Societal-Scale Human-AI Interaction Design? How Hospitals and Companies are Integrating Pervasive Sensing into Mental Healthcare

Title of the Paper

Societal-Scale Human-AI Interaction Design? How Hospitals and Companies are Integrating Pervasive Sensing into Mental Healthcare

Paper Information

• Research Domain: Human-Computer Interaction Design, Socio-Technical Systems, Applications of Artificial Intelligence in Healthcare
• Keywords: Human-Computer Interaction Design, Socio-Technical Systems, Artificial Intelligence, Mental Health, Healthcare Ecosystem

Research Background and Issues

• Issues or Challenges:

  • AI-driven Pervasive Sensing Platforms hold significant potential in the mental health domain but face challenges stemming from complex ecosystems, socio-technical dynamics, and concerns about privacy and equity.
  • There are methodological limitations in designing societal-scale AI platforms to maximize benefits while minimizing negative impacts.
  • Designers of such platforms often struggle to achieve the goal of comprehensive mental health systems, as design practices are constrained by specific data sources, intervention types, and evaluation metrics.

• Importance:

  • Mental health issues are increasingly severe worldwide, and AI technology offers potential solutions across various levels, from daily health monitoring to clinical diagnosis and treatment.
  • From both technical and societal-scale impact perspectives, the application of AI in mental health is a promising yet highly complex domain.

• Research Motivation and Related Work:

  • This study aims to explore the methods employed by designers of mental health AI platforms and how these methods address systemic complexities.
  • Additionally, the paper compares the differing goals between designers and stakeholders (such as patients and doctors).

Solution

• Methodology:

  • Researchers conducted interviews with designers of mental health technology platforms, doctors, and patients to comprehensively analyze how designers navigate systemic complexities.
  • The interviews collected insights into designers' design processes, strategies for addressing challenges, and the real needs from the perspectives of doctors and patients.

• Innovations:

  • Proposed a methodological framework for designing mental health AI platforms centered around three steps: data, intervention types, and system performance evaluation.
  • Unveiled the design tensions between AI and privacy/equity, emphasizing the need to consider multi-stakeholder, multi-layered requirements in societal-scale AI system design.

• Implementation Steps and Key Techniques:

  • Step 1: Diversify data sources. Designers aim to integrate various data types (clinical assessment data, social media data, patients' digital footprints, etc.) to enhance the diagnostic accuracy and usability of AI models.
  • Step 2: Diversify intervention methods. The goal is to develop a comprehensive AI system that includes both daily mental health recommendations and clinical treatments.
  • Step 3: Comprehensive evaluation methods. Gradually refine individual metrics to establish the platform's effectiveness, ultimately demonstrating its ability to improve patients' overall mental health.

Research Findings

• Specific Findings:

  • Described current work patterns of designers and the challenges they face in achieving the goals of mental health AI platforms.
  • Highlighted how limitations in data sources, intervention types, and single evaluation metrics lead platforms to deviate from their comprehensive objectives.

• Comparative Advantages Over Existing Solutions:

  • This study not only focuses on technical implementation but also delves into social and ethical issues in design practices and tensions among stakeholders.
  • Provides practical design feedback and theoretical insights for designing societal-scale AI platforms.

• Experimental or Evaluation Results:

  • Actions taken by platform designers reveal significant discrepancies between their goals and the expectations of patients and doctors. For example, designers tend to prioritize reducing face-to-face interaction time, while patients prefer more direct interaction.
  • Market forces driven by data have polarized application platforms (consumer-focused vs. clinically-focused applications), exacerbating fragmentation within the healthcare ecosystem.

• Limitations and Future Directions:

  • Current design practices fail to address issues of social equity, such as whether mental health AI can better serve minority groups or remote areas.
  • Lack of long-term methods for evaluating the overall effectiveness of the system may hinder accurate assessment of its true impact.
  • Future research should explore multi-layered, multi-stakeholder collaborative approaches (e.g., participatory design and simulation design) to further optimize societal-scale AI platforms.