MedAI-SciTS: Enhancing Interdisciplinary Collaboration between AI Researchers and Medical Experts

Research Background and Issues

• What problems or challenges did the authors identify?

  • Interdisciplinary collaboration between the medical and AI fields is challenging, primarily due to terminology barriers, methodological differences, misaligned goals, and resource allocation issues.
  • Medical practitioners lack trust and understanding of AI technologies, questioning the transparency of complex systems, while AI researchers are unfamiliar with clinical needs and real-world workflows.
  • Existing frameworks for the Science of Team Science (SciTS) are overly abstract and fail to adapt to the rapidly evolving and complex collaborative environment of medical-AI research.

• Why is this issue important?

  • AI technologies hold tremendous potential in the medical field, such as improving diagnostic accuracy and optimizing treatment, but complex interdisciplinary challenges hinder the efficient translation of scientific achievements.
  • Effectively addressing these issues can not only drive innovation but also accelerate the practical application of AI in healthcare, benefiting a broader patient population.

• Research Motivation and Related Work

  • The theoretical framework of SciTS serves as a foundation for addressing multidisciplinary collaboration issues, but existing tools are often generic or fragmented, making them insufficient for the specific needs of medical-AI research.
  • While fields like Human-Computer Interaction (HCI) and Computer-Supported Cooperative Work (CSCW) have proposed methods for knowledge sharing and collaboration, a comprehensive and targeted solution is still lacking.

Solution

• What methods or solutions did the authors propose?

  • The authors proposed the "Medical-AI Team Science Framework" (MedAI-SciTS), consisting of two components:
    1. Theoretical Framework: Comprising four stages (Foundation, SciTS Activation, Co-Design, and Validation), it clarifies the goals, challenges, and solutions for interdisciplinary collaboration.
    2. Practical Toolkit: Offering 12 core tools to support terminology translation, collaborative design, resource management, and real-time feedback.

• What are the innovative aspects of this solution?

  • By combining SciTS theory with user-centered approaches, the framework focuses on addressing specific issues in medical-AI interdisciplinary collaboration.
  • The toolkit integrates AI-enhanced features (e.g., personalized terminology analogies), an agile collaborative design platform, and a comprehensive resource management system, providing tailored support for medical and AI teams.

• What are the implementation steps and key technologies used?

  • The theoretical framework was designed through a literature review and formative research with 12 interdisciplinary experts.
  • Using platforms like Notion and Figma, 12 functional modules were developed, including a glossary, an agile collaboration platform, and a real-time progress management system.
  • Case Study: Participants underwent the Foundation, SciTS Activation, Co-Design, and Validation stages to evaluate the feasibility and effectiveness of the toolkit.

Research Outcomes

• What specific outcomes were achieved?

  • The tools demonstrated high usage frequency, with significant engagement around core functionalities such as the agile collaboration platform, resource management system, and glossary.
  • Successfully addressed eight major collaboration challenges across different stages, including reducing terminology barriers, improving communication flow, and fostering interdisciplinary collaboration.
  • The AI model for automatic adrenal CT image segmentation achieved a Dice coefficient of 87% and processed 702 cases within one week.

• What advantages does it have compared to existing solutions?

  • The systematic interdisciplinary collaboration framework significantly improved goal alignment and team communication efficiency.
  • The modular design of the tools offers flexibility, allowing customization based on team size and technical background.
  • The introduction of personalized AI features excelled in reducing terminology barriers and other challenges.

• What were the experimental or evaluation results?

  • Participants reported significantly improved collaboration satisfaction, including trust (increased from 3 to 6 points), motivation, and overall satisfaction, which reached a final score of 100%.
  • The case study demonstrated that the team co-created more efficient and contextually applicable AI solutions.

• Limitations and Future Directions

  • The small sample size and short study duration may limit the assessment of long-term sustainability.
  • The lack of a comparative baseline necessitates direct comparisons between MedAI-SciTS and other tools or methods.
  • Future directions include incorporating institutional support and dedicated funding factors, as well as developing diagnostic tools for collaborative environments to support larger-scale projects.

Conclusion

MedAI-SciTS provides a comprehensive and practical solution for interdisciplinary collaboration between the medical and AI fields, both theoretically and practically, and its effectiveness has been validated through case studies. Future optimization and expansion will further enhance its applicability and impact across broader domains.