RMRyan Miller

Ryan Miller

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Ryan Miller has a total of 35 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Detection of alcohol-involved driving via machine learning using eye-tracking and vehicle data

    by Ryan Miller, Andy Chestovich, Anshul Satav, Angela Eichelberger, Jonny Kuo, Michael Lenne, Timothy Brown

  • Clustering Roadway Departures from Impaired Driving Studies using Derived Features and Shape-Based Methods

    by Ryan Miller, Arsal Shaikh, Tannishtha Mukherjee, Chris Schwarz, Timothy Brown

    Lane departures remain a major contributor to fatal traffic accidents despite the increasing adoption of lane departure warning systems. Existing research has focused on the detection and mitigation of lane departures, while the characteristics of departures and their potential relationships with categories of driver impairment have received less attention. We apply a variety of unsupervised clustering approaches to a comprehensive set of lane departures from three drug-impaired driving studies involving combinations of cannabis and low-dose alcohol, alcohol, and cannabis. We find that departures are more frequent in active-drug conditions, ranging from 26% to 222% increases in mean departures per minute of driving depending on the study and condition. Across all studies, departures are most effectively grouped into two clusters based on duration and peak lateral deviation, with average silhouette scores of the most effective approaches being 0.778, 0.759, and 0.871, respectively, indicating high degrees of cohesion within clusters and separation between them. After clustering on the principal components derived from driving data from before, during, and after each departure we find clusters that differ significantly in their relative proportions of alcohol-dosed drivers ( p < 0.001). Shape-based approaches also identify two clusters of departures that differ in duration, maximum deviation, and timing of peak deviation. These clusters contain a significantly different composition of dosing conditions in the combined alcohol and cannabis study ( p = 0.015). Our results suggest lane departures can be effectively profiled and that leveraging their characteristics may benefit driver monitoring systems targeting the detection of drug-impaired driving.

  • Detecting Impaired Driving: Vehicle Inputs with the Greatest Influence on Accurate Identification

    by Ryan Miller, Bradley Carlton, Timothy Brown

    Alcohol and drug impairment are major causes of crashes and fatalities. Despite evidence of significant differences in lateral control, speed, and steering across impaired and sober states, leveraging this information for the automated detection of impairment using vehicle inputs remains a challenge. Using data from two driving simulator studies involving sources of impairment (cannabis, alcohol, and combined cannabis and alcohol) and a range of driving environments, we develop machine learning models to identify acute drug use using vehicle data with a focus on understanding which features exert the greatest influence on predictions across a variety of scenarios. Under a robust validation framework, we used SHapley Additive exPlanations to determine that features derived from lateral position were consistently among the most predictive, whereas vehicle speed, steering wheel rate, and brake force also made meaningful contributions. Features derived from lateral position were more predictive when calculated over longer periods (60 s) relative to shorter time spans (0.33 to 5 s). Leveraging subject-specific information tended to improve classification performance. For detecting recent cannabis use (~30-min postdose), receiver operating characteristic—area under the curve (AUC) improved from 0.0625 to 0.706 in a straight, two-lane rural road scenario, and for classifying combined cannabis and alcohol impairment ROC-AUC improved from 0.643 to 0.689 on a straight segment of a four-lane divided expressway. These findings highlight the potential for carefully selected features derived from vehicle inputs to contribute to automated impairment detection, and the need to consider driving environments and sources of impairment when using these features.

  • Feature‐specific inference for penalized regression using local false discovery rates

    by Ryan Miller, Patrick Breheny

    Penalized regression methods such as the lasso are a popular approach to analyzing high‐dimensional data. One attractive property of the lasso is that it naturally performs variable selection. An important area of concern, however, is the reliability of these selections. Motivated by local false discovery rate methodology from the large‐scale hypothesis testing literature, we propose a method for calculating a local false discovery rate for each variable under consideration by the lasso model. These rates can be used to assess the reliability of an individual feature, or to estimate the model's overall false discovery rate. The method can be used for any level of regularization. This is particularly useful for models with a few highly significant features but a high overall false discovery rate, a relatively common occurrence when using cross validation to select a model. It is also flexible enough to be applied to many varieties of penalized likelihoods including generalized linear models and Cox regression, and a variety of penalties, including the minimax concave penalty (MCP) and smoothly clipped absolute deviation (SCAD) penalty. We demonstrate the validity of this approach and contrast it with other inferential methods for penalized regression as well as with local false discovery rates for univariate hypothesis tests. Finally, we show the practical utility of our method by applying it to a case study involving gene expression in breast cancer patients.


