This SBIR phase I project will develop a Speech Therapy Robot (STR) to assist in the administration of evidence-based speech and language therapy to provide individualized monitoring of multiple clients simultaneously in a school setting. STR will use biologically plausible artificial intelligence models to prototype a system that is affordable, easy to use, portable and extensible to work with any number of students (clients) with disorder. Most children make some mistakes as they learn to say new words, but a speech sound disorder results when mistakes continue past a certain age. Speech sound disorders include problems with articulation (making sounds) and phonological processes (sound patterns), and it is one of the largest disabilities in the United States. Children with speech disorders are evaluated by a speech-language pathologist (SLP) and treated via speech-language intervention within the child's classroom (classroom-based) or outside of the classroom (pull-out). Multiple studies have demonstrated that classroom-based service is beneficial over pull- out service, but currently it is not widely practiced because of the many challenges facing SLPs: 1) it requires collaboration with classroom teachers and administrators who are not trained in speech-language pathology, 2) it can create a larger client-SLP ratio, 3) a small client-nonclient student in-class ratio, 4) unable to provide adequate intervention, 5) there is a large variation in severity of disorder within clients, 6) longer session hours over pull-out service. The American Speech-Language-Hearing Association (ASHA) recommendations caseloads should not exceed 40, but the median caseload is 50 in elementary and secondary schools, with high of 80 clients. This heavy workload for an SLP limits their capacity to provide effective treatment. Thus a robotic system capable of reducing the workload and assisting SLPs to provide improved individual care is highly desired by those in this field. In this Phase I SBIR, we will develop novel biologically plausible models to address these challenges by developing a robot-assisted therapy system capable of real time monitoring and assessment of client and client-provider interaction during the session, to determine client engagement, performance and to give feedback to providers in real time for improved treatment delivery. The biological models attempt to mimic the expert diagnostic capabilities of a SLP and extend it for use by non-SLPs to work with multiple clients at the same time. The solution will not require specialized training to use, allowing teachers to easily use it in their classrooms. In Phase I, we will demonstrate the feasibility and accuracy of STR. STR is not just a minor improvement over existing technologies but a technology and application that do not exist today. In Phase II, we will extend the capabilities towards a fully biologically plausible system to mimic expert human performance levels to develop a robotic system for speech-language therapy, this will be followed by clinical trials to ensure accuracy, efficacy of STR to facilitate evidence-based therapy. Variations of the system can be used towards phonology, morphology/syntax, pragmatics, language, fluency and/or vocabulary.
Overall the project provides direct relevance to public health by facilitating new insights through the development of a novel biologically plausible artificial intelligence system capable of real time monitoring and assessment of verbal therapy session content in real time to determine patient engagement, performance and give feedback to providers in real time to improve treatment delivery, in a school setting. The novel biologically plausible device will significantly impact the current known methods of in classroom evaluation, monitoring and treatment of speech disorders. The project can help in substantial improvement in patient client interaction, better treatment, lower burden on speech language pathologists, and will significantly impact the current known methods, technologies, treatments, and address critical barriers to progress in the field.