Leveraging a Large Language Model to Alleviate Caregiver Mental Burden in Dementia Care
Abstract (300 words maximum)
Providing care for individuals who suffer from Alzheimer’s Disease and Related Dementia (ADRD) can cause significant emotional and psychological strain on informal caregivers, causing chronic stress, anxiety, and burnout. Facing these ongoing mental health struggles can interfere with decision-making and ultimately compromise the quality of care they provide for their patients. This project proposes to utilize a large language model (LLM) to assess caregiver mental health with a patient behavior grading system. The system will use behavioral symptoms of the patient, who is being taken care of by the caregiver, such as agitation, wandering, or confusion, as an input. It will assess the stress level based on how distressing it is for the caregiver. This metric, along with information about behavior frequency, will be used to prompt the LLM, providing critical context about both the caregiver’s environment and their likely emotional state. Through this prompting, the LLM will generate responses that acknowledge the caregiver’s stress and offer tailored mental health support. This includes self-care reminders, suggestions to manage stress, and encouragement for caregivers to seek external support when needed. This mental health detecting module can offer tailored mental health support to help caregivers manage their stress, prevent burnout, and continue providing quality care to their loved ones with dementia.
Academic department under which the project should be listed
CCSE - Computer Science
Primary Investigator (PI) Name
Xinyue Zhang
Leveraging a Large Language Model to Alleviate Caregiver Mental Burden in Dementia Care
Providing care for individuals who suffer from Alzheimer’s Disease and Related Dementia (ADRD) can cause significant emotional and psychological strain on informal caregivers, causing chronic stress, anxiety, and burnout. Facing these ongoing mental health struggles can interfere with decision-making and ultimately compromise the quality of care they provide for their patients. This project proposes to utilize a large language model (LLM) to assess caregiver mental health with a patient behavior grading system. The system will use behavioral symptoms of the patient, who is being taken care of by the caregiver, such as agitation, wandering, or confusion, as an input. It will assess the stress level based on how distressing it is for the caregiver. This metric, along with information about behavior frequency, will be used to prompt the LLM, providing critical context about both the caregiver’s environment and their likely emotional state. Through this prompting, the LLM will generate responses that acknowledge the caregiver’s stress and offer tailored mental health support. This includes self-care reminders, suggestions to manage stress, and encouragement for caregivers to seek external support when needed. This mental health detecting module can offer tailored mental health support to help caregivers manage their stress, prevent burnout, and continue providing quality care to their loved ones with dementia.