Aims
Cognitive clinic attendees often receive written material to support the retention of information following a cognitive assessment. We aimed to improve the existing feedback letter template by involving end users and a large language model (LLM) to rapidly generate and evaluate prototype content.
Methods
We performed semi-structured interviews with patients and their support people at a specialist cognitive clinic prior to, and following, written feedback provision. Interviews were audio-recorded, transcribed and analysed thematically by two researchers. These themes were entered into a LLM (ChatGPT) to generate prototype letters. Clinic staff then critically appraised the output in three 30-minute design workshops before evaluation by an independent older adults’ consumer advisory group.
Results
Of the 10 interviews performed prior to feedback, five themes were identified: 1) reason for referral, 2) contextualised investigation results, 3) diagnosis and prognosis, 4) actions to modify risk of decline and 5) planning for the future. Of the eight interviews performed post written feedback, five themes were identified: 1) the need for clear indication of target audience, 2) clear letter structure and navigability, 3) simple language, 4) alignment between written and verbal feedback and 5) actions to modify risk of decline. The revised letter was reviewed favourably by older consumers.
Conclusion
Patients and support people value feedback that is personalised, accessible and aligned with verbal discussions with practical strategies for maintaining independence and brain health. Combining their needs and preferences with a LLM and tempering this with clinical expertise allows rapid production of prototypes for future evaluation.