Aims: The purpose of this study was to clarify the thought processes of nurses in performing nursing assessment. Methods: The participants comprised 20 nurses working in a surgery ward. Patient information on a case, ...Aims: The purpose of this study was to clarify the thought processes of nurses in performing nursing assessment. Methods: The participants comprised 20 nurses working in a surgery ward. Patient information on a case, including presenting illness, vital signs, and other findings from admission until 09:00 on the day after surgery, was shown to the participants. After reading the case report, the nurses presented their assessments. Based on these assessments, nursing problems, patient strengths, and patient information were identified. Nursing problems and patient strengths were described by various words and sentences, and were classified according to similar content. Results: The number of nursing problems ranged from 1 to 8 and patient strengths from 0 to 6 for each nurse. The mean number of nursing problems was 4.7 ± 1.8, and the mean number of patient strengths was 2.2 ± 1.4. The main nursing problems were respiratory complications, postoperative wound pain, and anxiety, and the main patient strength was family cooperation. Patient information as evidence of respiratory complications included history of smoking, chest radiography results, postoperative vital signs, sputum color and properties. Patient information as evidence of postoperative wound pain included complaints of pain, epidural anesthesia, use of patient-controlled anesthesia and its effect. Patient information indicating family cooperation included family structure, preoperative visits by family, and presence of family while providing informed consent. Significant differences were seen in the number of nursing problems and patient strengths according to cognitive style. Conclusions: Postoperative complications were the nursing problems most commonly extracted by nurses. To clarify nursing problems and patient strengths, the nurses made assessments on the basis of information such as patient complaints, vital signs, and test results. However, extracted nursing problems and patient strengths were diverse, suggesting that nursing problems and patient strengths as determined by nurses differed between individual nurses.展开更多
针对再制造知识多环节和歧义性等特点导致的传统抽取方法标注数据依赖性强、多跳关系解析能力不足等问题,提出一种基于大语言模型链式提示词的再制造工艺知识多粒度抽取方法,通过融合提示词工程与思维链推理,结合语义对齐机制,利用大语...针对再制造知识多环节和歧义性等特点导致的传统抽取方法标注数据依赖性强、多跳关系解析能力不足等问题,提出一种基于大语言模型链式提示词的再制造工艺知识多粒度抽取方法,通过融合提示词工程与思维链推理,结合语义对齐机制,利用大语言模型(large language model,LLM)实现粗粒度到细粒度知识的精准提取。首先,基于提示词工程引导LLM完成初始知识的粗粒度抽取,定位再制造工艺核心实体;其次,设计思维链推理框架,驱动LLMs解析实体间复杂逻辑关系,并通过余弦相似度实现异构语义对齐,提升细粒度知识的语义一致性与匹配精度。试验结果表明,链式提示词法的F1分数达88.0%,较传统方法提升超30%,且多跳关系覆盖率达89.2%,有效解决了传统技术对标注数据的依赖问题。展开更多
文摘Aims: The purpose of this study was to clarify the thought processes of nurses in performing nursing assessment. Methods: The participants comprised 20 nurses working in a surgery ward. Patient information on a case, including presenting illness, vital signs, and other findings from admission until 09:00 on the day after surgery, was shown to the participants. After reading the case report, the nurses presented their assessments. Based on these assessments, nursing problems, patient strengths, and patient information were identified. Nursing problems and patient strengths were described by various words and sentences, and were classified according to similar content. Results: The number of nursing problems ranged from 1 to 8 and patient strengths from 0 to 6 for each nurse. The mean number of nursing problems was 4.7 ± 1.8, and the mean number of patient strengths was 2.2 ± 1.4. The main nursing problems were respiratory complications, postoperative wound pain, and anxiety, and the main patient strength was family cooperation. Patient information as evidence of respiratory complications included history of smoking, chest radiography results, postoperative vital signs, sputum color and properties. Patient information as evidence of postoperative wound pain included complaints of pain, epidural anesthesia, use of patient-controlled anesthesia and its effect. Patient information indicating family cooperation included family structure, preoperative visits by family, and presence of family while providing informed consent. Significant differences were seen in the number of nursing problems and patient strengths according to cognitive style. Conclusions: Postoperative complications were the nursing problems most commonly extracted by nurses. To clarify nursing problems and patient strengths, the nurses made assessments on the basis of information such as patient complaints, vital signs, and test results. However, extracted nursing problems and patient strengths were diverse, suggesting that nursing problems and patient strengths as determined by nurses differed between individual nurses.
文摘针对再制造知识多环节和歧义性等特点导致的传统抽取方法标注数据依赖性强、多跳关系解析能力不足等问题,提出一种基于大语言模型链式提示词的再制造工艺知识多粒度抽取方法,通过融合提示词工程与思维链推理,结合语义对齐机制,利用大语言模型(large language model,LLM)实现粗粒度到细粒度知识的精准提取。首先,基于提示词工程引导LLM完成初始知识的粗粒度抽取,定位再制造工艺核心实体;其次,设计思维链推理框架,驱动LLMs解析实体间复杂逻辑关系,并通过余弦相似度实现异构语义对齐,提升细粒度知识的语义一致性与匹配精度。试验结果表明,链式提示词法的F1分数达88.0%,较传统方法提升超30%,且多跳关系覆盖率达89.2%,有效解决了传统技术对标注数据的依赖问题。
