Skip to content

LLM-Driven Daily Conversation for Cognitive Health in Older Adults

A Large Language Model-Driven Daily Conversation Cognitive Interaction Platform: Feasibility and Preliminary Efficacy in Older Adults at High Risk of Dementia

Status
Not yet recruiting
Phases
Unknown
Study type
Interventional
Source
ClinicalTrials.gov
Registry ID
NCT07751510
Acronym
LLM-DCI
Enrollment
75
Registered
2026-08-07
Start date
2026-09-01
Completion date
2027-06-30
Last updated
2026-08-07

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

Subjective Cognitive Decline

Keywords

Large Language Model; Conversational Cognitive Stimulation; Older Adults at High Risk of Dementia; Cognitive Function; Psychosocial Function

Brief summary

This study will develop and test a daily conversation platform that uses a large language model, a type of artificial intelligence, to support cognitive health in older adults at high risk of dementia. The platform will guide participants through voice-based conversations about topics that interest them. The conversations are designed to engage memory, attention, language, planning and problem-solving, and the ability to understand social situations. The platform will also analyze conversation data and create a report describing the participant's cognitive performance. The study has two stages. In the first stage, 15 participants will test the platform through two 20-minute conversation sessions. This group will include healthy younger adults, older adults with normal cognitive function, and older adults with cognitive difficulties. Participants will provide feedback on how easy and acceptable the platform is to use. In the second stage, 60 older adults at high risk of dementia will be randomly assigned to one of three groups. The groups will receive either conversation-based cognitive activities using the platform, interest-based news reading activities, or weekly health education while waiting to receive the intervention. The study activities will continue for 12 weeks. Researchers will compare changes in cognitive function, emotional well-being, loneliness, social connection, and user experience before and after the study. The main goals are to learn whether the platform is practical and acceptable for home use and whether it may help support cognitive and psychosocial health.

Detailed description

Cognitive decline is an important health concern in aging populations. Early cognitive stimulation that is meaningful and related to daily life may help older adults maintain cognitive and psychosocial function. However, many digital cognitive programs use repetitive exercises and provide limited opportunities for natural social interaction. This study aims to develop and evaluate a large language model-driven daily conversation cognitive interaction platform for older adults at high risk of dementia. The platform uses voice-based, semi-structured conversations based on topics that interest each participant. The conversations are designed to stimulate five cognitive areas: memory, attention, language, executive function, and social cognition. During each session, participants speak with the platform using natural conversation. The system converts speech into text and analyzes selected features of the participant's responses, such as response accuracy, language fluency, organization of ideas, attention to the conversation, and understanding of social situations. The system then generates a cognitive profile report. This report is intended to provide understandable information about performance across the five cognitive areas and is not intended to diagnose dementia. The study will be conducted in two stages. Stage 1 will evaluate the initial feasibility and usability of the platform. Fifteen participants will be enrolled, including five healthy younger adults, five older adults with normal cognitive function, and five older adults with cognitive difficulties. Each participant will complete two 20-minute conversation sessions. Participants will then complete questionnaires about system usability and intervention acceptability. Their comments and experiences will be used to improve the platform before the randomized trial. Stage 2 will use a single-blind, randomized controlled design. A total of 60 older adults at high risk of dementia will be randomly assigned in equal numbers to one of three groups: Large language model conversation group: Participants will complete voice-based cognitive conversations with the platform. The activities will include exercises targeting memory, attention, language, executive function, and social cognition. Interest-based news reading group: Participants will read news content selected according to their interests and complete questions related to the content. Waitlist control group: Participants will receive weekly health education materials during the initial study period. The interventions will continue for 12 weeks. Cognitive and psychosocial outcomes will be assessed before and after the intervention. The cognitive assessments will examine overall cognitive function and specific abilities such as memory, attention, language, processing speed, executive function, and social cognition. Psychosocial assessments will examine depressive symptoms, loneliness, and social connection. The study will also evaluate intervention completion, system usability, and intervention acceptability. The findings will be used to determine whether the platform is feasible for home-based use and whether it shows preliminary benefits for cognitive and psychosocial health in older adults at high risk of dementia.

Interventions

BEHAVIORALLLM Conversation Cognitive Interaction

A voice-based cognitive stimulation program delivered through a large language model-driven conversational platform. The platform uses topics selected according to each participant's interests and guides natural conversations designed to engage memory, attention, language, executive function, and social cognition. The platform provides topic summaries, asks questions, responds to participants' answers, and adjusts the depth of follow-up questions based on response quality. Conversation responses are converted into text and used to generate indicators of cognitive performance. The platform is intended for cognitive stimulation and does not provide a clinical diagnosis.

BEHAVIORALInterest-Based Knowledge Reading

A digital cognitive activity that provides short news and knowledge summaries based on participants' interests. The content includes both familiar and unfamiliar topics, as well as preferred and non-preferred topics, to provide varied cognitive challenges. Participants answer multiple-choice questions about each summary. The questions are designed to engage reading comprehension, attention, and memory for the presented information.

