Abstract
This thesis studies whether naturalistic driving data can help predict binary Clinical Dementia Rating (CDR) status while accounting for differences across vehicles. The final analytic dataset comprised 26,968 participant-weeks from 304 participants. Weekly driving features were derived from real-world telematics data and combined with four demographic covariates. Primary model comparisons used leave-one-participant-out (LOGO) cross-validation, with one individual held out at a time and pooled participant-level metrics used as the main reporting surface.
The main comparison includes six model families evaluated on the same dataset under a shared LOGO framework. Performance remained modest overall. GRU-DANN had the highest participant-level ROC AUC (0.599) and balanced accuracy (0.584), Logistic Regression had the highest sensitivity (0.523), and Random Forest was the strongest baseline on ranking-oriented metrics, with ROC AUC of 0.595 and PR AUC of 0.355. DANN did not clearly outperform simpler baselines.
Taken together, these results might suggest that naturalistic driving data may carry limited information related to cognitive status, but the present pipeline should be interpreted as an empirical comparison of modeling choices rather than as a clinical screening tool. The main contribution of this thesis is a comparison of baseline, domain-adversarial, and sequence-based modeling choices, together with a clearer account of what this dataset and pipeline can and cannot currently support.
Committee Chair
Alvitta Ottley
Committee Members
Nathan Jacobs, Ganesh Babulal
Degree
Master of Science (MS)
Author's Department
Computer Science & Engineering
Document Type
Thesis
Date of Award
Spring 5-6-2026
Creative Commons
This work is licensed under a Creative Commons Attribution 4.0 International License.
Language
English (en)
Author's ORCID
http://orcid.org/0009-0009-5495-2879
Recommended Citation
Gopala Reddy, Aadarsha, "Toward Vehicle-Agnostic Driving Signatures for Cognitive Impairment Prediction from Naturalistic Driving Data" (2026). McKelvey School of Engineering Theses & Dissertations. 1341.
https://openscholarship.wustl.edu/eng_etds/1341