评估人工智能为脊柱手术生成的患者教育材料:在ChatGPT和deepseek模型中对可读性和DISCERN质量的比较分析
Mi Zhou1, Yun Pan2, Yuye Zhang3
1Allied Health & Human Performance, University of South Australia, Adelaide, Australia.
International journal of medical informatics
|March 19, 2025
概括
与其他AI模型相比,DeepSeek-R1和ChatGPT-4o为脊柱手术提供了更易于阅读的患者教育材料. 虽然所有模型都提供了"公平"的质量信息,但这两种模型有望改善患者对医疗保健的理解和参与.
科学领域:
- 医疗保健中的人工智能
- 医疗信息的自然语言处理.
- 患者教育 材料开发 患者教育
背景情况:
- 以患者为中心的健康信息对于明智的决策至关重要,但在线资源往往缺乏质量和可读性,特别是在复杂的外科主题上.
- 越来越多的人工智能用于医疗信息,需要对其可靠性和可访问性进行评估,对于具有不同健康素养水平的患者来说是如此.
研究的目的:
- 评估四种自然语言处理 (NLP) 模型 (ChatGPT-4o,ChatGPT-o3 mini,DeepSeek-V3,DeepSeek-R1) 用于生成关于常见脊柱手术的患者教育材料.
- 通过使用DISCERN分数和使用Flesch-Kincaid指数来评估AI生成的外科内容的信息质量和可读性.
主要方法:
- 四个NLP模型的任务是为腰椎切除术,脊柱融合和解压膜切除术创建患者教育材料.
- 信息质量通过DISCERN分数来衡量,可读性通过弗莱什-金凯德等级水平 (FKGL) 来衡量.
主要成果:
- 深度搜索-R1显示了最高的可读性 (FKGL 7.2-9.0),其次是ChatGPT-4o. 聊天GPT-o3的可读性最低 (FKGL> 10.4).
- 所有的人工智能模型都获得了"公平"的质量评级 (DISCERN < 60),主要是由于缺乏引用引用.
结论:
- 当前的人工智能模型提供"公平"的质量患者教育材料,突出需要改进引用和个性化.
- 建议使用DeepSeek-R1和ChatGPT-4o来生成更易于阅读的外科信息,这可能会提高患者的参与度和结果.
相关概念视频
01:13Patient-centered Care
1.9K
Patient-centered care involves delivering care beyond inpatient hospitalization. Reflective practice can enhance a patient-centered approach. Reflective practice is a process of reasoning that considers all aspects of the present situation, including practicalities, learning from personal practice, and consideration of patient needs. Patients appreciate care decisions made while considering their input. Involving the patient in their care provides the patient with a sense of contribution rather...
1.9K
01:11Local Anesthetics: Clinical Application as Spinal Anesthesia
567
Spinal anesthetics are given during lower abdomen and limb surgeries to block sensory and motor neurons. They are administered in the mid to low lumbar regions, primarily acting on the cauda equina's nerve roots. The blockade level depends on the local anesthetic (LA) concentration. Usually, low LA concentrations are sufficient to block sensory fibers, while only high LA concentrations block motor fibers. Other factors like injection volume and speed, the patient's posture, and the drug...
567