PROJECT SUMMARY Current treatment options for addiction do not sufficiently address the clinical needs of patients. Due to low efficacy, misaligned treatment objectives, or lack of accessibility, patients are not able to durably abstain from substances or reduce their use to safer levels. Substitution therapy (ST) is a proven treatment approach and has resulted in massive global public health benefits for tobacco use disorder and opioid use disorder; however, the potential impact of ST is limited by concerns for causing more widespread harms. Thus substitution therapy is either underutilized or underexplored in other addictions due to risks of abuse, misuse, and diversion. Through implementing safe prescribing systems, health systems have solved for the provision of existing ST like methadone and buprenorphine. Nevertheless, there remains significant potential for broader access to current and novel ST treatments by utilizing advancement in software and hardware technology. Atman Therapeutics is an early-stage company seeking to develop novel therapeutics in addictive disorders. We are developing a platform to enable increased access to as well as novel development of ST. To achieve this objective, our platform uses cutting-edge machine learning methods, patient-centered research, and sophisticated hardware technology. Our technology will also enable the concurrent delivery of contingency management services, thus creating safe, engaging, and durably effective treatments for these complex neurobehavioral disorders. This Small Business Innovation and Research (SBIR) project aims to establish the feasibility of machine learning and integrated sensor technologies to confirm medication adherence in a diverse sample of patients. We will develop a deep-learning algorithm utilizing a feed-forward convolutional neural network (CNN), such as ResNet- 50, U-NET, PCU-Net, to enact classification of medication ingestion events. We will also demonstrate the feasibility of integrating a hardware microcontroller unit with multi-dimensional MEMS sensors into a device that can classify ingestion events in a highly sensitive and specific manner. These efforts will demonstrate the early feasibility of the Atman platform and form the basis for regulated development efforts of ST-enabling products and services.