基于词频统计法的医学新闻自动受控标引

分析了医学新闻信息利用的必要性及自动标引的发展现状,提出一种医学新闻文本自动受控标引方法,即以分词词表为基础词表,引入汉化MeSH词表建立标引词表,对中文医学新闻文本进行分词、词频统计和排序,过滤掉不在主题词表中的高频词后,选取词频最高的5个MeSH主题词用作标引词。     

标引关键词须知

标引关键词应针对文章的重点内容,请尽量使用最新版美国国立医学图书馆编辑的《IndexMedicus》中的医学主题词表（MeSH）内所列的词。     

标引关键词须知

标引关键词应针对文章的重点内容,请尽量使用最新版美国国立医学图书馆编辑的《IndexMedicus》中的医学主题词表（MeSH）内所列的词。如果最新的MeSH中还无相应的词,处理办法有：①可选用直接相关的几个关键词组配。     

标引关键词须知

标引关键词应针对文章的重点内容,请尽量使用最新版美国国市医学图书馆编辑的《Index Medicus》中的医学主题词表（MeSH）内所列的词。如果最新的MeSH中还无相应的词,处理办法有：①可选用直接相关的几个关键词组配。     

标引关键词须知

标引关键词应针对文章的重点内容,请尽量使用最新版美国国立医学图书馆编辑的《Index Medicus》中的医学主题词表（MeSH）内所列的词。     

标引关键词须知

标引关键词应针对文章的重点内容，请尽量使用最新版美国国立医学图书馆编辑的《Index Medicus》中的医学主题词表（MeSH）内所列的词。     

标引关键词须知

标引关键词应针对文章的重点内容,请尽量使用最新版美国国立医学图书馆编辑的《Index Medicus》中的医学主题词表（MeSH）内所列的词。     

关于关键词标引的要求

每篇论文需标引3—5个关键词,请尽量使用美国国立医学图书馆编著的最新版《IndexMedicus》中医学主题词表（MeSH）内所列的词。如果MeSH表中尚无相对应的词,可选用直接相关的几个主题词进行组配;     

标引关键词须知

标引关键词应针对文章的重点内容，请尽量使用最新版关同圈立医学图书馆编辑的《Index Medicus》中的医学主题词表（MeSH）内所列的词。如果最新的MeSH中还无相应的词，处理办法有：①可选用直接相关的几个关键词组配。②如果无法组配，可选用最直接的上位关键词。     

标引关键词须知

标引关键词应针对文章的重点内容，请尽量使用最新版美国国立医学图书馆编辑的《IndexMedicus））中的医学主题词表MeSH内所列的词。如果最新的MeSH中还无相应的词，处理办法有：（1）可选用直接相关的几个关键词组配；（2）如果无法组配，可选用最直接的上位关键词。     

Comparison of search strategies on CD Plus/MEDLINE.

OBJECTIVE: To compare two strategies for searching MEDLINE using the CD Plus/MEDLINE program on compact disc. DESIGN: Comparison study. INTERVENTIONS: Two search strategies were designed and executed for each of two topics (patient recruitment to clinical trials and attitudes of patients, the public and health care professionals toward clinical trials). Strategy A: searches based on key words selected from the medical subject heading (MeSH) tree structure. Strategy B: searches based on MeSH terms most frequently used to index a known set of relevant articles. Defined search restrictions were then applied. The effects of the restrictions on the absolute number of citations retrieved and on the proportion of relevant citations were assessed. OUTCOME MEASURES: Number of articles retrieved, number of relevant articles, precision and recall of each search strategy and overlap between strategies. MAIN RESULTS: Strategy A produced more citations than strategy B (recruitment 147 v. 38, attitude 366 v. 57) but had more inappropriate citations (recruitment 75 v. 17, attitude 265 v. 25). Both strategies produced 73 relevant recruitment citations and 101 relevant attitude citations. In the recruitment search although the precision did not differ significantly between strategies A and B the difference in recall was significant (98.6% v. 28.8% respectively, p less than 0.0001). In the attitude search strategy A had a lower precision than strategy B (27.6% v. 56.1%, p less than 0.0001) but a much higher recall (100% v. 31.7%, p less than 0.0001). CONCLUSIONS: Strategy A would be more valuable to researchers doing extensive reviews, whereas strategy B would be useful for the busy clinician who simply wants a few appropriate references quickly and is willing to sacrifice comprehensive retrieval in the interest of efficiency.     

