Artificial Intelligence (AI) is having a tremendous impact across most areas of science. Applications of AI in healthcare have the potential to improve our ability to detect, diagnose, prognose, and intervene on human disease. For AI models to be used clinically, they need to be made safe, reproducible and robust, and the underlying software framework must be aware of the particularities (e.g. geometry, physiology, physics) of medical data being processed. This work introduces MONAI, a freely available, community-supported, and consortium-led PyTorch-based framework for deep learning in healthcare. MONAI extends PyTorch to support medical data, with a particular focus on imaging, and provide purpose-specific AI model architectures, transformations and utilities that streamline the development and deployment of medical AI models. MONAI follows best practices for software-development, providing an easy-to-use, robust, welldocumented, and well-tested software framework. MONAI preserves the simple, additive, and compositional approach of its underlying PyTorch libraries. MONAI is being used by and receiving contributions from research, clinical and industrial teams from around the world, who are pursuing applications spanning nearly every aspect of healthcare.

At least 360 million people worldwide have disabling hearing loss that frequently causes difficulties in day-to-day conversations. Traditional technology (e.g., hearing aids) often fails to offer enough value, has low adoption rates, and can result in social stigma. Speechreading can dramatically improve conversational understanding, but speechreading is a skill that can be challenging to learn. To address this, we developed a novel speechreading acquisition framework that can be used to design Speechreading Acquisition Tools (SATs) -a new type of technology to improve speechreading acquisition. We interviewed seven speechreading tutors and used thematic analysis to identify and organise the key elements of our framework. We then evaluated our framework by using it to: 1) categorise every tutor-identified speechreading teaching technique, 2) critically evaluate existing conversational aids, and 3) design three new SATs. Through the use of SATs designed using our framework, the speechreading abilities of people with hearing loss around the world should be enhanced, thereby improving the conversational foundation of their day-to-day lives.

Subtitles can help improve the understanding of media content. People enable subtitles based on individual characteristics (e.g., language or hearing ability), viewing environment, or media context (e.g., drama, quiz show). However, some people fnd that subtitles can be distracting and that they negatively impact their viewing experience. We explore the challenges and opportunities surrounding interaction with real-time personalisation of subtitled content. To understand how people currently interact with subtitles, we frst conducted an online questionnaire with 102 participants. We used our fndings to elicit requirements for a new approach called Adaptive Subtitles that allows the viewer to alter which speakers have subtitles displayed in real-time. We evaluated our approach with 19 participants to understand the interaction trade-ofs and challenges within real-time adaptations of subtitled media. Our evaluation fndings suggest that granular controls and structured onboarding allow viewers to make informed trade-ofs when adapting media content, leading to improved viewing experiences. CCS CONCEPTS• Human-centered computing → Interaction paradigms.

Social media platforms are deeply ingrained in society, and they offer many different spaces for people to engage with others. Unfortunately, accessibility barriers prevent people with disabilities from fully participating in these spaces. Social media users commonly post inaccessible media, including videos without captions (which are important for people who are Deaf or Hard of Hearing) and images without alternative text (descriptions read aloud by screen readers for people who are blind). Users with motor impairments must find workarounds to deal with the complex user interfaces of these platforms, and users with cognitive disabilities may face barriers to composing and sharing information. We invited accessibility researchers, industry practitioners, and end-users with disabilities to come together at the Computer-Supported Cooperative Work conference (CSCW 2019) to discuss challenges and solutions for improving social media accessibility. Over the course of a day that included two panels and breakout sessions, the workshop attendees outlined four critical future research directions to progress on the path to accessible social media: tooling to support disabled people authoring content, developing more accessible formats/tools for new forms of interaction (e.g, Augmented and Mixed Reality), using communities to distribute accessibility labor, and ensuring machine learning systems are built on representative datasets for disability use-cases.

Situationally-induced impairments and disabilities (SIIDs) make it difficult for users of interactive computing systems to perform tasks due to context (e.g., listening to a phone call when in a noisy crowd) rather than a result of a congenital or acquired impairment (e.g., hearing damage). SIIDs are a great concern when considering the ubiquitousness of technology in a wide range of contexts. Considering our daily reliance on technology, and mobile technology in particular, it is increasingly important that we fully understand and model how SIIDs occur. Similarly, we must identify appropriate methods for sensing and adapting technology to reduce the effects of SIIDs. In this workshop, we will bring together researchers working on understanding, sensing, modelling, and adapting technologies to ameliorate the effects of SIIDs. This workshop will provide a venue to identify existing research gaps, new directions for future research, and opportunities for future collaboration. Submission instructionsWorkshop papers are limited to a MAXIMUM of 8 pages (incl. references

Social media platforms are deeply ingrained in society, and they offer many different spaces for people to engage with others. Unfortunately, accessibility barriers prevent people with disabilities from fully participating in these spaces. Social media users commonly post inaccessible media, including videos without captions (which are important for people who are deaf or hard of hearing) and images without alternative text (descriptions read aloud by screen readers for people who are blind). Users with motor Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for third-party components of this work must be honored. For all other uses, contact the owner/author(s).

Emoji are graphical symbols that appear in many aspects of our lives. Worldwide, around 36 million people are blind and 217 million have a moderate to severe visual impairment. This portion of the population may use and encounter emoji, yet it is unclear what accessibility challenges emoji introduce. We first conducted an online survey with 58 visually impaired participants to understand how they use and encounter emoji online, and the challenges they experience. We then conducted 11 interviews with screen reader users to understand more about the challenges reported in our survey findings. Our interview findings demonstrate that technology is both an enabler and a barrier, emoji descriptors can hinder communication, and therefore the use of emoji impacts social interaction. Using our findings from both studies, we propose best practice when using emoji and recommendations to improve the future accessibility of emoji for visually impaired people.

Although our sense of hearing, smell, and vision allow us to perceive things at a distance, the detection of many day-to day events relies exclusively on our hearing. For example, finding a ringing phone lost in a sofa, hearing a child cry in another room, and use of a car alarm to locate a vehicle in a car park. However, individuals with total or partial hearing loss have difficulty detecting the audible signals in these situations. We have developed VisAural, a system that converts audible signals into visual cues. Using an array of head-mounted microphones, VisAural detects the direction of a sound, and places LEDs at the periphery of the user's visual field to guide them to the source of the sound. We tested VisAural with nine people with hearing impairments and found that this approach holds great promise but needs to be made more responsive before it can be truly helpful.