Impact of acute glucose episodes on adherence to speed limits in the naturalistic setting for drivers with diabetes: An application of linear quantile mixed models

Abstract

Diabetes can cause complications from hypoglycemic and hyperglycemic episodes, impairing cognitive and motor skills essential for safe driving. Advances in low-cost sensors and wearable technologies have enabled naturalistic driving studies (NDS) that simulate real-world conditions while monitoring drivers’ blood glucose levels. This paper analyzes data from an NDS conducted in Nebraska, focusing on how drivers with Type 1 Diabetes Mellitus (T1DM) and Type 2 Diabetes Mellitus (T2DM), as well as control participants without diabetes, adhere to speed limits of 50–75 mph on highways. Alongside a conventional Linear Mixed Effects Model (LMM), we introduce a novel Linear Quantile Mixed Effects Model (LQMM) to evaluate six quantiles (τ = 0.10, 0.25, 0.50, 0.75, 0.85, and 0.90) of speed limit adherence during acute glucose episodes, including hyperglycemia and hypoglycemia. Findings show that hypoglycemia generally leads T1DM drivers to drive more cautiously and remain below speed limits. No significant effects of hypoglycemia or hyperglycemia were observed on T2DM drivers’ speed adherence, suggesting glycemic fluctuations may not substantially influence their behavior. Hyperglycemia was linked to increased caution among T1DM drivers, consistent with evidence of heightened physiological awareness in this group. Control drivers exceeded speed limits more often than those with diabetes, especially relative to T2DM drivers. Roadway characteristics (e.g., traffic flow and speed limits) and age also influence speed behavior, highlighting important contextual factors. By utilizing distribution-based methods like LQMMs that account for participant heterogeneity, this paper presents a nuanced view of speed control patterns, yielding new insights into how diabetes affects driving safety.