利用MeSH组配规则自动抽取表达特定语义关系句子的探索

目的：利用MeSH组配规则自动抽取文摘中表达特定语义关系的句子,为制定自然语言处理关系抽取模板以及句子水平的信息检索提供依据。方法：根据主题词组配规则,使用python语言从文摘数据中匹配出含有特定MeSH主题词概念的候选关系句,从中抽取出以描述概念间关系的短语或句子。邀请专家对100条候选关系句进行概念间语义关系人工标注,将得到的语义关系三元组作为评价金标准,与自动抽取出的概念间关系进行对比分析。将自动抽取的结果加以整理形成特定概念之间的语义关系表达。结果：对大量的自然文本句进行句法分析,批量识别出2个特定概念间语义关系抽取方法的准确率为87%,召回率为62%,F1=71.8%。结论：利用MeSH组配规则抽取表达特定语义关系句子的方法具有较高的准确率与召回率,对生物医学文本理解及医学知识发现等具有借鉴意义。     

用户对医学术语与用户词的认知差异

本文从用户健康信息学研究角度出发,对健康词汇熟悉度的测评方法进行了研究,并以此为基础研究用户对医学术语和用户词的认知差异。首先从用户搜索提问中获取用户词,从而形成用户词-医学术语概念对,设计评判词汇熟悉度的测评工具和调查问卷。然后,发放调查问卷532份(有效问卷503份),分析用户对医学术语和用户词的熟悉度差异,结果表明用户对医学术语和用户词存在着认知差异(P=0.003)。健康词汇熟悉度测评工具的建立,是中文健康词表构建的关键一环,用户的认知差异研究则是中文用户健康词表的重要研究前提和基础。     

In Search of Controlled Evidence for Health Care Quality Improvement

The purpose of this study was to measure the efficiency of simple searches in retrieving controlled evidence about specific primary health care quality improvement interventions and their effects. Searches were conducted to retrieve evidence on seven interventions and seven effect variables. Specific words and the closest Medical Subject Headings (MeSH) recommended by professional librarians were used to search the MEDLINE database. Searches were restricted to the MeSH publication type “randomized controlled trial.” Two reviewers independently judged retrieved citations for relevancy to the selected interventions and effects. In selecting MeSH terms, the average agreement among librarians was 64.3% (±26.1) for interventions and 57.1% (±19.9) for effects. Analysis of the 755 retrieved reports showed that MeSH term searches had an overall recall rate of 58% while the same rate for textword searches was significantly lower (11%, p < .001). The difference in overall precision rates was nonsignificant (26% versus 33%, p = .15). In the group of MeSH searches, overall precision and recall was significantly lower for effects than for interventions (12% versus 52%, p < .001 and 41% versus 69%, p < .001). Two textwords appeared in more than 25% of the benchmark collection: reminder (25.7%) and cost (25.0%). The results of this study indicate that information needs for health care quality improvement cannot be met by simple literature searches. Certain MeSH terms and combinations of textwords yield moderately efficient recall and precision in literature searches for health care quality improvement. Clinicians and physician executives gaining direct access to bibliographic database could probably be better served by structured indexing of critical aspects of randomized controlled clinical trials: design, sample, interventions, and effects.     

Translating guidelines into practice: a systematic review of theoretic concepts, practical experience and research evidence in the adoption of clinical practice guidelines

