AI-Based Automated Speech Therapy Tools for Persons with Speech Sound Disorders: A Systematic Review of the Literature

Основные понятия

This systematic literature review analyzes research on AI-based automated speech therapy tools for individuals with speech sound disorders, examining the types of disorders addressed, the level of autonomy achieved, the modes of intervention, and the effectiveness compared to conventional speech therapy.

Аннотация

This systematic literature review analyzed 24 research studies on AI-based automated speech therapy tools for individuals with speech sound disorders (SSD) published between 2007 and 2022. The key findings are:

1. Types of SSD Addressed:

   • The most frequently addressed SSD were articulation disorders, hearing impairment, dysarthria, and motor speech disorders.
   • However, 50% of the studies did not address any specific SSD, proposing generalized tools.

2. Level of Autonomy:

   • Most studies proposed fully automated AI-based speech therapy tools without considering the role of speech-language pathologists, caregivers, and other stakeholders.

3. Modes of Intervention:

   • The most common modes of intervention were mobile-based and computer-based applications, often incorporating serious games and augmented reality.

4. Effectiveness:

   • Only a few studies compared the effectiveness of their automated tools with conventional speech therapy provided by experts, indicating a need for more rigorous evaluation.

The review highlights the growing interest in AI-based automated speech therapy tools, but also identifies several areas for future research, such as developing tools for underrepresented languages, involving stakeholders in the design process, and conducting more comprehensive evaluations of the effectiveness of these tools compared to expert-led speech therapy.

Статистика

Articulation disorders were the most frequently addressed SSD, addressed in 3 studies.
Hearing impairment, dysarthria, and motor speech disorders were each addressed in 2 studies.
12 out of 24 studies did not address any specific SSD, proposing generalized tools.
79 out of 91 unique authors (86.81%) contributed to only one paper in the included studies.
11 studies were conducted in Europe, 6 in North America, and 5 in Asia.
The most prevalent language targeted was English (10 studies), followed by Spanish (4 studies).
Only 4 out of 24 studies compared the effectiveness of their automated tools with conventional speech therapy provided by experts.

Цитаты

"Almost all studies proposed fully automated AI-based speech therapy tools suggesting that researchers did not emphasize the role of caretaker, parents, family members, and SLPs."
"Very few studies have compared their proposed system's effectiveness with expert SLPs."

Ключевые выводы из

AI-Based Automated Speech Therapy Tools for persons with Speech Sound Disorders: A Systematic Literature Review

by Chinmoy Deka... в arxiv.org 04-22-2024

https://arxiv.org/pdf/2204.10325.pdf

Дополнительные вопросы

What strategies can be employed to better involve speech-language pathologists, caregivers, and other stakeholders in the design and development of AI-based automated speech therapy tools?

In order to better involve speech-language pathologists (SLPs), caregivers, and other stakeholders in the design and development of AI-based automated speech therapy tools, several strategies can be implemented:

Collaborative Design Workshops: Organize workshops where SLPs, caregivers, and developers can collaborate in the design process. This allows for the incorporation of expert knowledge from SLPs and insights from caregivers about the needs and challenges faced by individuals with speech sound disorders.

User-Centered Design Approach: Adopt a user-centered design approach that involves stakeholders in every stage of the development process. This ensures that the final product meets the needs and expectations of the end-users.

Feedback Loops: Establish feedback mechanisms to gather input from SLPs, caregivers, and individuals undergoing therapy. This feedback can be used to iterate on the design and improve the effectiveness of the automated tools.

Training and Education: Provide training and education sessions for SLPs and caregivers on how to effectively use the AI-based tools. This empowers them to integrate the tools into their practice and support individuals with speech sound disorders.

Pilot Testing: Conduct pilot testing with SLPs and caregivers to evaluate the usability, functionality, and impact of the automated tools in real-world settings. This feedback can inform further refinements and enhancements.

Ethical Considerations: Address ethical considerations such as data privacy, security, and the ethical use of AI in speech therapy. Involving stakeholders in these discussions ensures that the tools are developed and deployed responsibly.

By implementing these strategies, AI-based automated speech therapy tools can be designed in a way that aligns with the needs and preferences of SLPs, caregivers, and individuals with speech sound disorders.

How can the effectiveness of these automated tools be more rigorously evaluated to ensure they provide outcomes comparable or superior to conventional speech therapy?

To rigorously evaluate the effectiveness of AI-based automated speech therapy tools and ensure they provide outcomes comparable or superior to conventional speech therapy, the following approaches can be taken:

Randomized Controlled Trials (RCTs): Conduct RCTs comparing the outcomes of individuals receiving therapy through AI-based tools versus traditional speech therapy. This gold standard research design helps establish causal relationships and measure the effectiveness of the automated tools.

Longitudinal Studies: Implement longitudinal studies to track the progress of individuals over an extended period. This allows for the assessment of long-term outcomes and the sustainability of improvements achieved through automated therapy.

Outcome Measures: Define clear and standardized outcome measures to assess the impact of the automated tools on speech sound disorders. These measures can include improvements in articulation, phonological skills, and overall communication abilities.

Comparative Studies: Conduct comparative studies that directly compare the effectiveness of AI-based tools with conventional speech therapy in similar populations. This head-to-head comparison can provide valuable insights into the relative efficacy of the two approaches.

User Satisfaction Surveys: Administer user satisfaction surveys to individuals undergoing therapy, SLPs, and caregivers to gather feedback on their experiences with the automated tools. High satisfaction levels often correlate with positive outcomes.

Data Analysis: Utilize advanced data analysis techniques to analyze the effectiveness of the automated tools, such as machine learning algorithms to identify patterns and trends in therapy outcomes.

By employing these rigorous evaluation methods, researchers and developers can confidently assess the effectiveness of AI-based automated speech therapy tools and ensure that they deliver outcomes that are on par with or surpass traditional speech therapy.

What opportunities exist for developing AI-based automated speech therapy tools for underrepresented languages and deploying them in regions with limited access to speech-language pathology services?

Developing AI-based automated speech therapy tools for underrepresented languages and deploying them in regions with limited access to speech-language pathology services presents several opportunities:

Language Customization: AI technologies can be tailored to specific languages by training speech recognition models on datasets in those languages. This customization enables the development of automated tools for underrepresented languages.

Remote Access: AI-based tools can be deployed through telepractice, allowing individuals in remote or underserved areas to access speech therapy services without the need for in-person sessions with SLPs.

Mobile Applications: Develop mobile applications that deliver automated speech therapy in underrepresented languages. These apps can reach a wider audience and provide convenient access to therapy resources.

Community Partnerships: Collaborate with local communities, organizations, and healthcare providers to introduce AI-based speech therapy tools in regions with limited access to traditional services. This can help bridge the gap in speech therapy provision.

Low-Cost Solutions: Create cost-effective AI-based tools that are affordable and accessible to individuals in underserved areas. This can democratize access to speech therapy services and improve outcomes for those with speech sound disorders.

Training Programs: Establish training programs for local healthcare professionals and educators to use AI-based tools effectively in providing speech therapy. This builds capacity within the community to support individuals with speech disorders.

Research Collaborations: Foster research collaborations between academic institutions, technology developers, and local stakeholders to address the unique challenges of deploying AI-based tools in underrepresented languages and regions.

By leveraging these opportunities, AI-based automated speech therapy tools can be developed and deployed effectively in underrepresented languages and regions with limited access to speech-language pathology services, ultimately improving the quality of care for individuals with speech sound disorders.