JMIR Publications

ABSTRACT

Background:

Older adults have a high rate of loneliness, which contributes to increased medical morbidity and mortality. Peer support interventions combine persons with a similar struggle or condition to create an environment of mutual support and have demonstrated positive benefits across a broad range of medical conditions and demographics. Peer support programs contrast the usual provider-based clinical care models because they offer more direct support for empowerment, highlighting existing skill sets and intrinsic motivation. Peer support can be effectively delivered in various formats, including digital forms.

Objective:

This study examines a novel service delivering peer support through an anonymous, real-time digital experience using trained moderators to guide peer-to-peer interactions. The experience of a cohort of 699 adults aged 65 years and older who engaged in an anonymous, chat-based, digital peer support was analyzed to determine 1) if participation led to the measurable aggregate change in loneliness and optimism and 2) the impact of peers on changes in loneliness and optimism.

Methods:

Participants were each prompted with a single question: “What’s your struggle?” and from their free-text response, were matched on their self-identified area of struggle using a proprietary artificial intelligence model to participate in anonymous, digital, chat-based exchanges with peers. User messages from peer conversations were analyzed to quantitatively measure the change in each specific emotion using a third-party, public natural language processing model (GPT-4). The analysis of the change in emotion was initially performed at the level of an individual user and then averaged across all users with similar emotion types to produce a collective trend for each emotion. To evaluate the impact of peers on the loneliness and optimism trends, we perform propensity matching to align the moderator+single user and moderator+small group chat cohorts and then compare the emotion trends between the matched cohorts.

Results:

Loneliness and optimism trends significantly improve after 8-10 minutes into the chat. We observed a significant improvement in the loneliness and optimism trends between the moderator and a small group chat compared to the moderator and a single-user chat cohort.

Conclusions:

Chat-based peer support may be a viable intervention to help address loneliness in older adults and present an alternative to traditional care. The promising results support the need for further study to expand the evidence for such cost-effective options. Clinical Trial: N/A