Leveraging the J48 Algorithm to Inform Community-Based AI Solutions for African American Dementia Caregiving
Sunmoo Yoon1, Melissa Patterson2, Frederick Sun3
1General Medicine, Columbia University, New York, NY.
None:
We applied the J48 machine learning algorithm to build models that identify demographic and caregiving factors associated with perceived risks and benefits of community-based AI solutions for African American family caregiving. 572 diverse family members of a person with Alzheimer's disease and related dementias (ADRD) participated in this online survey in the U.S. The J48 algorithm (C4.5) identified race as the primary predictor for AI support, with African Americans favoring AI-enabled hospital-based diagnostic testing and faith-based apps regardless of their demographic or caregiving factors. Conversely, risk perceptions were heightened among highly educated White family members (aged 25-34) for clinical AI and among younger family members (ages 18-34) for community-based meal-related apps. Overall, model's F-measures (0.83) and PRC areas (0.74) confirm that community-based AI preference is driven more by cultural context and specific use cases than by general caregiving circumstances. While African Americans are willing to support the development of AI applications, enthusiasm does not extend uniformly across all community settings. While scientists should prioritize AI in clinical and faith-based settings, they must exercise caution in the nutritional domain where algorithms may perpetuate bias.
Related Concept Videos
Alzheimer's Disease: Treatment
Dementia l: Introduction
Alzheimer Disease l: Introduction
Alzheimer's Disease: Overview
The clinical diagnosis of AD hinges on the presence of memory and other cognitive impairments. Biomarkers, such as changes in Aβ and tau...
Documentation in Long-Term and Home Healthcare Setting
Long-Term Care Facilities
Dementia
The progression of dementia is generally gradual.

