GPS driving: a digital biomarker for preclinical Alzheimer disease

Bayat, Sayeh; Babulal, Ganesh M.; Schindler, Suzanne E.; Fagan, Anne M.; Morris, John C.; Mihailidis, Alex; Roe, Catherine M. · 2021 · OpenAlex-citations

DOI: 10.1186/s13195-021-00852-1

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates the potential of naturalistic driving data, captured via Global Positioning System (GPS) devices, to serve as a digital biomarker for preclinical Alzheimer’s disease (AD). Preclinical AD is characterized by underlying brain pathology, such as amyloid accumulation, in individuals who remain cognitively normal. Current diagnostic methods rely on invasive and costly procedures like lumbar puncture or imaging. The authors hypothesize that subtle functional changes associated with early AD pathology manifest in complex behaviors like driving, offering a low-cost, non-invasive alternative for early detection. The research utilized a cohort of 139 cognitively normal older adults (aged 65+) from longitudinal studies at the Washington University Knight Alzheimer Disease Research Center. Participants were classified into two groups based on cerebrospinal fluid (CSF) biomarkers: 64 with preclinical AD (defined by low Aβ42/Aβ40 ratios) and 75 without. Naturalistic driving data was collected over one year using in-vehicle GPS loggers installed in participants’ personal vehicles. The study analyzed a comprehensive set of GPS-derived indicators describing driving performance (e.g., speed, acceleration, jerk) and driving space (e.g., trip distance, radius of gyration, entropy). Four Random Forest machine learning models were trained to distinguish between the two groups using different combinations of input features: (1) age and APOE ε4 status only, (2) driving indicators only, (3) driving indicators and age, and (4) driving indicators, age, and APOE ε4 status. The results demonstrated that GPS driving data could effectively identify preclinical AD. The model using driving indicators alone achieved an F1 score of 0.82 and an area under the receiver operating curve (AUC) of 0.82. Performance improved significantly when demographic and genetic factors were included; the model combining driving indicators with age achieved an F1 score of 0.88 and an AUC of 0.94. The final model, incorporating driving indicators, age, and APOE ε4 status, achieved the highest performance with an F1 score of 0.91 and an AUC of 0.96. Feature importance analysis revealed that APOE ε4 status and age were the strongest predictors, followed by driving-specific metrics, particularly average jerk (smoothness of acceleration/deceleration), number of night trips, and radius of gyration. The findings suggest that GPS-based driving analysis is a promising, accurate, and accessible digital biomarker for identifying preclinical AD. The study highlights that while driving features alone provide robust prediction, combining them with age and genetic risk factors yields the highest accuracy. This approach offers a scalable, non-invasive method for early detection, potentially enabling earlier intervention. The authors note that while the study was limited to a specific geographic region and sample size, the results support further investigation into using naturalistic driving data alongside emerging blood-based biomarkers for broader clinical application.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• mci dementia driving