Can use of the LipReader communication app improve communication in voiceless intensive care patients?

Arm 1: LipReader intervention Participants will use LipReader during a single supervised study session conducted at the bedside in the intensive care unit or step-down setting. LipReader is a smartphone-based visual speech recognition application designed to support communication for individuals who are unable to use their voice. The application uses the device’s built-in front-facing camera to capture lip movements during silent speech and applies an artificial intelligence model to predict silently mouthed phrases, which are then presented as on-screen text and/or synthesised speech output. The system is designed for short, functional utterances relevant to bedside care and everyday patient needs. For this Phase 1 study, LipReader will be used only under supervision of the research team and only for non-critical, functional communication. It is non-invasive, does not provide diagnostic or therapeutic advice, does not replace clinical judgement or standard communication pathways, and will not be used to support goals-of-care discussions or other high-stakes clinical decision-making. LipReader is used for research purposes only in this study. Each participant will complete a brief, standardised orientation to LipReader and the comparator communication method before commencing the communication tasks. Participants will act as their own comparator by completing communication attempts using both LipReader and standard bedside communication tools during the same supervised session. All participants will complete the LipReader condition and the comparator condition in the same session. The order of communication methods will be standardised, with LipReader completed first followed by the comparator communication method. A short rest break of approximately 5 minutes will be offered between communication methods, or longer if required due to fatigue, distress or clinical care needs. Participants will communicate a set of pre-agreed functional phrases. Each communication attempt will have an allocated time window of up to 2 minutes. The time from commencement of each communication attempt to correct identification of the participant’s intended message will be recorded. If the intended message is not correctly identified within the allocated time window, the attempt will be recorded as unsuccessful for that attempt. The total study session will last approximately 20 to 30 minutes. If a session is interrupted by clinical care, fatigue or distress, the interruption will be documented and the session may be paused or terminated as clinically appropriate. LipReader will be used on a research-owned, password-protected smartphone configured by the research team. Participants will not use personal devices, will not be required to log in or enter identifiable information, and no conversational content will be retained as part of the clinical record. Adherence will be monitored directly by the research team during the single study session.