In-Vehicle Real-Time Cannabis Influenced Driving Detection
Trial statusRecruiting
Trial phaseNot applicable
Trial typeInterventional
Biological sexAll
Age21+
SponsorUniversity of Bern
About this trial
The goal of this clinical trial is to evaluate whether in-vehicle sensor data can be used to detect cannabis-impaired driving in healthy adult recreational cannabis users.
The study aims to assess whether changes in vehicle, driver, and physiological sensor data can distinguish sober driving from cannabis-impaired driving, and how driving performance changes from baseline to approximately 1 to 6 hours after controlled cannabis consumption.
Researchers will compare driving behavior and in-vehicle sensor data from participants who receive controlled cannabis administration with data from a randomized reference group without cannabis exposure, to determine whether cannabis-related impairment driving can be identified on the basis of machine learning.
Participants will complete screening and baseline assessments and drive an instrumented vehicle on a closed test track under sober conditions. Participants assigned to the experimental arm will receive controlled cannabis administration, while participants in the reference arm will receive no intervention. All participants will perform repeated standardized driving sessions over several hours and complete traffic-medical, traffic-psychological, and in-vehicle pre-driving tests. Biological samples and in-vehicle sensor data will be collected throughout the study.
Eligibility criteria
This trial accepts healthy volunteersQualifiers
Informed consent as documented by signature
Recreational cannabis-consumption (more than once per month)
In possession of a definite Swiss or European Union (EU) driving license
At least 21 years old
Disqualifiers
Health concerns where cannabis consumption is contra-indicated (such as: high blood pressure, psychiatric problems (e.g. psychosis, depression, attention-deficit conditions, etc.)
Cannabis-abstinence or excessive consumption, , assessed using the Cannabis Use Disorders Identification Test-Revised (CUDIT-R)
For women: pregnancy or breastfeeding or if intention to become pregnant during study period (time between telephone screening and study day (visit 2)
Alcohol misuse or excessive alcohol consumption habits/risky drinking behaviour, assessed using the Alcohol Use Disorders Identification Test (AUDIT) and/or phosphatidylethanol (PEth) in capillary blood > 200 ng/mL at first visit
Trial design
Design model
Parallel
Treatments tested in this trial
Cannabis (THC)
DrugParticipants assigned to the experimental arm receive a single, controlled inhalative administration of cannabis by smoking a THC-containing joint (target dose 0.67 mg THC per kg body weight; cannabis flowers with 15-18% THC).
Treatment groups
45 Participants
are divided into 2 treatment groupsGroup A: Controlled Cannabis AdministrationExperimental treatment 1 intervention
Group B: Reference (No Cannabis)No intervention 0 interventions
Trial outcomes
Primary outcomes
1
Diagnostic accuracy (AUROC) of a multimodal machine-learning model for detection of cannabis-impaired driving
Area under the receiver operating characteristic curve (AUROC) of a single machine-learning classifier that integrates multimodal in-vehicle data (including vehicle controller area network \[CAN\] data, driver monitoring camera \[DMC\] features, and physiological signals). All modalities are combined into one predictive model, and performance is reported as one aggregated AUROC value distinguishing non-impaired (sober) driving from cannabis-impaired driving.
Time frame
Baseline (sober driving) and up to 6 hours after cannabis administration in the experimental arm, with matched time points in the reference arm.Secondary outcomes
1
Diagnostic accuracy (AUROC) using CAN data
AUROC for detecting cannabis-impaired driving using vehicle controller area network (CAN) data only.
Time frame
Baseline (sober driving) and approximately 1-2, 3-4, and 5-6 hours after cannabis administration in the experimental arm, with matched time points in the reference arm.2
Diagnostic accuracy (AUROC) using DMC data
AUROC for detecting cannabis-impaired driving using driver monitoring camera (DMC) data only.
Time frame
Baseline (sober driving) and approximately 1-2, 3-4, and 5-6 hours after cannabis administration in the experimental arm, with matched time points in the reference arm.3
Diagnostic accuracy (AUROC) of a physiology-based machine-learning model for detection of cannabis-impaired driving
AUROC of a single machine-learning classifier trained and evaluated using aggregated physiological features only, derived from signals including heart rate, heart-rate variability, oxygen saturation, electrodermal activity, skin temperature, and respiration-related measures. All physiological signals are combined into one predictive model, and performance is reported as a single AUROC value distinguishing non-impaired (sober) driving from cannabis-impaired driving.
Time frame
Baseline (sober driving) and approximately 1-2, 3-4, and 5-6 hours after cannabis administration in the experimental arm, with matched time points in the reference arm.4
Change in driving behavior derived from vehicle CAN data
Change in driving behavior between sober driving and post-cannabis driving, derived from vehicle controller area network (CAN) signals, including steering, braking, acceleration, and velocity-related measures.
Time frame
Baseline (sober driving) and approximately 1-2, 3-4, and 5-6 hours after cannabis administration in the experimental arm, with matched time points in the reference arm.Sponsors and contacts
Click on the lead sponsor to view all of their trials.
Source ClinicalTrials.gov