BEHAVIORALWaitlist Health Education

General health education materials for older adults delivered through the LINE platform. The materials provide information related to healthy aging and general health management but do not include structured conversational cognitive stimulation or interest-based reading exercises.

Sponsors

Taipei Medical University Shuang Ho Hospital
Lead SponsorOTHER

Study design

Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
PREVENTION
Masking
SINGLE (Outcomes Assessor)

Eligibility

Sex/Gender
ALL
Age
18 Years to 90 Years
Healthy volunteers
Yes

Inclusion criteria

Stage 1: Feasibility Study Participants must belong to one of the following groups: Healthy younger adults aged 18 to 40 years. Older adults with normal cognitive function aged 65 years or older and a Montreal Cognitive Assessment score of 24 or higher. Older adults with cognitive impairment aged 65 years or older and a Montreal Cognitive Assessment score of 18 to 23. Participants must have completed at least elementary school, corresponding to 6 or more years of education. Participants must be able to read and write Chinese, communicate in Chinese, and understand the study content. Participants must be able to use a tablet or smartphone independently or with assistance from a family member. Participants must have sufficient hearing and vision, with usual corrective devices if needed, to complete the study tasks. Stage 2: Preliminary Efficacy Study Participants must be aged 65 years or older. Participants must be considered at high risk of dementia by meeting at least one of the following conditions: 1. Subjective cognitive decline: A Subjective Cognitive Decline Questionnaire score greater than 7 out of 24. A response of yes to both of the following questions: Compared with your previous usual condition, do you have memory problems? Have your family members or close contacts noticed memory problems compared with your previous condition? Performance within the normal range on standardized cognitive tests, defined as scores no more than 1.5 standard deviations below age- and education-adjusted norms on the Color Trails Test and memory-related assessments. 2. Mild cognitive impairment: A diagnosis of mild cognitive impairment made by a neurologist or psychiatrist according to accepted diagnostic criteria. Evidence of memory performance at least 1.5 standard deviations below age- and education-adjusted normative values on at least one subtest of the Wechsler Memory Scale, Third Edition. 3. Limited social interaction: Living alone, living with a person who cannot provide care, or living with a person who is absent for at least 3 days per week; or A score below 12 on the 6-item Lubben Social Network Scale. 4. Chronic disease: A physician diagnosis of hypertension or diabetes for at least 1 year; and A need for ongoing medication or lifestyle management. Participants must have completed at least elementary school, corresponding to 6 or more years of education. Participants must be able to read and write Chinese, communicate in Chinese, and understand news and study-related content. Participants must be able to use a tablet or smartphone independently or with assistance from a family member. Participants must have sufficient hearing and vision, with usual corrective devices if needed, to complete the conversation and reading tasks.

Exclusion criteria

A diagnosis of dementia or a severe neurodegenerative disorder, such as moderate-to-advanced Parkinson disease or Huntington disease. A history of a severe psychiatric disorder, such as schizophrenia or severe bipolar disorder, or current acute psychiatric symptoms that may interfere with understanding or responding during conversations. A major neurological event within the previous 6 months, such as stroke or severe traumatic brain injury, that may substantially affect cognitive or language function. Severe hearing, speech, or language impairment that prevents basic verbal communication, even when assistive devices are used. An acute medical illness, active infection, or any condition considered by a physician to make study participation inappropriate. Inability to complete a 20-minute conversation task or unwillingness to allow collection of interaction data.

Design outcomes

Primary

MeasureTime frameDescription
Verbal fluency testChange from Baseline at 13 weeksThe Verbal Fluency Test assesses language production and executive function. Participants will name as many items as possible from specified semantic categories, including fruits and vegetables, within 60 seconds. The outcome is the number of correct, non-repeated words produced. Higher scores indicate better verbal fluency. The primary outcome will be the change in the total number of correct words from baseline to the post-intervention assessment.
Chinese Verbal Learning TestChange from Baseline at 13 weeksThe Chinese Version Verbal Learning Test assesses verbal learning and memory. A list of 9 two-character nouns is presented across 4 learning trials. After each trial, participants recall as many words as possible. One point is given for each correctly recalled word. The total learning score is the sum of correct responses across the 4 learning trials and ranges from 0 to 36. Higher scores indicate better verbal learning. The primary outcome will be the change in the total learning score from baseline to the post-intervention assessment.

