Siddhartha Jonnalagadda

Text summarization in the biomedical domain

OBJECTIVE The amount of information for clinicians and clinical researchers is growing exponentially. Text summarization reduces information as an attempt to enable users to find and understand relevant source texts more quickly and effortlessly. In recent years, substantial research has been conducted to develop and evaluate various summarization techniques in the biomedical domain. The goal of this study was to systematically review recent published research on summarization of textual documents in the biomedical domain. MATERIALS AND METHODS MEDLINE (2000 to October 2013), IEEE Digital Library, and the ACM digital library were searched. Investigators independently screened and abstracted studies that examined text summarization techniques in the biomedical domain. Information is derived from selected articles on five dimensions: input, purpose, output, method and evaluation. RESULTS Of 10,786 studies retrieved, 34 (0.3%) met the inclusion criteria. Natural language processing (17; 50%) and a hybrid technique comprising of statistical, Natural language processing and machine learning (15; 44%) were the most common summarization approaches. Most studies (28; 82%) conducted an intrinsic evaluation. DISCUSSION This is the first systematic review of text summarization in the biomedical domain. The study identified research gaps and provides recommendations for guiding future research on biomedical text summarization. CONCLUSION Recent research has focused on a hybrid technique comprising statistical, language processing and machine learning techniques. Further research is needed on the application and evaluation of text summarization in real research or patient care settings.

Efficient Extraction of Protein-Protein Interactions from Full-Text Articles

Proteins and their interactions govern virtually all cellular processes, such as regulation, signaling, metabolism, and structure. Most experimental findings pertaining to such interactions are discussed in research papers, which, in turn, get curated by protein interaction databases. Authors, editors, and publishers benefit from efforts to alleviate the tasks of searching for relevant papers, evidence for physical interactions, and proper identifiers for each protein involved. The BioCreative II.5 community challenge addressed these tasks in a competition-style assessment to evaluate and compare different methodologies, to make aware of the increasing accuracy of automated methods, and to guide future implementations. In this paper, we present our approaches for protein-named entity recognition, including normalization, and for extraction of protein-protein interactions from full text. Our overall goal is to identify efficient individual components, and we compare various compositions to handle a single full-text article in between 10 seconds and 2 minutes. We propose strategies to transfer document-level annotations to the sentence-level, which allows for the creation of a more fine-grained training corpus; we use this corpus to automatically derive around 5,000 patterns. We rank sentences by relevance to the task of finding novel interactions with physical evidence, using a sentence classifier built from this training corpus. Heuristics for paraphrasing sentences help to further remove unnecessary information that might interfere with patterns, such as additional adjectives, clauses, or bracketed expressions. In BioCreative II.5, we achieved an f-score of 22 percent for finding protein interactions, and 43 percent for mapping proteins to UniProt IDs; disregarding species, f-scores are 30 percent and 55 percent, respectively. On average, our best-performing setup required around 2 minutes per full text. All data and pattern sets as well as Java classes that extend third-party software are available as supplementary information (see Appendix).

Towards Effective Sentence Simplification for Automatic Processing of Biomedical Text

The complexity of sentences characteristic to biomedical articles poses a challenge to natural language parsers, which are typically trained on large-scale corpora of non-technical text. We propose a text simplification process, bioSimplify, that seeks to reduce the complexity of sentences in biomedical abstracts in order to improve the performance of syntactic parsers on the processed sentences. Syntactic parsing is typically one of the first steps in a text mining pipeline. Thus, any improvement in performance would have a ripple effect over all processing steps. We evaluated our method using a corpus of biomedical sentences annotated with syntactic links. Our empirical results show an improvement of 2.90% for the Charniak-McClosky parser and of 4.23% for the Link Grammar parser when processing simplified sentences rather than the original sentences in the corpus.

