A mobile application (chatbot) to automate delivery of sleep‐dependent memory tasks

• Published in: Alzheimer's & dementia Vol. 19; no. S18
• Main Authors: Lam, Aaron Kin Fu, D'Rozario, Angela, Bradford, Dana, Ireland, David, Fripp, Jurgen, Simonetti, Simone, Mills, Peta, Naismith, Sharon L
• Format: Journal Article
• Language: English
• Published: 01.12.2023
• ISSN: 1552-5260; 1552-5279
• DOI: 10.1002/alz.078658

Background Prior work has shown that sleep is critical for memory through a process called sleep‐dependent memory (SDM). However, SDM appears to diminish with age, and may be further compromised in those with mild cognitive impairment (MCI) and dementia. Typically, studies examining SDM have been small, often restricted by the need for administration in the sleep laboratory. Consequently, understanding of SDM decrements in MCI has been limited, hindering targeted developments of treatments for SDM. We aimed to develop an app‐based ‘chatbot’ for measuring SDM, allowing data to be obtained at scale and in response to treatments without the need for sleep laboratory attendance. Method An initial prototype of the chatbot was developed for Android and iOS utilising a 32‐item word‐pair task demonstrated to be suitable for people with MCI. The task delivery included pre‐sleep learning and memory and post‐sleep recall. To reduce content‐specific practice effects, an alternative word‐pair list was included for repeat testing. Co‐design focus group methodologies were used to ascertain usability and acceptability for older people with MCI who attended a memory clinic. Participants (N = 11) attended one of two 90‐minute focus groups in which semi‐structured questions probed technology use and gathered feedback for app improvement. Results Participant feedback validated that core useability and content features of the chatbot were appropriately developed for the specific use of older adults at risk of dementia. Participants were comfortable with the chatbot obtaining demographics and habitual sleep and wake times and providing notifications for completing the task pre‐ and post‐sleep. No negative feedback was received on the layout of the instructions for the test or delivery of the word‐pairs task. Suggested improvements included ability to defer testing and customise notifications, and provision of performance feedback or rewards for compliance. This information was used to further develop the chatbot. Conclusion The SDM chatbot app will now undergo further clinical and psychometric testing to determine how performance changes in relation to, rest‐activity rhythms, clinical features, and sleep and dementia biomarkers.