TITLE: AI Voice Companion Reduces Depression, Anxiety in Nursing Home Residents

AI Voice Companion Eases Mental Health Challenges in Long-Term Care

Nursing home residents interacting with an AI-driven voice-based telephone companion showed significant reductions in depressive and anxiety symptoms, particularly those with elevated baseline levels, according to a new feasibility study.

The companion, known as Meela, was developed to combat loneliness, depression and anxiety among older adults in post-acute and long-term care facilities. Published in the Journal of the American Medical Directors Association (JAMDA), the study involved 28 residents, with 23 completing a four-week intervention featuring scheduled and on-demand phone conversations with the AI system.

No internet access, screens or additional devices were necessary, making the technology accessible to a population often facing barriers to digital tools. Participants were selected based on their ability to hear, speak clearly, use a telephone and engage in conversation, while understanding they were interacting with an artificial intelligence system and providing informed consent.

During calls, Meela employed speech recognition to transcribe responses, processed them via a conversational language model and responded through synthesized voice. Interactions followed a structured format, including greetings, check-in questions and open-ended prompts. The AI retained details from prior conversations, such as personal interests or family mentions, to personalize future exchanges.

Outcomes were assessed using the Patient Health Questionnaire-9 (PHQ-9) for depression and the Generalized Anxiety Disorder-7 (GAD-7) scale for anxiety. Residents scoring 9 or higher on the PHQ-9 — indicative of moderate to severe depression — experienced an average reduction of 5.7 points after the intervention. Overall, anxiety scores fell by a median of 2.3 points.

"The system was designed with older adults in mind, requiring no digital literacy or specialized equipment," the researchers noted. "Its use of familiar infrastructure, combined with a self-directed call model and minimal staff involvement, makes it feasible for implementation across skilled nursing, assisted living and independent living environments."

Challenges in Scaling Mental Health Support

Loneliness and mental health issues are prevalent in nursing homes and assisted living facilities, where operators often face difficulties in expanding psychosocial interventions and sustaining staff-resident engagement. Traditional approaches can be resource-intensive, limiting their reach.

The study positions AI-driven voice companions as a promising, scalable and low-barrier alternative for emotional support. By leveraging existing telephone systems, Meela minimizes implementation hurdles and reduces the burden on overextended staff.

However, researchers emphasized limitations, including the small sample size of 23 completers, absence of a control group and brief four-week duration. These factors restrict the ability to draw broad conclusions on long-term efficacy or generalizability across diverse populations.

"Findings suggest that AI-driven voice companions are feasible and could offer a scalable, low-barrier option to emotional support in nursing homes," the study concluded. "But larger controlled trials are needed to confirm effectiveness."

Broader Implications for Long-Term Care

As the aging population grows, demand for innovative mental health solutions in long-term care settings is intensifying. The success of Meela aligns with ongoing efforts to integrate technology into elder care without alienating users unfamiliar with digital interfaces.

Similar initiatives, such as expanded mental health services in skilled nursing facilities and federal proposals for better access to behavioral health for older adults, underscore the urgency of addressing these issues. While Meela's phone-based model offers immediate practicality, future research must explore its integration with comprehensive care plans and potential effects on overall resident well-being.

The study was conducted in a controlled feasibility context, with participants from post-acute and long-term care environments. Researchers called for expanded investigations to validate results in larger, more varied groups, potentially paving the way for widespread adoption.