1 - (SYM 25) Validity and User Acceptability of Alcohol Use Detection via Wrist-worn Transdermal Biosensor

Background: Alcohol use monitoring forms the backbone of addiction intervention. But the identification of objective measures for monitoring drinking has represented a major challenge for researchers. Transdermal sensors offer the possibility of continuous, objective, and unobtrusive assessment of drinking. While older-generation sensors employed bulky ankle-worn designs and sparse sampling schedules, new-generation sensors feature sleek designs and rapid sampling, well suited for use in healthcare and treatment contexts. The current study represents the first large-scale test of these novel devices, employing a combined laboratory-ambulatory design to examine transdermal sensor acceptability and drinking detection accuracy in real-time.

Methods: Participants consisted of 150 healthy drinkers recruited from the local community (ages 21-54; 49% White; 45% Female; 22% Hispanic). Participants attended three laboratory sessions involving the experimental manipulation of alcohol dose, rate of consumption, and environmental dosing conditions. Participants also wore transdermal sensors and provided breathalyzer readings in the field over 14 days. Transdermal sensor readings were converted into real-time estimates of drinking via machine learning. Participants' perceptions of sensor acceptability were assessed post-study.

Results: Acceptability was high, with 77% of participants indicating they would be willing to wear the device beyond the study endpoint. Itching and sleep interference emerged as the most widely reported issues with transdermal sensors. Regarding sensor validity, models indicated strong transdermal accuracy for real-time drinking detection across both laboratory and field contexts (AUROC, 0.966, 95% CI, 0.956-0.972; Sensitivity, 89.8%; Specificity, 90.6%). Models aimed at differentiating high-risk (=0.08%) drinking levels yielded intermediate (AUROC, 0.738; 95% CI, 0.698-0.777; only drinking episodes) to strong (AUROC, 0.941, 95% CI, 0.929-0.954; all data) accuracy levels.

Conclusions: Results indicate strong accuracy for new-generation transdermal sensors in the real-time detection of drinking across diverse contexts. Results further indicate generally strong user acceptability for these sensors. The accuracy of these sensors across longer time periods remains an important direction for future study. However, results offer promising initial evidence for useful transdermal sensor application across prevention, clinical trial, and just-in-time intervention contexts.