering analysis.

Methods

A total of 64 patients with ABI (59% cerebral hemorrhage, 34% cerebral infarction, 7% traumatic brain injury; mean age 64±13 years; 81% male) underwent comprehensive neuropsychological assessments and driving simulator evaluations before and after training. Multiple factor analysis was applied to integrate pre- and post-training variables and reduce dimensionality. Hierarchical Clustering on Principal Components was then performed to classify patients based on training effect patterns. The Kruskal-Wallis test and post hoc multiple comparisons were used to assess differences in background factors among the clusters.

Results

Three distinct clusters were identified: Cluster 1 (n=32) exhibited consistently high performance in reaction and city-driving tasks, Cluster 2 (n=19) demonstrated prolonged reaction times but showed significant improvements in city-driving tasks after training, and Cluster 3 (n=13) demonstrated severe city-driving errors and limited post-training improvement. Neuropsychological assessments revealed significant differences among the clusters (p < 0.05), with Cluster 1 consistently outperforming Clusters 2 and 3 across multiple cognitive domains, including attention, cognitive flexibility, visuospatial abilities, memory, and executive function.

Conclusion

Neuropsychological assessments may serve as predictors of both baseline driving performance and post-training improvements. Tailoring interventions to individual cognitive profiles, particularly focusing on attention, visuospatial abilities, and executive function, may enhance the efficacy of simulator-based rehabilitation and support the safe resumption of driving. Future longitudinal studies should examine how targeted cognitive training might improve driving performance in patients with different cognitive profiles.