A Mechanism Randomised Controlled Trial of a Three-Agent LLM-Augmented mHealth Intervention for Late-Life Loneliness in Older Adults
Study Overview
Status
Detailed Description
Loneliness in later life is associated with adverse physical and mental health outcomes. Conversational applications based on large language models have been proposed as a scalable form of support, but most published evaluations remain at the pilot feasibility stage, and few have used a comparator that holds interface richness, attention, and novelty constant. Without such a comparator, any observed benefit cannot be attributed to the language model rather than to the surrounding experience.
This trial addresses that gap using an architecturally matched comparator. Both arms deliver the identical mobile application: the same visual interface, navigation, three named conversational companions, tool layer, and in-app assessment prompts. The sole difference is the backend that generates companion response content. In the Hybrid arm, responses are generated by a large language model using each companion's system prompt, conversation history, and user input, and the arm exhibits language-model-distinctive conversational behaviours including anchoring on specific content, cross-session memory, explicit admission of unfamiliarity, mixed-content routing, and generative summarisation. In the Rule-based arm, responses are produced by a template system driven by keyword matching and conversational state, without cross-session memory and without those behaviours; conversational content in this arm is not transmitted to any external service.
Seventy-two community-dwelling Cantonese-speaking adults aged 60 or above who score at least 2 on the De Jong Gierveld Emotional and Social Loneliness Scale are allocated 1:1, stratified by baseline emotional loneliness using permuted blocks within stratum, with allocation concealment. A designated staff member performs randomisation and onboarding and does not collect outcome data. The Principal Investigator conducts all in-person assessments while unaware of allocation. Participants are not informed which version they use and are debriefed at study completion.
The intervention period is four weeks, followed by four weeks of follow-up with continued application access and continuous usage logging. The primary analysis is a longitudinal 1-1-1 multilevel mediation model with arm as the between-person predictor, session-level perceived responsiveness as the within-person mediator, and emotional loneliness as the outcome; the indirect effect is tested using Monte Carlo confidence intervals. The study is powered for the indirect effect rather than for confirmatory testing of the between-arm main effect, which is reported as an effect estimate with a 95% confidence interval. A falsifiable specification of the proposed mechanism is that an arm advantage appears on emotional loneliness but not on social loneliness.
A three-layer safety architecture operates in both arms: a Cantonese-calibrated distress detector that surfaces Hong Kong crisis resources in real time and alerts the research team; a daily 100% audit of Thought Exercise events against pre-specified scope criteria; and weekly Principal Investigator review of all flags and audits. Pre-specified pause criteria are an intervention-related serious adverse event, boundary-crossed events exceeding 10%, or distress detector recall below 0.90 on a real Cantonese corpus.
Study Type
Enrollment (Estimated)
Phase
- Not Applicable
Contacts and Locations
Study Contact
- Name: Junyu Zhao
- Phone Number: 852-60919592
- Email: zhaojyxs@connect.hku.hk
Participation Criteria
Eligibility Criteria
Ages Eligible for Study
- Adult
- Older Adult
Accepts Healthy Volunteers
Description
Inclusion Criteria:
- Aged 60 years or above
- Self-identified Cantonese as primary language of communication
- Community-dwelling in Hong Kong (not in residential care)
- De Jong Gierveld Emotional and Social Loneliness Scale total score of 2 or above at screening, indicating at least moderate loneliness
- Owns a smartphone, or willing to use a study-provided device
- Able to demonstrate understanding of the study using the teach-back method
- Willing to provide written informed consent
Exclusion Criteria:
- Acute suicidality, defined as a PHQ-9 item 9 score above 1 with reported active intent
- Currently receiving formal psychiatric treatment
- Severe hearing or visual impairment precluding application use even with accommodation
- Self-reported diagnosis of dementia or significant cognitive impairment
- Unable to demonstrate understanding of the study using the teach-back method
- Prior participation in the feasibility phase of this research programme
Study Plan
How is the study designed?
Design Details
- Primary Purpose: Basic Science
- Allocation: Randomized
- Interventional Model: Parallel Assignment
- Masking: Double
Arms and Interventions
Participant Group / Arm |
Intervention / Treatment |
|---|---|
|
Experimental: Hybrid arm (LLM-driven)
Participants receive the mobile application with companion responses generated by a large language model backend, using each companion's system prompt, conversation history, and user input.
