MICROLENS - A DESIGN-CENTRED APPROACH TO AI IN HEALTHCARE

Abstract

Abstract AIMS MicroLens aims to co-develop tools and functions that enable patients & caregivers to view and understand their own images. METHOD We used an explicit design-led approach, foregrounding user needs. Over 7 months we used a series of 4 meetings, written reports and online questionnaires capturing the views of 20 staff, patient and caregiver participants (55% patients & caregivers; primary & metastatic brain tumours) to explore feasibility and requirements for an AI-enhanced online medical image viewing tool. RESULTS There were two distinct patterns of preferred usage: smartphone-based “snapshot” approach vs. online plat- form with detailed review of all images. Participants wanted to use it as a “scan photo album” and said help in understanding the scans would be useful, but the desired level of technical performance was relatively low: patients and caregivers would mostly be happy with labelling of common simple structures and sides, rather than detailed annotation. Further benefits and concerns raised were ability to • share with family members who could not attend hospital appointments. • review their scans outside of an appointment. • better understand images. • Use tools to assist in the interpretation reducing staff workload • image-share causing more work for staff CONCLUSIONS Patients, caregivers and staff all expressed a clear view that an AI-enhanced online tool to allow patients to view and understand their images would be useful. However, there were significantly different use cases between different patients, and the level of functionality that patients & caregivers expressed a need for was substantially simpler than that originally envisaged. We are currently implementing a prototype.