Alcohol Digital Twins: Can BAC, EtG, EtS and Urine Alcohol Reconstruct Drinking Time?

A 2026 study shows why reconstructing drinking time from a single BAC can be difficult—and how multiple alcohol markers may help. Podéus and colleagues developed a physiological “digital twin” that jointly models blood and breath ethanol, urine alcohol concentration (UAC), and the blood metabolites ethyl glucuronide (EtG) and ethyl sulfate (EtS). The model may help test whether a proposed drinking history is compatible with measured samples. It is not a method for identifying one indisputable time of consumption.

What is a physiological digital twin?

In this context, a digital twin is a mathematical representation of how ethanol moves through a person’s body and is transformed into metabolites. The model contains stomach, intestinal, plasma, tissue, liver and bladder components. It accounts for drinking conditions, food and physiological characteristics, then predicts concentration-time profiles for multiple markers.

BAC and BrAC primarily describe recent ethanol exposure; EtG and EtS are ethanol metabolites with different appearance and elimination patterns; UAC follows urine formation and bladder dynamics. These are related signals, not interchangeable clocks. The investigators also situate phosphatidylethanol (PEth) within the broader modelling framework, but the main forensic examples emphasize BAC, UAC, EtG and EtS.

How was the model evaluated?

The authors integrated experimental observations from ten prior studies, including sequential drinking protocols involving different beverages and doses. They compared model predictions with an independent validation dataset and examined new sequential-drinking data from two individuals. They also tested an alternative hypothetical drinking history using computer simulation.

The useful insight is not that every marker yields a unique answer. In the simulated comparison, some portions of a BAC curve looked consistent with two different drinking histories, while EtG, EtS or urine alcohol predictions differed during particular intervals. Multiple biological trajectories can therefore place additional constraints on a proposed chronology.

Could the model evaluate a claimed post-incident drink?

Potentially. The authors discuss the so-called “hip-flask” scenario, in which alcohol is claimed to have been consumed after an incident rather than before. In principle, a model can compare the biomarker patterns expected under competing drinking scenarios. But model output is conditional on reliable sample collection times, analytical measurements, drinking-history assumptions and suitable individual parameters.

The paper’s forensic case study used an in-silico alternative drinking scenario, not a prospectively validated method for deciding contested legal cases. A plausible model fit should not be represented as proof of when an individual drank.

What were the limitations?

The authors reported discrepancies in some peak concentrations, predicted faster BAC clearance than was observed in part of the independent validation, and noted that EtG or EtS elimination could be misestimated in particular phases. Some underlying studies did not fully record drink timing, food or nonalcoholic beverages. Incomplete or inaccurate inputs widen the plausible range of outcomes.

Forensic use should preserve the distinction between observed measurements, model-based estimates and disputed historical assertions. Sensitivity analyses, uncertainty ranges, assay validation and competing scenarios are essential if modelling is used to support testimony.

What records matter in an actual case?

Review the original BAC/BrAC and metabolite reports, specimen matrices and collection times, chain of custody, calibration and quality-control records, claimed consumption times and amounts, food and fluid intake, and the model version with its parameter settings. A quantitative answer without a reproducible input record is difficult to audit. For background on traditional calculations, see our guide to retrograde BAC extrapolation; for marker-specific interpretation, see EtG, EtS and PEth in forensic toxicology.

Frequently asked questions

Can EtG or EtS alone identify the exact time someone consumed alcohol?

No. These metabolites provide time-dependent information, but no single concentration is a unique timestamp. Their meaning depends on matrix, sampling time, assay characteristics and the assumed drinking pattern.

Does a digital twin eliminate uncertainty in retrograde BAC calculations?

No. It makes assumptions and uncertainty more explicit, potentially using more of the available evidence. Missing or disputed information can still produce multiple plausible explanations.

Was the 2026 model validated as a courtroom proof of drinking time?

No. The study reported research validation and illustrative reconstructions. It did not establish universal forensic error rates for contested cases.

Primary research

Podéus H, Simonsson C, Jakobsson G, et al. A digital twin framework for forensic reconstruction of alcohol intake via fast and slow metabolite kinetics. Scientific Reports. 2026;16:9336. doi:10.1038/s41598-026-44093-4.