Recipient organisationNorthumbria NHS Healthcare Foundation Trust
Funding£27K
PeriodSept 2025 — Sept 2026
In plain English
AI plain-English summary
A caregiver sits down with a tablet and answers questions from an AI about their relative’s memory and behaviour, generating a structured report for the memory clinic before the patient’s appointment. This matters because collecting a detailed history from a family member is essential for diagnosing dementia, but it is often rushed, inconsistent, and emotionally draining for caregivers. The LUMEN prototype uses large language models to standardise this conversation, asking clear questions and producing a summary that clinicians can review. The study will test the tool with 20–30 patient-caregiver pairs and 8–10 clinicians in real NHS memory clinics, measuring usability and clinical usefulness through surveys and interviews. If the tool proves feasible and acceptable, it could streamline memory clinic workflows, reduce caregiver burden, and help clinicians reach earlier, more accurate dementia diagnoses. The research is a small-scale feasibility study—it will generate pilot data to support larger trials and future funding applications, not yet change clinical practice.
View original technical description
LUMEN (Large Language Model for Understanding and Monitoring Elderly Neurocognition) is an AI-driven conversational prototype designed to support the dementia diagnostic process by collecting structured collateral histories from caregivers prior to memory clinic assessments. Obtaining detailed and accurate collateral histories is critical for diagnosis but can be inconsistent, time-consuming, and emotionally challenging for caregivers. LUMEN aims to standardise this process, reduce caregiver burden, and enhance diagnostic efficiency by using large language models (LLMs) to guide structured history-taking. This feasibility study with a nested qualitative evaluation will assess LUMEN’s usability, acceptability, and feasibility in real-world clinical settings. We will recruit 20–30 patient-caregiver dyads, with caregivers using LUMEN to generate a collateral history after attending a memory clinic appointment. Usability will be measured using the System Usability Scale (SUS) and National Aeronautics and Space Administration Task Load Index (NASA-TLX), while qualitative semi-structured interviews will explore user experience, accessibility, and barriers to adoption. In parallel, 8–10 clinicians will review LUMEN-generated histories to evaluate their clinical utility, completeness, and relevance. Aligned co-production workshops, held in partnership with community groups to ensure diverse representation, will explore issues around language and culture, refining the tool’s interface, accessibility, and output structure for integration into NHS memory clinics. By exploring feasibility, usability, and clinician perceptions, this study will generate critical pilot data to support future intermediate fellowship applications and larger-scale evaluations. Ultimately, LUMEN has the potential to enhance early dementia diagnosis and streamline memory clinic workflows, making AI-assisted collateral history collection a scalable and clinically viable solution.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know