Predicting the Degree of Distracted Driving Based on fNIRS Functional Connectivity: A Pilot Study

Ogihara, Takahiko; Tanioka, Kensuke; Hiroyasu, Tomoyuki; Hiwa, Satoru · 2022 · Crossref

DOI: 10.3389/fnrgo.2022.864938

archive: archived pipeline: cataloged

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

Summary

This pilot study addresses the challenge of predicting the degree of distracted driving by analyzing brain activity during actual vehicle operation. While distracted driving is a major cause of traffic accidents, few studies have measured neural correlates during real car driving due to the complexity of the task compared to simulator-based experiments. The research aims to develop a predictive model that estimates driver distraction levels using functional near-infrared spectroscopy (fNIRS) data, specifically leveraging functional connectivity (FC) to predict behavioral responses. The motivation is to create systems that can monitor attentional states to prevent accidents and to elucidate the neural basis of mind-wandering during complex multitasking. The study involved twelve healthy male participants who drove a small electric vehicle on a 40-meter oval course for 15 minutes at 20 km/h. Random beep tones were presented at intervals of 20–40 seconds, requiring participants to brake and decelerate to 10 km/h. Brake reaction time (BRT) served as the objective behavioral index of distraction, while fNIRS measured oxyhemoglobin concentration changes across 44 channels. The researchers constructed individual regression models for each participant, using FC matrices (calculated from 10, 15, or 20-second windows) as explanatory variables and BRT as the target. A bootstrap-based feature selection method identified stable FC edges, which were then used in ordinary least squares regression. Model performance was evaluated using leave-one-out cross-validation, with hyperparameters optimized to maximize prediction accuracy on held-out test data. Results indicated that the regression models successfully reduced the dimensionality of FC data from 946 features to fewer than the sample size. For 11 of the 12 participants, the models achieved statistically significant correlations between predicted and measured BRTs, with an average mean absolute error of 5.58 × 10² ms. Hierarchical clustering of the selected FC edges revealed five distinct clusters of prediction models. A common feature across all clusters was the involvement of connections between the dorsal attention network (DAN) and the sensory-motor network (SMN), as well as connections between the DAN and the ventral attention network (VAN). These findings suggest that these specific network interactions are essential for predicting distraction levels in complex driving tasks. The significance of this work lies in demonstrating that fNIRS-based functional connectivity can effectively predict the degree of distracted driving in real-world conditions. The identification of specific neural network interactions (DAN-SMN and DAN-VAN) provides insight into the neural mechanisms underlying attentional lapses during driving. These results support the development of advanced driver assistance systems that monitor cognitive states to enhance road safety and contribute to the broader understanding of the neural basis of mind-wandering and attentional control.

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 qwen3.8-27b-gittensor on 2026-08-23; verification: pending re-verification.

Topics

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

• neuro workload indices
• distraction detection algorithms
• cognitive
• mind wandering

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: physiological data
• Theoretical Contribution: computational model, theory or model