#72 Notable Clinical Interest
Emerging findings or policy developments worth monitoring closely.
Clinicians assessing patients’ substance use and impairment risk can now leverage validated predictive factors identified through machine learning to better identify individuals at high risk for driving under the influence of alcohol or cannabis. This evidence-based approach enables more targeted screening and counseling interventions during clinical encounters, potentially reducing motor vehicle accidents and injuries in the patient population. Understanding these key predictors helps clinicians provide personalized risk assessment and prevention strategies rather than relying on generic substance use warnings.
Researchers employed machine learning algorithms to identify predictive factors for driving under the influence of alcohol or cannabis, findings published in Alcohol: Clinical & Experimental Research. By analyzing patterns in behavioral and demographic data, the study determined which variables most strongly correlate with impaired driving risk in users of these substances. These computational models could help clinicians identify high-risk patients who may benefit from counseling about driving safety and substance use consequences. The work also has implications for public health screening and may inform substance use disorder assessment protocols in clinical settings. Understanding which patient characteristics predict dangerous driving behavior allows clinicians to have more targeted conversations about the real-world harms of cannabis and alcohol use, particularly among individuals at elevated risk.
“What this machine learning work shows us is that we can now identify which patients are genuinely at risk for impaired driving after cannabis use, rather than relying on outdated assumptions that all cannabis users drive unsafely. In my practice, this means I can have more precise conversations with patients about their individual risk factors and help them make informed decisions about timing and dosing relative to driving, which is ultimately what evidence-based cannabis medicine requires.”
🚗 While machine learning approaches offer promise in identifying patterns associated with impaired driving risk, clinicians should recognize that predictive models trained on retrospective data may not capture the full complexity of acute cannabis impairment or individual variability in tolerance and metabolism. The generalizability of these models depends heavily on the populations and driving contexts in which they were developed, and important confounders such as polydrug use, sleep deprivation, and driving experience may not be adequately weighted in algorithmic predictions. From a clinical standpoint, these findings underscore the need for brief, evidence-based screening conversations with patients about cannabis use and driving—particularly given the expanding legalization landscape—rather than relying on predictive algorithms to identify at-risk individuals. Clinicians should counsel patients that unlike alcohol, there is no validated roadside or point-of-care test for cannabis impairment, making honest patient education about subjective effects and driving safety a practical
💬 Join the Conversation
Have a question about how this applies to your situation? Ask Dr. Caplan →
Want to discuss this topic with other patients and caregivers? Join the forum discussion →
Have thoughts on this? Share it: