Using artificial intelligence (CognoSpeak™) in memory assessments: a GP interview study

BACKGROUND: The memory assessment pathway (MAP) for people with subjective memory deficits (dementia, mild cognitive impairment, and other diagnoses) is under huge strain and new diagnostic technologies have been identified as a high priority for research. AIM: To investigate the views of GPs on the MAP, and on how an artificial intelligence tool (CognoSpeak™) could be implemented. DESIGN & SETTING: Qualitative interview study in a large region of the NHS (South Yorkshire) in England. METHOD: Eighteen GPs were recruited using convenience sampling to undertake semi-structured interviews, which were analysed using reflexive thematic analysis (demographic data were monitored to ensure diversity). RESULTS: GPs think that the MAP has system-wide problems, and that GPs are overworked yet underutilised. They highlighted assessment and referral dilemmas, and the perspectives of patients and families. When asked about implementation of CognoSpeak™, they gave their thoughts on the optimal sites of implementation. The GPs also highlighted barriers and difficulties, as well as the opportunities and benefits, and they made proposals for the future development of CognoSpeak™. CONCLUSION: GPs thought effective implementation of CognoSpeak™ could save time, expedite diagnosis, free-up much needed capacity, and improve the longitudinal assessment of people with mild cognitive impairment. A major concern among GPs was the potential for unintended consequences such as creating additional unfunded work. They also felt it could exacerbate difficulties at the intersections between subjective memory deficits and other factors such as low mood, excess alcohol consumption, learning difficulties, or language and culture. They were concerned about poor access to technology among older and more economically deprived people.