Advancing Patient-Centered Shared Decision-Making with AI Systems for Older Adult Cancer Patients

Title of the Paper

Advancing Patient-Centered Shared Decision-Making with AI Systems for Older Adult Cancer Patients

Paper Information

• Subject Area: Application of Artificial Intelligence in Shared Decision-Making for Older Adult Cancer Patients
• Keywords: Shared Decision-Making, Clinical Decision-Making, Older Adults, Cancer Care, Risk Communication, Artificial Intelligence, Health Education, Patient Engagement, Healthcare Efficiency

Research Background and Problem Statement

• Identified Problems/Challenges:
  • Medical decision-making is complex and time-sensitive.
  • Older adult cancer patients often have lower health literacy, reduced memory, and limited information processing abilities, making it difficult for them to fully participate in decision-making.
  • Physicians face constraints in patient communication due to staff shortages and time pressures.
• Significance:
  • Effective Shared Decision-Making (SDM) practices can significantly enhance patient satisfaction and improve treatment outcomes.
  • Emerging technologies such as Artificial Intelligence (AI) have the potential to bridge the communication gap between patients and physicians.
• Research Motivation and Related Work:
  • Through a foundational literature review, the authors identified design and implementation issues in existing SDM systems.
  • While AI models are widely applied in clinical settings, their use for cancer decision-making in older patients remains largely theoretical, with limited practical implementation.

Proposed Solution

• Core Approach:
  • Introduce a patient-centered SDM AI system (i-SDM) specifically designed for older adult cancer patients.
  • The system includes features such as treatment option comparisons, survival predictions, risk assessments, and other patient-relevant factors.
  • Employ a phased interface to gradually present information to patients and physicians, fostering effective dialogue.
• Innovative Aspects:
  • Utilize AI technologies (e.g., GPT-4 model and LightGBM algorithm) to generate personalized medical information and explanations.
  • Integrate system design with interaction habits of older adults, offering clear and intuitive graphical presentations.
  • Emphasize educational functionality in clinical settings to help patients deeply understand treatment pathways.
• Implementation Steps and Techniques:
  1. Phase One: Conduct interviews with patients and physicians to gather key SDM factors for prototype design.
  2. Phase Two: Develop the i-SDM system, integrating survival prediction models and generating comprehensible medical language.
  3. Testing and Evaluation: Assess system acceptability and effectiveness through usability interviews with patients and physicians.

Research Outcomes

• Specific Results:
  • The i-SDM system significantly improved older patients' understanding of complex treatment options, reducing the communication burden on clinicians.
  • The average patient interaction time was 23 minutes, indicating high acceptance and engagement with the system.
• Advantages Compared to Existing Solutions:
  • Reduced physician workload: Clear presentation of complex information saves communication time.
  • Enhanced patient empowerment: Patients can actively participate in treatment decisions, improving satisfaction and adherence.
  • Improved health education: Helps older adults better understand the critical impacts of cancer treatments.
• Experimental Results:
  • Participants generally provided positive feedback on the interface and information design, noting its effectiveness in disseminating cancer treatment knowledge.
  • A total of 316 interaction tags were collected, with approximately 46% being positive responses, indicating participants' approval of the system's functionality.
• Limitations and Future Directions:
  • The sample size was small and concentrated on highly educated, high-income groups, lacking validation for diverse populations (e.g., low-income or low-health-literacy patients).
  • The model's predictive capability for complex comorbidities was limited and requires further optimization.
  • Future research could explore customized solutions for different types of cancer and validate efficacy in real clinical workflows.

Conclusion

This study demonstrates the feasibility and potential of AI in facilitating Shared Decision-Making for older adult cancer patients. The i-SDM system enhances patient engagement through education and information optimization while alleviating physicians' communication burden. It provides a reliable foundation and design direction for future applications of health technologies in older patient populations. Future work should focus on testing the system with broader and more diverse user groups and improving AI model performance for complex cases.