PlainHealth: A Multimodal Service Suite for Cognitively Accessible Healthcare Communication
Abstract
Health information is often difficult for patients, relatives, and caregivers to understand because of technical language, limited health literacy, and cognitive interface barriers. This paper presents PlainHealth, a work-in-progress multimodal service suite that uses generative AI and task-specific language technologies to support accessible healthcare communication in clinical consultations, residential care, and home settings. The suite offers reusable services for audio capture, multi-recipient report generation, plain-language and easy-to-read adaptation, accessible infographic generation, and assistive care. It also includes a quality-status service that communicates output quality, traceability, and human-review status for AI-generated content. Services are delivered through cognitively accessible interfaces designed in accordance with relevant standards. The service-oriented implementation selects models and resources according to communication goals, accessibility requirements, and safety risks. The paper presents the architecture, implementation strategy, interaction design, and quality-control mechanisms, with particular attention to cognitive accessibility, human oversight, explainability, traceability, and responsible use.
