Sharing the Care: Investigating How Conversational AI Might Facilitate Coordination Among Home Care Workers and Family Caregivers

Paper Title

Sharing the Care: Investigating How Conversational AI Might Facilitate Coordination Among Home Care Workers and Family Caregivers

Publication Info

• Topic area: Coordination of home-based care using AI-driven conversational agents.
• Keywords: Home care, family caregivers, home care workers, shared care, conversational AI, large language models, care coordination, human-centered design, AI accountability, relational care.

Background and Problem

• Problem / challenge: Coordination among family caregivers (FCs) and home care workers (HCWs) who care for the same care recipient (CR) is complex and error-prone, with limited tools to streamline communication, task management, and shift handovers.
• Significance: Addressing coordination challenges can reduce caregiver stress, improve care quality, and ensure better health outcomes for CRs.
• Motivation and related work: Prior research has largely focused on either FCs or HCWs individually, neglecting the interdependent roles these groups play in shared care. Existing technologies often fail to account for relational and emotional aspects of caregiving or introduce burdensome data work.

Solution

• Proposed approach: LLM-driven conversational AI agents to support shared care coordination among FCs and HCWs.
• Novelty:
  1. Empirical data capturing perspectives of both HCWs and FCs on shared care coordination.
  2. Insights into the potential of conversational AI agents to mediate communication and task management in non-clinical, high-stakes home care contexts.
  3. Design recommendations for agents that augment human capabilities while preserving relational care and addressing accountability.
• Procedure and key techniques:
  • Conducted qualitative video elicitation sessions with 17 participants (8 HCWs, 9 FCs).
  • Explored scenarios depicting AI agents assisting with care task tracking, shift handovers, health monitoring, and documentation.
  • Analyzed data using structural coding and thematic analysis to identify challenges and opportunities for AI integration.

Results

• Concrete findings:
  • Participants viewed AI agents as promising tools for streamlining communication, bridging language gaps, and supporting onboarding of new caregivers.
  • Agents could reduce documentation burdens through real-time voice inputs and multimedia support.
  • Participants emphasized the need for agents to signal uncertainty, enable error reporting, and complement human judgment.
• Advantage over baselines: Unlike existing tools, LLM-based agents can interpret unstructured updates, offer language translation, and provide context-sensitive guidance, opening new design possibilities for shared care coordination.
• Experiments / evaluation:
  • Video elicitation scenarios demonstrated practical roles for agents in care coordination tasks.
  • Participants provided feedback on usability, feasibility, and concerns about errors, privacy, and relational impacts.
• Limitations and future work:
  • Study excluded CRs due to ethical considerations; future research should engage CRs directly.
  • Scenarios depicted idealized agent behavior; real-world deployments may encounter additional challenges.
  • Findings are based on a U.S. urban context; further studies are needed in rural and non-U.S. settings.

Summary

This study explored how LLM-driven conversational AI agents might facilitate shared care coordination among family caregivers and home care workers. Through qualitative video elicitation sessions, participants highlighted opportunities for agents to streamline communication, reduce documentation burdens, and support shift handovers, while emphasizing the importance of preserving relational care and addressing AI accountability. The findings provide actionable design recommendations for developing agents that complement human caregiving, bridge stakeholder gaps, and enable attentive, humanistic care in sensitive home care contexts.