Using machine learning for real-time BAC estimation from a new-generation transdermal biosensor in the laboratory

• Published in: Drug and alcohol dependence 2020-11, Vol.216, p.108205-108205, Article 108205
• Main Authors: Fairbairn, Catharine E., Kang, Dahyeon, Bosch, Nigel
• Format: Article
• Language: English
• Subjects: Addictions; Adult; Alcohol; Alcohol Drinking - blood; Alcohol use; Alcohols; Algorithms; Biosensing Techniques - instrumentation; Biosensing Techniques - methods; Biosensor; Biosensors; Blood alcohol concentration; Breath tests; Breath Tests - methods; Dosage; Drinking; Drinking behavior; Ethanol - analysis; Failure rates; Family Characteristics; Female; Humans; Laboratories; Laboratory research; Learning algorithms; Machine Learning; Male; Model accuracy; Prototypes; Real time; Sampling; Sensors; Smartphone; Technology; Technology assessment; Time series; Transdermal; Trees
• ISSN: 0376-8716; 1879-0046
• DOI: 10.1016/j.drugalcdep.2020.108205

•We examined a new-generation, smartphone-integrated alcohol biosensor in the lab.•Machine learning was used to convert biosensor data into estimates of BAC.•The new sensor demonstrated strong capabilities for detecting episodes of drinking.•BAC estimates for the new sensor were more accurate than those from an older device. Transdermal biosensors offer a noninvasive, low-cost technology for the assessment of alcohol consumption with broad potential applications in addiction science. Older-generation transdermal devices feature bulky designs and sparse sampling intervals, limiting potential applications for transdermal technology. Recently a new-generation of transdermal device has become available, featuring smartphone connectivity, compact designs, and rapid sampling. Here we present initial laboratory research examining the validity of a new-generation transdermal sensor prototype. Participants were young drinkers administered alcohol (target BAC = .08 %) or no-alcohol in the laboratory. Participants wore transdermal sensors while providing repeated breathalyzer (BrAC) readings. We assessed the association between BrAC (measured BrAC for a specific time point) and eBrAC (BrAC estimated based only on transdermal readings collected in the immediately preceding time interval). Extra-Trees machine learning algorithms, incorporating transdermal time series features as predictors, were used to create eBrAC. Failure rates for the new-generation prototype sensor were high (16 %–34 %). Among participants with useable new-generation sensor data, models demonstrated strong capabilities for separating drinking from non-drinking episodes, and significant (moderate) ability to differentiate BrAC levels within intoxicated participants. Differences between eBrAC and BrAC were 60 % higher for models based on data from old-generation vs new-generation devices. Model comparisons indicated that both time series analysis and machine learning contributed significantly to final model accuracy. Results provide favorable preliminary evidence for the accuracy of real-time BAC estimates from a new-generation sensor. Future research featuring variable alcohol doses and real-world contexts will be required to further validate these devices.