OBJECTIVE: To recommend effective strategies for implementing clinical practice guidelines (CPGs). DATA SOURCES: The Research and Development Resource Base in Continuing Medical Education, maintained by the University of Toronto, was searched, as was MEDLINE from January 1990 to June 1996, inclusive, with the use of the MeSH heading "practice guidelines" and relevant text words. STUDY SELECTION: Studies of CPG implementation strategies and reviews of such studies were selected. Randomized controlled trials and trials that objectively measured physicians' performance or health care outcomes were emphasized. DATA EXTRACTION: Articles were reviewed to determine the effect of various factors on the adoption of guidelines. DATA SYNTHESIS: The articles showed that CPG dissemination or implementation processes have mixed results. Variables that affect the adoption of guidelines include qualities of the guidelines, characteristics of the health care professional, characteristics of the practice setting, incentives, regulation and patient factors. Specific strategies fell into 2 categories: primary strategies involving mailing or publication of the actual guidelines and secondary interventional strategies to reinforce the guidelines. The interventions were shown to be weak (didactic, traditional continuing medical education and mailings), moderately effective (audit and feedback, especially concurrent, targeted to specific providers and delivered by peers or opinion leaders) and relatively strong (reminder systems, academic detailing and multiple interventions). CONCLUSIONS: The evidence shows serious deficiencies in the adoption of CPGs in practice. Future implementation strategies must overcome this failure through an understanding of the forces and variables influencing practice and through the use of methods that are practice- and community-based rather than didactic.     

A Document Clustering and Ranking System for Exploring MEDLINE Citations

ObjectiveA major problem faced in biomedical informatics involves how best to present information retrieval results. When a single query retrieves many results, simply showing them as a long list often provides poor overview. With a goal of presenting users with reduced sets of relevant citations, this study developed an approach that retrieved and organized MEDLINE citations into different topical groups and prioritized important citations in each group.DesignA text mining system framework for automatic document clustering and ranking organized MEDLINE citations following simple PubMed queries. The system grouped the retrieved citations, ranked the citations in each cluster, and generated a set of keywords and MeSH terms to describe the common theme of each cluster.MeasurementsSeveral possible ranking functions were compared, including citation count per year (CCPY), citation count (CC), and journal impact factor (JIF). We evaluated this framework by identifying as “important” those articles selected by the Surgical Oncology Society.ResultsOur results showed that CCPY outperforms CC and JIF, i.e., CCPY better ranked important articles than did the others. Furthermore, our text clustering and knowledge extraction strategy grouped the retrieval results into informative clusters as revealed by the keywords and MeSH terms extracted from the documents in each cluster.ConclusionsThe text mining system studied effectively integrated text clustering, text summarization, and text ranking and organized MEDLINE retrieval results into different topical groups.     

MeSH增补概念表中的中草药术语使用

增补概念表(Supplementary Concept Records,SCRs)通过与Me SH的映射关系形成了对Me SH主题词有效的补充与扩展。通过Me SH Browser获得SCRs中的相关中草药术语,从国别、单味药、中药提取成分、复方、经方(经典方剂)与自拟方、使用频次等角度进行分析,揭示了中草药相关增补概念的现状及存在的问题,提出了要推进中医药术语规范化及研究人员发文和检索时要使用规范术语等建议。     

Recommending MeSH terms for annotating biomedical articles

Background

Due to the high cost of manual curation of key aspects from the scientific literature, automated methods for assisting this process are greatly desired. Here, we report a novel approach to facilitate MeSH indexing, a challenging task of assigning MeSH terms to MEDLINE citations for their archiving and retrieval.

Methods

Unlike previous methods for automatic MeSH term assignment, we reformulate the indexing task as a ranking problem such that relevant MeSH headings are ranked higher than those irrelevant ones. Specifically, for each document we retrieve 20 neighbor documents, obtain a list of MeSH main headings from neighbors, and rank the MeSH main headings using ListNet–a learning-to-rank algorithm. We trained our algorithm on 200 documents and tested on a previously used benchmark set of 200 documents and a larger dataset of 1000 documents.

Results

Tested on the benchmark dataset, our method achieved a precision of 0.390, recall of 0.712, and mean average precision (MAP) of 0.626. In comparison to the state of the art, we observe statistically significant improvements as large as 39% in MAP (p-value <0.001). Similar significant improvements were also obtained on the larger document set.

Conclusion

Experimental results show that our approach makes the most accurate MeSH predictions to date, which suggests its great potential in making a practical impact on MeSH indexing. Furthermore, as discussed the proposed learning framework is robust and can be adapted to many other similar tasks beyond MeSH indexing in the biomedical domain. All data sets are available at: http://www.ncbi.nlm.nih.gov/CBBresearch/Lu/indexing.