Wang Mini Han, Jiang Xudong, Zeng Peijin, Li Xinyue, Chong Kelvin Kam-Lung, Hou Guanghui, Fang Xiaoxiao, Yu Yang, Yu Xiangrong, Fang Junbin, Pan Yi
Zhuhai People's Hospital (The Affiliated Hospital of Beijing Institute of Technology, Zhuhai Clinical Medical College of Jinan University), Zhuhai, China.
Department of Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Kowloon, Hong Kong SAR, China.
Front Artif Intell. 2025 Feb 13;8:1517918. doi: 10.3389/frai.2025.1517918. eCollection 2025.
The rapid evolution of the Internet of Things (IoT) and Artificial Intelligence (AI) has opened new possibilities for public healthcare. Effective integration of these technologies is essential to ensure precise and efficient healthcare delivery. This study explores the application of IoT-enabled, AI-driven systems for detecting and managing Dry Eye Disease (DED), emphasizing the use of prompt engineering to enhance system performance.
A specialized prompt mechanism was developed utilizing OpenAI GPT-4.0 and ERNIE Bot-4.0 APIs to assess the urgency of medical attention based on 5,747 simulated patient complaints. A Bidirectional Encoder Representations from Transformers (BERT) machine learning model was employed for text classification to differentiate urgent from non-urgent cases. User satisfaction was evaluated through composite scores derived from Service Experiences (SE) and Medical Quality (MQ) assessments.
The comparison between prompted and non-prompted queries revealed a significant accuracy increase from 80.1% to 99.6%. However, this improvement was accompanied by a notable rise in response time, resulting in a decrease in SE scores (95.5 to 84.7) but a substantial increase in MQ satisfaction (73.4 to 96.7). These findings indicate a trade-off between accuracy and user satisfaction.
The study highlights the critical role of prompt engineering in improving AI-based healthcare services. While enhanced accuracy is achievable, careful attention must be given to balancing response time and user satisfaction. Future research should optimize prompt structures, explore dynamic prompting approaches, and prioritize real-time evaluations to address the identified challenges and maximize the potential of IoT-integrated AI systems in medical applications.
物联网（IoT）和人工智能（AI）的快速发展为公共医疗保健带来了新的可能性。有效整合这些技术对于确保精确高效的医疗服务至关重要。本研究探讨了物联网支持、人工智能驱动的系统在检测和管理干眼病（DED）中的应用，强调使用提示工程来提高系统性能。
利用OpenAI GPT - 4.0和ERNIE Bot - 4.0 API开发了一种专门的提示机制，以根据5747条模拟患者投诉评估医疗关注的紧迫性。采用来自变换器的双向编码器表示（BERT）机器学习模型进行文本分类，以区分紧急和非紧急情况。通过服务体验（SE）和医疗质量（MQ）评估得出的综合得分来评估用户满意度。
提示查询和未提示查询之间的比较显示，准确率从80.1%显著提高到99.6%。然而，这种提高伴随着响应时间的显著增加，导致SE得分下降（从95.5降至84.7），但MQ满意度大幅提高（从73.4升至96.7）。这些发现表明在准确性和用户满意度之间存在权衡。
该研究强调了提示工程在改善基于人工智能的医疗服务中的关键作用。虽然可以实现更高的准确性，但必须谨慎注意平衡响应时间和用户满意度。未来的研究应优化提示结构，探索动态提示方法，并优先进行实时评估，以应对已识别的挑战，并最大限度地发挥物联网集成人工智能系统在医疗应用中的潜力。
