Predicting Social Isolation in Older Adults with Cognitive Decline
Author Information
Author(s): Kang Bada, Park Min Kyung, Kim Jennifer, Yoon Seolah, Heo Seok-Jae, Kang Chaeeun, Hong Dahye
Primary Institution: Yonsei University, Seoul, Republic of Korea
Hypothesis
This study aimed to develop and validate machine learning models to predict social interaction and loneliness levels among older adults with subjective cognitive decline and mild cognitive impairment.
Conclusion
The study demonstrated that machine learning models can effectively predict social interaction and loneliness in older adults at risk for dementia.
Supporting Evidence
- The Random Forest model was the most suitable for predicting social interaction level with an AUC of 0.935.
- The Gradient Boosting Machine model was the most suitable for predicting high levels of loneliness with an AUC of 0.887.
- Low physical movement in the morning was associated with low social interaction.
- Decreased sleep quality at night was linked to high levels of loneliness.
Takeaway
Researchers used computers to help figure out how lonely older people might feel and how much they interact with others, which can help prevent them from feeling isolated.
Methodology
The study used surveys for demographic and health-related data, along with mobile Ecological Momentary Assessment for real-time measurement of social interaction and loneliness, and wrist-worn actigraphy for sleep and activity data.
Participant Demographics
Community-dwelling older adults, with 67 having subjective cognitive decline and 32 having mild cognitive impairment.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website