Using Artificial Intelligence To Monitor Medication Adherence in Opioid Replacement Therapy

Overview

This study uses an artificial intelligence platform to automatically confirm medication ingestion. The Health Insurance Portability and Accountability Act (HIPAA)-compliant platform can be downloaded as an 'app' onto any smartphone to automate directly observed therapy (Automated DOT®). Real-time patient adherence data are encrypted and automatically sent to a centralized web-based dashboard for use by healthcare professionals or research staff. Unlike Facetime® or Skype®, the system relies on computer vision algorithms to confirm the process of medication administration; no human review is necessary. The purpose of this study is to evaluate the feasibility and acceptability, and measure the accuracy, of the AiCure platform ("platform") in patients being treated for opioid dependence with Zubsolv® over the course of 12 weeks. The following aims will be tested: 1) to assess the feasibility and acceptability to both participants and study staff in using AiCure to monitor medication adherence; 2) to evaluate the acceptability of using AiCure to optimize care pathways; and 3) to measure the reliability and validity of AiCure in detecting interruptions in treatment. To assess feasibility and acceptability of the platform, we will measure rates of physician satisfaction and user acceptance. Optimization of care pathways will be measured by assessing the sustainability of AiCure use over 12 weeks (retention rates) and measuring illicit opioid use (urine drug screens) compared to historical data. Reliability and validity of AiCure will be measured by comparing AiCure adherence against pharmacokinetic data. All participants will be requested to take each of their prescribed doses using the app. Participants will be able to download the app onto their own smartphone or will be provisioned a device at the start of the study. The data captured during the medication ingestion process will be automatically encrypted and stored on the participant smartphone and uploaded wirelessly to a cloud-based dashboard. If a participant is non-adherent (missed dose, incorrect dosage) or if suspicious behavior is detected, an automated alert will be sent to study staff via email or SMS to prompt immediate intervention. In addition, all participants will receive treatment as usual.

This study will employ a multi-site, single-arm design. A total of approximately 50-100 participants - patients stable for at least 2 weeks on their current opioid replacement medication - will be recruited for the study. All participants will receive their doctor's treatment-as-usual. Patients not currently prescribed Zubsolv® will be switched to Zubsolv®. Study visits include a screening visit, one baseline visit (which ideally will occur between 7 and 14 days after the screening visit), and bi-weekly visits for the 12 weeks (six visits) following the baseline visit. During the baseline visit, participants will be trained on how to use the AiCure app. Training consists of a number of interactive training steps to teach the participant how to use the app correctly. Participants will be provided with three placebo tablets for the training. Study participants will be reimbursed to cover their time and transportation costs in accordance with Institutional Review Board (IRB) guidelines. Participants will receive contingency management (CM) to reinforce regular use of the app. Orexo AB will provide the study drug, Zubsolv®, to all participants throughout the 12-week treatment duration. For the length of the study, participants will be requested to take each dose of their prescribed Zubsolv® regimen using the AiCure app. Each medication administration event will be saved onto the participant's smartphone and encrypted data (including de-identified video and time and date of administration) will be automatically transmitted to the centralized dashboard. Research staff will have access to the dashboard to view real-time and detailed dosing histories for each participant. Access to the dashboard is roles-based and password-protected. If a participant does not dose using the AiCure app (misses/skips a dose), self-reports on the device or over the phone, or is tagged for suspicious behavior, the participant will receive a combination of automated SMS text messages and tailored SMS text messages / phone calls from research staff based on the pre-defined escalation protocol.