Comprehensive temporal information detection from clinical text: medical events, time, and TLINK identification

BACKGROUND Temporal information detection systems have been developed by the Mayo Clinic for the 2012 i2b2 Natural Language Processing Challenge. OBJECTIVE To construct automated systems for EVENT/TIMEX3 extraction and temporal link (TLINK) identification from clinical text. MATERIALS AND METHODS The i2b2 organizers provided 190 annotated discharge summaries as the training set and 120 discharge summaries as the test set. Our Event system used a conditional random field classifier with a variety of features including lexical information, natural language elements, and medical ontology. The TIMEX3 system employed a rule-based method using regular expression pattern match and systematic reasoning to determine normalized values. The TLINK system employed both rule-based reasoning and machine learning. All three systems were built in an Apache Unstructured Information Management Architecture framework. RESULTS Our TIMEX3 system performed the best (F-measure of 0.900, value accuracy 0.731) among the challenge teams. The Event system produced an F-measure of 0.870, and the TLINK system an F-measure of 0.537. CONCLUSIONS Our TIMEX3 system demonstrated good capability of regular expression rules to extract and normalize time information. Event and TLINK machine learning systems required well-defined feature sets to perform well. We could also leverage expert knowledge as part of the machine learning features to further improve TLINK identification performance.

Coreference analysis in clinical notes: a multi-pass sieve with alternate anaphora resolution modules.

OBJECTIVE This paper describes the coreference resolution system submitted by Mayo Clinic for the 2011 i2b2/VA/Cincinnati shared task Track 1C. The goal of the task was to construct a system that links the markables corresponding to the same entity. MATERIALS AND METHODS The task organizers provided progress notes and discharge summaries that were annotated with the markables of treatment, problem, test, person, and pronoun. We used a multi-pass sieve algorithm that applies deterministic rules in the order of preciseness and simultaneously gathers information about the entities in the documents. Our system, MedCoref, also uses a state-of-the-art machine learning framework as an alternative to the final, rule-based pronoun resolution sieve. RESULTS The best system that uses a multi-pass sieve has an overall score of 0.836 (average of B(3), MUC, Blanc, and CEAF F score) for the training set and 0.843 for the test set. DISCUSSION A supervised machine learning system that typically uses a single function to find coreferents cannot accommodate irregularities encountered in data especially given the insufficient number of examples. On the other hand, a completely deterministic system could lead to a decrease in recall (sensitivity) when the rules are not exhaustive. The sieve-based framework allows one to combine reliable machine learning components with rules designed by experts. CONCLUSION Using relatively simple rules, part-of-speech information, and semantic type properties, an effective coreference resolution system could be designed. The source code of the system described is available at https://sourceforge.net/projects/ohnlp/files/MedCoref.

Pooling annotated corpora for clinical concept extraction

BackgroundThe availability of annotated corpora has facilitated the application of machine learning algorithms to concept extraction from clinical notes. However, high expenditure and labor are required for creating the annotations. A potential alternative is to reuse existing corpora from other institutions by pooling with local corpora, for training machine taggers. In this paper we have investigated the latter approach by pooling corpora from 2010 i2b2/VA NLP challenge and Mayo Clinic Rochester, to evaluate taggers for recognition of medical problems. The corpora were annotated for medical problems, but with different guidelines. The taggers were constructed using an existing tagging system MedTagger that consisted of dictionary lookup, part of speech (POS) tagging and machine learning for named entity prediction and concept extraction. We hope that our current work will be a useful case study for facilitating reuse of annotated corpora across institutions.ResultsWe found that pooling was effective when the size of the local corpus was small and after some of the guideline differences were reconciled. The benefits of pooling, however, diminished as more locally annotated documents were included in the training data. We examined the annotation guidelines to identify factors that determine the effect of pooling.ConclusionsThe effectiveness of pooling corpora, is dependent on several factors, which include compatibility of annotation guidelines, distribution of report types and size of local and foreign corpora. Simple methods to rectify some of the guideline differences can facilitate pooling. Our findings need to be confirmed with further studies on different corpora. To facilitate the pooling and reuse of annotated corpora, we suggest that – i) the NLP community should develop a standard annotation guideline that addresses the potential areas of guideline differences that are partly identified in this paper; ii) corpora should be annotated with a two-pass method that focuses first on concept recognition, followed by normalization to existing ontologies; and iii) metadata such as type of the report should be created during the annotation process.

A distributional semantics approach to simultaneous recognition of multiple classes of named entities

Named Entity Recognition and Classification is being studied for last two decades. Since semantic features take huge amount of training time and are slow in inference, the existing tools apply features and rules mainly at the word level or use lexicons. Recent advances in distributional semantics allow us to efficiently create paradigmatic models that encode word order. We used Sahlgren et als permutation-based variant of the Random Indexing model to create a scalable and efficient system to simultaneously recognize multiple entity classes mentioned in natural language, which is validated on the GENIA corpus which has annotations for 46 biomedical entity classes and supports nested entities. Using distributional semantics features only, it achieves an overall micro-averaged F-measure of 67.3% based on fragment matching with performance ranging from 7.4% for “DNA substructure” to 80.7% for “Bioentity”.

