Abstract Patients affected by opioid use disorder (OUD) face numerous challenges, ranging from stigma to policy barriers that impede their access to care. Even after engaging OUD patients in treatment, retention rates are disappointingly low due to a multitude of factors. Despite the widespread adoption of virtual counseling and treatment solutions, such as telemedicine programs using medications for opioid use disorder (MOUD), during the COVID-19 pandemic, addiction treatment still requires in-person visits for regular drug testing. However, this requirement can lead to non-compliance and being dropped from what is often their only therapeutic option available. Innovative and FDA-cleared opioid use testing methods that can be performed outside of laboratory environments address these issues and hold the promise of increasing retention rates and improving clinical outcomes. Tenvos’ product will be integrated into a smartphone app, enabling patients to test for opioid use from the comfort of their home and send instant results to the clinician without the need for an in-person visit. The proposed solution will not only improve clinical outcomes in the outpatient OUD treatment setting but can be adopted in many other settings due to its convenience and can, potentially, unlock new workflows that were not feasible before. The FDA-regulated SaMD itself, which is a ML-based solution, will be exposed through an API enabling its integration into a large number of systems and applications. The API can be integrated into virtual health solutions and provide opioid use assessment real-time during an appointment. It can be integrated into a wearable device and serve as a remote patient monitoring device that notifies the treating physician in case of a relapse. It can also be used for population health and risk stratification by integrating into case management calls at managed care organizations to identify individuals who may benefit from OUD treatment. The specific aims for the Phase I set of activities include: Aim 1: Collecting clinical data with gold standard test results and patient voice samples to study the correlation between them. Aim 2: Validating and fine-tuning the machine learning models with the collected clinical data. Aim 3: Engaging with the FDA Center of Devices and Radiological Health (CDRH) through a pre-submission process to obtain feedback on the most appropriate pathway, such as PMA, De Novo, or 510(k).