Secondary

MeasureTime frameDescription
Digit Span subtestChange from Baseline at 13 weeksThe Digit Span test assesses verbal attention, short-term memory, and working memory. It includes forward and backward digit span tasks. Each correctly repeated sequence receives 1 point. The forward and backward raw scores are added to produce a total score ranging from 0 to 32. Higher scores indicate better verbal attention and working memory. The outcome will be the change in total score from baseline to the post-intervention assessment.
Cookie Theft PictureChange from Baseline at 13 weeksThe Cookie Theft Picture is a well-known visual stimulus used to assess language comprehension and expressive language abilities. The picture shows a mother washing dishes beside a kitchen sink while the water is overflowing. Two children are standing on a stool and attempting to take cookies from a cupboard. While the participant describes the scene, the assessor evaluates language production, including grammatical structure, word choice, speech fluency, and syntactic errors. The participant is shown the picture and instructed, "Please tell me as completely as possible what is happening in this picture." The task ends when the participant indicates that the description is complete or stops speaking for more than 10 to 15 seconds.
Digit Symbol Substitution TestChange from Baseline at 13 weeksThe Digit Symbol Substitution Test assesses attention and information processing speed. Participants use a key to match numbers with corresponding symbols and complete as many items as possible within 90 seconds. The score is the number of correctly completed items. Higher scores indicate better attention and processing speed. The outcome will be the change in score from baseline to the post-intervention assessment.
Stroop Color and Word TestChange from Baseline at 13 weeksThe Stroop Color and Word Test assesses selective attention and inhibitory control. In the incongruent condition, participants identify the ink color while ignoring the meaning of the printed word. The score is the number of correct responses completed within the specified test period. Higher scores indicate better selective attention and inhibitory control. The outcome will be the change in score from baseline to the post-intervention assessment.
Color Trails TestChange from Baseline at 13 weeksThe Color Trails Test Part 1 assesses visual attention and information processing speed. Participants connect numbered circles from 1 to 25 in ascending order as quickly as possible. The outcome is the completion time measured in seconds. Lower completion times indicate better performance. The outcome will be the change in completion time from baseline to the post-intervention assessment. The Color Trails Test Part 2 assesses complex attention, cognitive flexibility, and executive function. Participants connect numbered circles in ascending order while alternating between 2 colors. The outcome is the completion time measured in seconds. Lower completion times indicate better performance. The outcome will be the change in completion time from baseline to the post-intervention assessment.
Theory of Mind ScaleChange from Baseline at 13 weeksThe Theory of Mind Scale assesses the ability to understand other people's thoughts, intentions, and emotions. It includes a verbal task scored from 0 to 30 and a nonverbal picture task scored from 0 to 20. The scores are added to produce a total score ranging from 0 to 50. Higher scores indicate better theory of mind ability. The outcome will be the change in total score from baseline to the post-intervention assessment.
Geriatric Depression Scale- 15Change from Baseline at 13 weeksThe 15-item Geriatric Depression Scale assesses depressive symptoms in older adults. Each item is scored as 0 or 1, producing a total score ranging from 0 to 15. Higher scores indicate more severe depressive symptoms. The outcome will be the change in total score from baseline to the post-intervention assessment.
University of California, Los Angeles, Loneliness ScaleChange from Baseline at 13 weeksThe University of California, Los Angeles Loneliness Scale assesses perceived loneliness and social isolation. The scale contains 20 items rated from 1 to 4. Total scores range from 20 to 80. Higher scores indicate greater loneliness. The outcome will be the change in total score from baseline to the post-intervention assessment.
Chinese Happiness InventoryChange from Baseline at 13 weeksThe 10-item Chinese Happiness Inventory assesses subjective happiness and well-being during the previous 3 months. Item scores are added to produce a total score. Higher scores indicate greater subjective happiness and well-being. The outcome will be the change in total score from baseline to the post-intervention assessment.
Visual Analogue Scale of Subjective Cognitive FunctionChange from Baseline at 13 weeksParticipants rate their perceived cognitive function on a 100-millimeter horizontal visual analogue scale. The left endpoint, 0 millimeters, represents the worst perceived cognitive function, and the right endpoint, 100 millimeters, represents the best perceived cognitive function. The score is the distance in millimeters from the left endpoint to the participant's mark and ranges from 0 to 100. Higher scores indicate better perceived cognitive function. The outcome will be the change in score from baseline to the post-intervention assessment.
System Usability ScaleWithin 1 week after completion of the 12-week intervention (Week 13)The System Usability Scale assesses participants' perceived usability of the digital intervention. It contains 10 items rated on a 5-point scale. Item scores are converted and multiplied by 2.5 to produce a total score ranging from 0 to 100. Higher scores indicate better perceived usability.
Acceptability of Intervention MeasureWithin 1 week after completion of the 12-week intervention (Week 13)The Acceptability of Intervention Measure assesses participants' perceived acceptability of the assigned intervention. It contains 4 items rated on a 5-point scale. The mean of the 4 item scores ranges from 1 to 5. Higher scores indicate greater intervention acceptability.

Contacts

CONTACTHsinchieh Lee
kyoko70385@gmail.com0929722265
PRINCIPAL_INVESTIGATORHsinchieh Lee

Department of Physical Medicine and Rehabilitation, Shuang Ho Hospital, Taipei Medical University, Taipei

Outcome results

None listed

Source: ClinicalTrials.gov · Data processed: Aug 8, 2026