Discovering opinion leaders for medical topics using news articles

BackgroundRapid identification of subject experts for medical topics helps in improving the implementation of discoveries by speeding the time to market drugs and aiding in clinical trial recruitment, etc. Identifying such people who influence opinion through social network analysis is gaining prominence. In this work, we explore how to combine named entity recognition from unstructured news articles with social network analysis to discover opinion leaders for a given medical topic.MethodsWe employed a Conditional Random Field algorithm to extract three categories of entities from health-related new articles: Person, Organization and Location. We used the latter two to disambiguate polysemy and synonymy for the person names, used simple rules to identify the subject experts, and then applied social network analysis techniques to discover the opinion leaders among them based on their media presence. A network was created by linking each pair of subject experts who are mentioned together in an article. The social network analysis metrics (including centrality metrics such as Betweenness, Closeness, Degree and Eigenvector) are used for ranking the subject experts based on their power in information flow.ResultsWe extracted 734,204 person mentions from 147,528 news articles related to obesity from January 1, 2007 through July 22, 2010. Of these, 147,879 mentions have been marked as subject experts. The F-score of extracting person names is 88.5%. More than 80% of the subject experts who rank among top 20 in at least one of the metrics could be considered as opinion leaders in obesity.ConclusionThe analysis of the network of subject experts with media presence revealed that an opinion leader might have fewer mentions in the news articles, but a high network centrality measure and vice-versa. Betweenness, Closeness and Degree centrality measures were shown to supplement frequency counts in the task of finding subject experts. Further, opinion leaders missed in scientific publication network analysis could be retrieved from news articles.

Website Sharing in Online Health Communities: A Descriptive Analysis

Background An increasing number of people visit online health communities to seek health information. In these communities, people share experiences and information with others, often complemented with links to different websites. Understanding how people share websites can help us understand patients’ needs in online health communities and improve how peer patients share health information online. Objective Our goal was to understand (1) what kinds of websites are shared, (2) information quality of the shared websites, (3) who shares websites, (4) community differences in website-sharing behavior, and (5) the contexts in which patients share websites. We aimed to find practical applications and implications of website-sharing practices in online health communities. Methods We used regular expressions to extract URLs from 10 WebMD online health communities. We then categorized the URLs based on their top-level domains. We counted the number of trust codes (eg, accredited agencies’ formal evaluation and PubMed authors’ institutions) for each website to assess information quality. We used descriptive statistics to determine website-sharing activities. To understand the context of the URL being discussed, we conducted a simple random selection of 5 threads that contained at least one post with URLs from each community. Gathering all other posts in these threads resulted in 387 posts for open coding analysis with the goal of understanding motivations and situations in which website sharing occurred. Results We extracted a total of 25,448 websites. The majority of the shared websites were .com (59.16%, 15,056/25,448) and WebMD internal (23.2%, 5905/25,448) websites; the least shared websites were social media websites (0.15%, 39/25,448). High-posting community members and moderators posted more websites with trust codes than low-posting community members did. The heart disease community had the highest percentage of websites containing trust codes compared to other communities. Members used websites to disseminate information, supportive evidence, resources for social support, and other ways to communicate. Conclusions Online health communities can be used as important health care information resources for patients and caregivers. Our findings inform patients’ health information–sharing activities. This information assists health care providers, informaticians, and online health information entrepreneurs and developers in helping patients and caregivers make informed choices.

Using Empirically Constructed Lexical Resources for Named Entity Recognition

Because of privacy concerns and the expense involved in creating an annotated corpus, the existing small-annotated corpora might not have sufficient examples for learning to statistically extract all the named-entities precisely. In this work, we evaluate what value may lie in automatically generated features based on distributional semantics when using machine-learning named entity recognition (NER). The features we generated and experimented with include n-nearest words, support vector machine (SVM)-regions, and term clustering, all of which are considered distributional semantic features. The addition of the n-nearest words feature resulted in a greater increase in F-score than by using a manually constructed lexicon to a baseline system. Although the need for relatively small-annotated corpora for retraining is not obviated, lexicons empirically derived from unannotated text can not only supplement manually created lexicons, but also replace them. This phenomenon is observed in extracting concepts from both biomedical literature and clinical notes.