Self-management for Chronic Illness: A Scoping Review on Designing Virtual Assistants for Patient-Centered Care

Paper Title

Self-management for Chronic Illness: A Scoping Review on Designing Virtual Assistants for Patient-Centered Care

Publication Info

• Topic area: Virtual assistant design for chronic illness self-management
• Keywords: Chronic illness, virtual assistants, patient-centered care, AI integration, participatory design, self-management, holistic care, reductionist care, health technology, human-centered AI

Background and Problem

• Problem / challenge: Virtual assistants (VAs) for chronic illness self-management often fail to deliver patient-centered care (PCC), prioritizing reductionist approaches (e.g., adherence, behavior change) over holistic care (e.g., relationality, empowerment). AI integration has not resolved this gap and may exacerbate it by focusing on technical personalization at the expense of meaningful patient involvement.
• Significance: Chronic illness affects over one-third of adults annually, requiring individualized, holistic care approaches to improve quality of life, symptom management, and patient empowerment. Misaligned VA designs risk reinforcing narrow, clinical care models that overlook patients' lived experiences.
• Motivation and related work: Prior research has emphasized the importance of participatory design and patient-centeredness in health technologies. However, most VAs for chronic illness self-management fail to integrate patients meaningfully into the design process, and AI-driven systems often prioritize algorithmic optimization over holistic care needs.

Solution

• Proposed approach: A scoping review of 55 studies to analyze how patient-centeredness is enacted in the design and development of VAs for chronic illness self-management, with a focus on care conceptualizations, system functionalities, and patient participation.
• Novelty:
  1. Mapping of care dimensions implemented in VAs, distinguishing between reductionist and holistic approaches.
  2. Categorization of patient roles in the VA design process, from non-involvement to co-design.
  3. Design implications for expanding care dimensions and improving patient involvement in AI-driven VAs.
• Procedure and key techniques:
  • Conducted a scoping review following PRISMA-ScR guidelines.
  • Analyzed 55 studies from 2011–2025, covering both AI and non-AI VAs.
  • Thematically analyzed care concepts, functionalities, and patient participation levels.
  • Categorized care approaches into reductionist (adherence, behavior change, proaction) and holistic (health literacy, relationality, autonomy, empowerment).

Results

• Concrete findings:
  • 53% of VAs prioritized reductionist care (e.g., adherence, behavior change), while 47% emphasized holistic care (e.g., relationality, empowerment).
  • AI-powered VAs were more likely to adopt reductionist approaches (64%) compared to non-AI VAs (43%).
  • Patient participation was limited: 45% of studies involved patients only as evaluators, and 31% did not involve patients at all.
  • Co-design approaches were rare (14.5%) but led to more holistic and patient-centered care concepts.
• Advantage over baselines:
  • Non-AI VAs demonstrated higher patient involvement and more holistic care approaches compared to AI-driven systems.
  • AI-driven VAs often conflated technical personalization with patient-centeredness, failing to address relational and psychosocial care needs.
• Experiments / evaluation:
  • Studies included usability tests, focus groups, interviews, and feasibility studies.
  • AI technologies used included machine learning (ML), natural language processing (NLP), and large language models (LLMs).
  • Non-AI VAs relied more on two-way conversational formats and participatory design methods.
• Limitations and future work:
  • Limited inclusion of non-English studies and grey literature may bias findings toward Western academic perspectives.
  • Future research should explore industry practices, diverse cultural contexts, and adaptive co-design methods for chronic illness populations.

Summary

This scoping review highlights significant gaps in the design of virtual assistants (VAs) for chronic illness self-management, particularly in delivering patient-centered care (PCC). While AI integration has introduced advanced personalization features, it has not enhanced holistic care or meaningful patient involvement. Non-AI VAs demonstrated higher alignment with PCC principles, emphasizing relational and psychosocial care. The review provides a mapping of care dimensions, a categorization of patient roles, and design implications to guide future VA development. These findings underscore the need for participatory design practices and adaptive methodologies to ensure VAs align with the lived experiences and holistic care needs of chronic illness patients.