Language-model-distinctive conversational behaviours (content anchoring, cross-session memory, admission of unfamiliarity, mixed-content routing, generative summarisation) are operationally present.
Recommended use is at least three sessions per week over four weeks; actual use is at the participant's discretion and is logged.
|
Mobile application providing three named Cantonese-language conversational companions plus four in-app tools (Action Loop, Thought Exercise, Education, Progress).
Companion replies are generated turn-by-turn by a large language model (DeepSeek-V3, accessed via Firebase Cloud Functions), conditioned on the companion's system prompt, retrieved conversation history, and current user input.
This arm can therefore produce five conversational behaviours the comparator cannot: anchoring on specific content the participant has just said, memory carried across separate sessions, explicit admission of unfamiliarity, routing of single messages containing mixed emotional and informational content, and generative summarisation.
Occurrences are tagged in system logs and analysed as a cumulative within-arm exposure variable.
Conversational text is transmitted to the model provider for response generation; participants are advised at consent not to include identifying information.
Recommended use is a
|
|
Active Comparator: Rule-based arm (template-driven)
Participants receive an application identical in interface, navigation, companion personae, tool layer, and in-app assessment prompts, with companion responses produced by a template system driven by keyword matching and conversational state.
Cross-session memory and the language-model-distinctive behaviours are absent.
Conversational content is not transmitted externally.
Recommended use and logging are identical to the Hybrid arm.
|
Mobile application identical to the experimental arm in interface, navigation, companion names and personae, tool layer (Action Loop, Thought Exercise, Education, Progress), notification schedule, and all in-app assessment prompts.
The sole difference is the response-generation backend: companion replies are assembled from pre-written templates selected by keyword matching and conversational state, with no language model involved.
The system retains no memory across sessions, cannot anchor on unanticipated content, cannot admit unfamiliarity outside scripted cases, and cannot generate novel summaries.
Conversational content is not transmitted to any external service.
This is an attention- and interface-matched active comparator rather than a waitlist or usual-care control: participants receive equivalent contact time, interface richness, tool access, and prompting schedule.
Recommended use, expected daily duration, and logging are identical to the experimental arm.
|
What is the study measuring?
Primary Outcome Measures
Outcome Measure |
Measure Description |
Time Frame |
|---|---|---|
|
Session-level perceived responsiveness (primary mediator)
Time Frame: After each conversation session, Weeks 1 through 4
|
Brief perceived-responsiveness measure administered in-app immediately after each companion conversation.
Single-item sliders scored 1-7 covering Understanding, Validation, Caring and Insensitivity.
Higher scores indicate greater perceived responsiveness.
This is the within-person mediator in the pre-specified primary mediation model, not an efficacy endpoint.
|
After each conversation session, Weeks 1 through 4
|
|
Change in emotional loneliness (De Jong Gierveld emotional subscale)
Time Frame: Baseline (Week 0) and Week 4
|
De Jong Gierveld Emotional and Social Loneliness Scale, 3-item emotional subscale.
Score range 0-3; higher scores indicate greater emotional loneliness.
Change from baseline to Week 4. This is the outcome variable in the pre-specified primary mediation model.
The trial is powered for the indirect effect via session-level perceived responsiveness and is not powered for confirmatory testing of the between-arm difference, which is reported as an effect estimate with a 95% confidence interval.
|
Baseline (Week 0) and Week 4
|
Collaborators and Investigators
Sponsor
Study record dates
Study Major Dates
Study Start (Estimated)
Primary Completion (Estimated)
Study Completion (Estimated)
Study Registration Dates
First Submitted
First Submitted That Met QC Criteria
First Posted (Actual)
Study Record Updates
Last Update Posted (Actual)
Last Update Submitted That Met QC Criteria
Last Verified
More Information
Terms related to this study
Keywords
Additional Relevant MeSH Terms
Other Study ID Numbers
- EA260523
Drug and device information, study documents
Studies a U.S. FDA-regulated drug product
Studies a U.S. FDA-regulated device product
This information was retrieved directly from the website clinicaltrials.gov without any changes. If you have any requests to change, remove or update your study details, please contact register@clinicaltrials.gov. As soon as a change is implemented on clinicaltrials.gov, this will be updated automatically on our website as well.