A portable brain imaging device (fNIRS) detected THC impairment with 90% accuracy and only 5% false positives — dramatically outperforming field sobriety tests which had 34% false positive rates.
Law enforcement, traffic safety policymakers, cannabis users concerned about driving, and impairment detection researchers
What the researchers found
Resting-state fNIRS achieved ROC-AUC=0.87, accuracy=0.90, and false-positive rate=0.05 for THC impairment detection vs. FST ROC-AUC=0.75, accuracy=0.69, and false-positive rate=0.34 — all differences statistically significant (p<.005).
Why it matters
Current THC impairment detection relies on biased subjective tests — a portable, objective brain imaging approach could revolutionize cannabis impairment testing for law enforcement and workplace safety.
The numbers in context
183 participants; fNIRS: AUC=0.87, accuracy=0.90, FPR=0.05; FST: AUC=0.75, accuracy=0.69, FPR=0.34; precision difference=0.23 (p<.001); accuracy difference=0.15 (p<.001); FPR difference=-0.25 (p<.001).
How the study worked
Double-blind, randomized, crossover trial of 183 cannabis users receiving oral synthetic THC (5-80mg) or placebo, with fNIRS brain scans and field sobriety tests at baseline, 100min, and 200min post-dose, analyzed with machine learning classifiers.
What this study cannot tell us
Laboratory setting with synthetic THC pills differs from real-world cannabis smoking; single-site study; fNIRS hardware still needs miniaturization for field use; trained on regular cannabis users.
How to read the evidence
Rigorous double-blind randomized crossover design published in JAMA Network Open with large sample and machine learning validation, though laboratory setting limits real-world generalizability.
When this study was published
Published in 2026 in JAMA Network Open, representing a major advance in objective cannabis impairment detection technology.
The bigger picture
The 34% false positive rate of field sobriety tests means 1 in 3 unimpaired people could be wrongly identified as impaired — fNIRS reduces this to 1 in 20, with enormous implications for criminal justice fairness.
Questions still open
- Can fNIRS devices be made portable enough for roadside use? Will the classifier work for occasional users or different consumption methods? How will courts handle neural impairment evidence?
Common questions
Can a brain scanner tell if someone is impaired by cannabis?
Why are field sobriety tests bad for detecting cannabis impairment?
Read the original research
Detection of Δ9-Tetrahydrocannabinol Impairment Using Resting-State Functional Near-Infrared Spectroscopy: A Randomized Clinical Trial.
JAMA network open, 9(1), e2556647
Citation
Berchansky, Moshe; Evins, A Eden; Evohr, Bryn; Himmelsbach, Zachary; Pachas, Gladys N; Karunakaran, Keerthana Deepti; Laufer Goldshtein, Bracha; Ozana, Nisan; Gilman, Jodi M. (2026). Detection of Δ9-Tetrahydrocannabinol Impairment Using Resting-State Functional Near-Infrared Spectroscopy: A Randomized Clinical Trial.. JAMA network open, 9(1), e2556647. https://doi.org/10.1001/jamanetworkopen.2025.56647
Explore the wider topic
- How Long Does THC Stay in Your System?
- How to Pass a Drug Test for Weed: What Actually Works
- Military, DOT, and Security Clearance Drug Testing for Cannabis
- Quitting Weed for a Drug Test: Timeline and Tips
- Cannabis DUI Laws: How Impairment Is Tested and What the Penalties Are
- Can People Tell I'm High? Signs of Cannabis Use Others Actually Notice
- Does THC Affect Driving? Impairment Science Beyond the Legal Question