AI in Global Health: The View from the Front Lines

Title of the Paper

AI in Global Health: The View from the Front Lines

Bibliographic Information

• Subject Area: Applications and societal impact of artificial intelligence in global health
• Keywords: Artificial intelligence, social good, healthcare, India, HCI4D, qualitative research

Research Background and Problem Statement

• Identified Problems or Challenges:

  • Artificial intelligence (AI) technologies are widely regarded as tools to promote social good in global health, but their implementation may exacerbate marginalization and privacy violations.
  • Despite the growing attention to AI in health systems of the Global South, the voices of frontline healthcare workers (especially women) and their service communities are rarely represented in existing discussions.
  • Many AI applications in health focus on isolated problems (e.g., diagnosis or disease prediction) and lack a deep understanding of implementation contexts and broad engagement with local communities.

• Why This Problem Is Important:

  • The majority of the world's population resides in resource-constrained, marginalized, or developing environments, making AI solutions targeting these regions critical for achieving sustainable development goals.
  • Frontline health services play a critical role in the healthcare systems of Global South countries. Understanding their workflows, cultural contexts, and power structures is essential for designing and implementing effective AI systems.

• Research Motivation and Related Work:

  • The authors are motivated to analyze AI from the perspective of real frontline healthcare workers, providing concrete recommendations on how to design and implement AI in low-resource settings to promote social good.
  • Related work includes academic research (e.g., technology applications in community health projects) and gray literature (white papers published by governments, nonprofits, and industry) discussing AI applications in global health.

Proposed Solution

• Proposed Methods or Solutions:

  • Conduct thematic discourse analysis to uncover the main applications, driving motivations, stakeholders, and local participation in AI for frontline health.
  • Develop a deep understanding of how to design AI systems that address inequalities in power structures while supporting the daily tasks of data-driven frontline healthcare workers through actionable insights.

• Innovative Contributions:

  • Combine AI perspectives with ethnographic data from frontline healthcare workers and marginalized communities, centering on their experiences to redefine "social good" through the lenses of social justice, post-development theory, and transnational feminism.
  • Propose critical perspectives for improving the design of different types of AI systems and practical steps for deploying AI in resource-constrained settings.

• Implementation Steps and Techniques:

  • Employ a combination of thematic discourse analysis (analyzing 347 academic and gray literature documents) and three years of ethnographic data from resource-constrained communities in Delhi, India.
  • Systematically map frontline health workflows (e.g., data collection, disease response, psychological support) and analyze potential interaction points with AI technologies.

Research Findings

• Specific Findings:

  • Identified key application areas of AI in global health, such as diagnosis and screening, disease prediction, and monitoring.
  • Highlighted design opportunities for AI systems that align with frontline health workflows, such as improving service quality through optimized resource allocation and reducing repetitive tasks through automation.
  • Raised ethical concerns for data-driven AI systems, including issues of data privacy, transparency, and ownership.

• Advantages Over Existing Solutions:

  • Emphasized the importance of ethnographic data in understanding complex local contexts and frontline workflows, avoiding overly simplistic or impractical AI designs.
  • Advocated for prioritizing user agency (e.g., frontline healthcare workers) in design, from avoiding mandatory workflows to valuing local knowledge and practices.

• Experimental or Evaluation Results:

  • Literature analysis revealed that AI applications are predominantly focused on high-resource datasets or mainstream health issues, with limited relevance to resource-constrained environments in the Global South.
  • Ethnographic research demonstrated how frontline healthcare workers navigate complex sociopolitical ecosystems and suggested ways AI could better support these workflows.

• Limitations and Future Directions:

  • Limitations: The literature analysis only included English-language publications, potentially overlooking relevant research in other languages. The ethnographic data was primarily based on the Delhi region in India, which may not fully represent other countries or regions.
  • Future Directions: Deepen research on the ethical relationship between frontline healthcare work and AI; strengthen interdisciplinary collaborations to drive impactful social good designs; develop environmentally friendly AI systems to reduce resource consumption.