  • Influence of cannabis use history on the impact of acute cannabis smoking on simulated driving performance during a distraction task

    by Ryan Miller, Tim Brown, Julia Wrobel, Michael J. Kosnett, Ashley Brooks-Russell

    Objective: This research explores the driving performance of people who use cannabis daily or occasionally during distraction tasks performed following acute cannabis use. Methods: Healthy adults aged 25 to 45 years with different cannabis usage histories were recruited to participate in a within-subjects controlled experiment using a car-based driving simulator. Participants were classified as having daily use (n = 31), occasional use (1 or 2 times per week; n = 24), or no-use (n = 30). Participants completed a practice drive followed by four 5–10 minute driving scenarios during the baseline period. Participants then smoked self-procured cannabis flower ad libitum for up to 15 minutes. Thirty minutes later, they completed four additional 5–10 minute scenarios. Scenarios were paired according to difficulty and randomized across the baseline and post-use periods. Each scenario contained between 0 and 3 repetitions of a distraction task where the participant was prompted by an audio message to select an app from a 4 × 5 grid displayed on a mounted tablet, a step that would require briefly looking away from the roadway. Measures of driving performance (lane departures, standard deviation of lateral position) were assessed during the five-second period following the audio trigger and analyzed using generalized linear mixed models. Results: Those with a pattern of occasional use were significantly more likely to experience a lane departure during distraction periods after acute cannabis use relative to baseline (OR = 3.71, p = 0.04, CI = 1.04, 13.17), while those with daily use did not exhibit a similar increase (OR = 1.56, p = 0.43, CI = 0.52, 4.64). Changes in departure risk were significantly greater for the occasional use group compared to no-use (p = 0.02), but not for the daily use group compared to no-use (p = 0.18). However, following acute use, those who use daily exhibited decreases in speed relative to baseline in comparison to the changes observed in the no-use group (p = 0.02), while differences between occasional and no-use did not reach statistical significance (p = 0.052). Differences in standard deviation of lateral position were not statistically significant, likely due to the short duration of tasks. Conclusions: These results find the largest potential safety concerns associated with a pattern of occasional use, who displayed an increase in lane departures after acute cannabis smoking. Those in the daily use group decreased their speed, which may be interpreted as compensation for drug effects. Further research is needed to understand the effects during longer and more complex secondary tasks.


  • Impact of cannabis and low alcohol concentration on divided attention tasks during driving

    by Ryan E. Miller, Timothy L. Brown, Stella Lee, Ishaan Tibrewal, Gary G. Gaffney, Gary Milavetz, Rebecca L. Hartman, David A. Gorelick, Richard Compton, Marilyn A. Huestis

    To assess divided-attention performance when driving under the influence of cannabis with and without alcohol. Three divided-attention tasks were performed following administration of placebo, cann...


  • Marginal false discovery rate control for likelihood‐based penalized regression models

    by Ryan E. Miller, Patrick Breheny

    The popularity of penalized regression in high‐dimensional data analysis has led to a demand for new inferential tools for these models. False discovery rate control is widely used in high‐dimensional hypothesis testing, but has only recently been considered in the context of penalized regression. Almost all of this work, however, has focused on lasso‐penalized linear regression. In this paper, we derive a general method for controlling the marginal false discovery rate that can be applied to any penalized likelihood‐based model, such as logistic regression and Cox regression. Our approach is fast, flexible and can be used with a variety of penalty functions including lasso, elastic net, MCP, and MNet. We derive theoretical results under which the proposed method is valid, and use simulation studies to demonstrate that the approach is reasonably robust, albeit slightly conservative, when these assumptions are violated. Despite being conservative, we show that our method often offers more power to select causally important features than existing approaches. Finally, the practical utility of the method is demonstrated on gene expression datasets with binary and time‐to‐event outcomes.