Millions of Americans receive evidence-based counseling for substance use problems each year. Many evidence-based treatments for substance abuse are ?talk based? therapies, such as motivational interviewing (MI), but the existing research-based methodology for evaluating counseling quality is to record sessions and use human rating teams to evaluate them. However, using humans as the assessment tool via behavioral coding is prohibitive in cost and time, can be error prone, and is virtually never used in the real world. Technology is needed that can analyze the speech patterns and spoken language of counseling sessions, provide automatic and intuitive quality scores, and summarize these in actionable feedback. Rapid, performance-based quality metrics could support training, ongoing supervision, and quality assurance for millions of evidence-based counseling sessions for substance abuse each year. Lyssn.io is a start-up targeting the development of implementation-focused technology to support evidence-based counseling. Our goal is to develop innovative health technology solutions that are objective, scalable, and cost efficient. Lyssn.io includes expertise in speech signal processing, machine learning, user-centered design, software engineering, and clinical expertise in evidence-based counseling. Previous NIH-funded research laid a computational foundation for generating MI quality metrics from speech and language features in MI sessions, and led to a prototype of a clinical software support tool, the Counselor Observer Ratings Expert for MI (CORE-MI). The current Fast-Track SBIR proposal includes Phase I, which will focus on understanding clinical workflows, assessing usability, and initial validation of machine learning of MI fidelity measures in the opioid treatment program at Evergreen Treatment Services (ETS) clinic in Seattle, WA. Phase II will focus on robust validation of the speech and language technologies underlying the CORE-MI tool, and development of scalable supervision protocols that integrate CORE-MI supported feedback for counselors. Finally, we will conduct a quasi-experimental evaluation of CORE-MI supported supervision and training at a second ETS clinic in the Puget Sound, focusing on acceptability, usability, and adoption, the impact on supervision, improved MI fidelity and preliminary evidence of increased client retention. The successful execution of this project will break the reliance on human judgment for providing performance-based feedback to MI and will massively expand the capacity to train, supervise, and provide quality assurance in MI for substance abuse.
Most evidence-based treatments for substance abuse are in-person psychotherapy and counseling interventions, such as motivational interviewing. There are currently no methods for evaluating the quality of such counseling interventions in the real world to support training, supervision, and quality assurance. Building on an existing prototype, Lyssn.io ? a technology start-up focused on scalable and cost-efficient human-centered technologies ? will enhance and evaluate a cloud-based, HIPAA-compliant clinical support software tool that uses automated speech recognition and machine learning in an community based opioid replacement clinic.
