To compare the performance of the DeepSeek model with experienced ICU clinicians in responding to inquiries from patient-family members and evaluate the impact of AI-assisted communication on patient-family satisfaction, specifically focusing on the measurement methods used.
Key Findings:
DeepSeek outperformed clinicians in information completeness, clarity, empathy, and patient-family satisfaction (p < 0.05), indicating significant advantages in communication.
66.7% of family members in the DeepSeek group reported satisfaction with the responses, highlighting the model's effectiveness.
No significant differences were found in medical accuracy or safety assessment between DeepSeek and clinician responses, suggesting comparable reliability.
Interpretation:
DeepSeek provided more complete, clear, and empathetic responses than clinicians, leading to higher perceived family satisfaction, which may influence future communication strategies in ICUs.
Limitations:
The study was a standardized evaluation rather than a real-world clinical communication intervention, limiting its applicability.
The findings are based on simulated responses and may not reflect actual clinical scenarios, which could affect the reliability of the results.
Conclusion:
DeepSeek may support ICU communication, but clinician oversight is essential for individualized discussions. Further real-world studies are needed to confirm safety, reliability, and the model's applicability in clinical settings.
