Machine Learning for Assistive Technology

According to the World Health Organization (WHO), in 2022 more than 1.3 billion people – or 1 in 6 people worldwide – experienced significant disabilities. Various assistive technologies (ATs) have been created to help individuals with physical, sensory, or cognitive impairments to live and work more effectively and independently in all aspects of their lives. Intelligent ATs, which use artificial intelligence and machine learning techniques, have been incorporated into smart homes, resulting in a profusion of systems, chatbots, augmentative communication devices, smart wheelchairs, and brain–computer interfaces. However, existing ATs present many challenges that remain to be addressed, including limited capabilities, accessibility, and/or accuracy. Even more seriously, ATs often lack the customization required to adapt to changes in the situation or the abilities of the individual.

Machine learning (ML) is arguably the most prominent branch of artificial intelligence (AI), covering many exciting research and industrial innovations that provide more efficient, effective, and automated algorithms to deal with large-scale data in a wide variety of disciplines (e.g., computer vision, neuroscience, speech recognition, language processing, human–computer interaction, health informatics, medical image analysis, recommender systems, fraud detection, etc.). Advances in ML provide opportunities to transform the landscape of existing ATs. For example, ML-powered gesture-based prediction provides better speed and precision in analyzing and deciphering complex communication, expressions, and visual behaviors. ML can also be utilized to adapt to changes in an individual’s situation and abilities by integrating health data. Therefore, ML/AI will likely be a key component in future intelligent AT applications.

Nevertheless, there are still challenges and barriers in designing and using ML-powered ATs for various impairments. For example: (1) How to collect large and diverse datasets of people with disabilities (PWDs) to train ML models, while respecting the privacy issues and reflecting the uniqueness of such individuals? (2) How to make ML-powered ATs more accurate and adaptable for children with multiple developmental impairments, to match their developing abilities? (3) How to design ML-powered ATs that continuously assist individuals with changing security and privacy concerns? (4) Regarding usability, how to develop ATs that have minimum requirements from users?

This Research Topic welcomes original research papers that examine these questions. Topics of interest include, but are not limited to:

- Novel ML-powered applications of ATs
- Novel ML algorithms for ATs
- Datasets for ML-based ATs
- Human subject studies to identify specific requirements for ATs
- Security and usability of ATs
- Privacy and ethical issues of ATs
- Design and prototypes of context-aware, personalized ATs
- Intelligent personal digital assistants
- Robot assistants
- Human–Computer Interfaces (HCI) and Brain–Computer Interfaces (BCI) for ML-powered assistive systems
- Multimodal ML-powered assistive systems
- Design principles and methods for ATs
- Off-the-shelf components and open-source tools for ATs
- QoS evaluation metrics for assistive systems
- Wearable assistive devices
- Serious games for ATs
- Ambient assisted living, and active and assisted living technologies
- Egocentric view ATs
- Natural language processing (NLP) for ATs
- Biometrics technology for ATs

Keywords: machine learning, assistive technologies, privacy and security, usability, accessibility

Important Note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.