Driving Performance as a Marker of Cognitive Status: A Machine Learning‐ based Approach
Bibliographic record
Abstract
BACKGROUND: Driving is essential for maintaining independence and social engagement in many older adults, serving as both a practical mode of transportation and a symbol of autonomy. Aging, however, is associated with cognitive changes and increased risk of cognitive impairment (e.g. mild cognitive impairment and dementia). These changes in cognition can affect the complex skills required to drive and previous studies have suggested that changes to driving behavior can serve as an early marker of cognitive decline. This study aimed to determine whether machine learning can detect cognitive status (cognitively impaired vs. cognitively unimpaired) based on driving performance. METHOD: Eight cognitively healthy older adults and seven older adults with diagnosed cognitive impairments (mild cognitive impairment and very mild dementia) drove through various everyday scenarios in a high-fidelity driving simulator. Driving performance metrics such as steering wheel angles, lane deviation, acceleration, and braking intensity were analyzed. Statistical features, including means, standard deviations, ranges, and cumulative changes, were extracted using sliding windows. A Random Forest model was developed as the primary predictive tool to detect cognitive status, given its ability to handle non-linear relationships, and feature interactions. Model performance was compared with other machine learning models. Feature importance in the Random Forest model was assessed by evaluating each feature's contribution to reducing impurity during tree construction. A refined Random Forest model trained with the top five features was also evaluated. RESULT: The Random Forest model achieved the highest accuracy (66.67%) using all features, followed closely by Gradient Boosting (66.07%), with K-Nearest Neighbors (63.10%), Decision Trees (62.50%), Support Vector Machine (60.12%), and Logistic Regression (58.93%). Feature importance analysis identified key predictors: cumulative and minimum acceleration, maximum and minimum steering wheel angle, and lane gap range. Using these top five features, the Random Forest model's accuracy improved to 70.83%. CONCLUSION: This study demonstrates the feasibility of using machine learning to classify cognitive status based on driving performance. Findings suggest that driving metrics have potential as a tool for detecting cognitive decline. In the future, we will extend this approach to detect pre-clinical cognitive decline in individuals with subjective cognitive decline, assess AI's reliability, and support timely interventions.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".