Opioid Detection Using Smartphone-Based Eye-Scanning

Sensors, vol. 25, pp. 5467

Abstract

Opioids are known to constrict pupils, and mobile phone-based self-administered eye scanning (MPSES) offers a potential method for monitoring opioid use in real-world settings. A clinical trial with 12 volunteers measured pupil size using MPSES under different light conditions (approx. 50 or approx. 500 lux) in the lab and over a week at home. Each participant made approximately 21 home tests, 12 in the lab without oxycodone and 16 in the lab after oxycodone intake. At the second visit the participants received a single dose of 20 mg oxycodone, and their pupil size was monitored hourly for 5 h. The pupil size after oxycodone intake was compared to drug-naïve tests performed at the lab and at home. Logistic regression models were built using measured pupil size and light conditions measured by the phone during each test, and a dichotomous variable indicating tests before or after oxycodone dosing as the outcome. The model demonstrated high classification accuracy (AUC = 0.94), with 82% true positives, 9% false positives, 91% true negatives and 18% false negatives. Misclassifications were largely due to difficulties measuring pupil size in individuals with corneal arcus, causing most of the false positive findings, and other interindividual differences. This shows that MPSES, including monitoring of ambient light conditions, can effectively detect opioid use within the 50-500 lux range. Our study paves the way for using MPSES to detect opioid use.

Authors 5

1. Kiki W. K. Kuijpers corresponding Aachen Department of Anesthesiology, Leiden University Medical Center

   Leiden University Medical Center

   Affiliation as printed

   Department of Anesthesiology, Leiden University Medical Center, 2333 ZA Leiden, The Netherlands

2. Karl Andersson

   University of Skövde

   Affiliation as printed

   Digital Health Research, Skövde University, SE54128 Skövde, Sweden

   Skillsta Teknik Design och Kvalitet AB, Vänge, SE75578 Uppsala, Sweden

3. Albert Dahan Aachen Centre for Human Drug Research (CHDR)

   Outcomes Research Consortium · Centre for Human Drug Research

   Affiliation as printed

   Centre for Human Drug Research, 2333 CL Leiden, The Netherlands

   MediD Consultancy Group, 1077 XD Amsterdam, The Netherlands

   Outcomes Research Consortium, Houston, TX 77030, USA

4. Markku Hämäläinen
   Affiliation as printed

   Kontigo Care AB, SE75320 Uppsala, Sweden

5. Monique van Velzen corresponding Aachen Department of Anesthesiology, Leiden University Medical Center

   Leiden University Medical Center

   Affiliation as printed

   Department of Anesthesiology, Leiden University Medical Center, 2333 ZA Leiden, The Netherlands