A single drop of untreated urine can be screened for illicit drugs in about five minutes, according to research described November 18 by South Korea’s National Research Council of Science and Technology. The KRIBB-led team combines a nanostructured gold surface with AI, reporting approximately 80% accuracy in identifying drug-positive samples in tests using real urine.
The real-urine experiments included samples from patients taking non-narcotic painkillers or multiple medications. Those substances create a complicated background for identifying illicit drugs. The team used the LightGBM machine-learning model to distinguish drug-specific signal fingerprints from metabolic waste and medication residues.
The institutional release does not break the accuracy figure into false positives and false negatives. That leaves an important distinction unresolved: how often the platform incorrectly flags a sample versus missing a drug-positive one.
Separate experiments used rats. In those tests, the researchers demonstrated that drug components could be analyzed in urine collected within six hours after administration. That timing finding concerns the animal experiments, not the real-urine accuracy result.
Gold strengthens the signal; AI identifies the pattern
The platform starts with a thin gold surface covered in densely packed, grass-like pillars at the nanoscale. The team roughened those pillars through selective etching. When drug molecules settle on the surface, their characteristic vibrational light signals become stronger. The researchers report amplification by millions of times or more.
LightGBM then handles recognition, separating the drug fingerprints from other signals in the sample. This combination allows analysis without chemical pretreatment. The approach targets a practical obstacle in laboratory testing, which requires several preparation steps and specialized equipment, according to the research release.
Eun-Kyung Lim at the Korea Research Institute of Bioscience and Biotechnology led the work. The study, “Interpretable Screening of Illicit Drugs in Raw Urine via LightGBM-Assisted SERS,” appeared in September 2026 in ACS Measurement Science Au, before this week’s institutional release.
By further developing the platform into a portable diagnostic device for on-site use, we hope to transform drug screening and emergency diagnostics.
Eun-Kyung Lim, study lead at KRIBB
The team identifies two potential uses for a compact measurement device: on-site screening by law enforcement and rapid assessment of drug intoxication in hospital emergency departments. Both require further development. Portability remains a goal for the research, not an available product.
Reader comments
Newest comments first. Replies stay oldest first.