Dataset Information

A modular framework for biomedical concept recognition.

ABSTRACT: BACKGROUND: Concept recognition is an essential task in biomedical information extraction, presenting several complex and unsolved challenges. The development of such solutions is typically performed in an ad-hoc manner or using general information extraction frameworks, which are not optimized for the biomedical domain and normally require the integration of complex external libraries and/or the development of custom tools. RESULTS: This article presents Neji, an open source framework optimized for biomedical concept recognition built around four key characteristics: modularity, scalability, speed, and usability. It integrates modules for biomedical natural language processing, such as sentence splitting, tokenization, lemmatization, part-of-speech tagging, chunking and dependency parsing. Concept recognition is provided through dictionary matching and machine learning with normalization methods. Neji also integrates an innovative concept tree implementation, supporting overlapped concept names and respective disambiguation techniques. The most popular input and output formats, namely Pubmed XML, IeXML, CoNLL and A1, are also supported. On top of the built-in functionalities, developers and researchers can implement new processing modules or pipelines, or use the provided command-line interface tool to build their own solutions, applying the most appropriate techniques to identify heterogeneous biomedical concepts. Neji was evaluated against three gold standard corpora with heterogeneous biomedical concepts (CRAFT, AnEM and NCBI disease corpus), achieving high performance results on named entity recognition (F1-measure for overlap matching: species 95%, cell 92%, cellular components 83%, gene and proteins 76%, chemicals 65%, biological processes and molecular functions 63%, disorders 85%, and anatomical entities 82%) and on entity normalization (F1-measure for overlap name matching and correct identifier included in the returned list of identifiers: species 88%, cell 71%, cellular components 72%, gene and proteins 64%, chemicals 53%, and biological processes and molecular functions 40%). Neji provides fast and multi-threaded data processing, annotating up to 1200 sentences/second when using dictionary-based concept identification. CONCLUSIONS: Considering the provided features and underlying characteristics, we believe that Neji is an important contribution to the biomedical community, streamlining the development of complex concept recognition solutions. Neji is freely available at http://bioinformatics.ua.pt/neji.

SUBMITTER: Campos D

PROVIDER: S-EPMC3849280 | biostudies-literature | 2013

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A modular framework for biomedical concept recognition.

Campos David D Matos Sérgio S Oliveira José Luís JL

BMC bioinformatics 20130924

<h4>Background</h4>Concept recognition is an essential task in biomedical information extraction, presenting several complex and unsolved challenges. The development of such solutions is typically performed in an ad-hoc manner or using general information extraction frameworks, which are not optimized for the biomedical domain and normally require the integration of complex external libraries and/or the development of custom tools.<h4>Results</h4>This article presents Neji, an open source framew ...[more]

PMID: 24063607

Similar Datasets

Project description:Knowledge graphs (KGs) are an important tool for representing complex relationships between entities in the biomedical domain. Several methods have been proposed for learning embeddings that can be used to predict new links in such graphs. Some methods ignore valuable attribute data associated with entities in biomedical KGs, such as protein sequences, or molecular graphs. Other works incorporate such data, but assume that entities can be represented with the same data modality. This is not always the case for biomedical KGs, where entities exhibit heterogeneous modalities that are central to their representation in the subject domain. We aim to understand how to incorporate multimodal data into biomedical KG embeddings, and analyze the resulting performance in comparison with traditional methods. We propose a modular framework for learning embeddings in KGs with entity attributes, that allows encoding attribute data of different modalities while also supporting entities with missing attributes. We additionally propose an efficient pretraining strategy for reducing the required training runtime. We train models using a biomedical KG containing approximately 2 million triples, and evaluate the performance of the resulting entity embeddings on the tasks of link prediction, and drug-protein interaction prediction, comparing against methods that do not take attribute data into account. In the standard link prediction evaluation, the proposed method results in competitive, yet lower performance than baselines that do not use attribute data. When evaluated in the task of drug-protein interaction prediction, the method compares favorably with the baselines. Further analyses show that incorporating attribute data does outperform baselines over entities below a certain node degree, comprising approximately 75% of the diseases in the graph. We also observe that optimizing attribute encoders is a challenging task that increases optimization costs. Our proposed pretraining strategy yields significantly higher performance while reducing the required training runtime. BioBLP allows to investigate different ways of incorporating multimodal biomedical data for learning representations in KGs. With a particular implementation, we find that incorporating attribute data does not consistently outperform baselines, but improvements are obtained on a comparatively large subset of entities below a specific node-degree. Our results indicate a potential for improved performance in scientific discovery tasks where understudied areas of the KG would benefit from link prediction methods.

Project description:ImportancePrioritization and funding for health initiatives, including biomedical innovation, may not consistently target unmet public health needs.ObjectiveTo (1) develop a quantitative, databased framework to identify and prioritize opportunities for biomedical product innovation investments based on a multicriteria decision-making model (MCDM) that includes comprehensive measures of public health burden and health care costs, and (2) pilot test the model.Design, setting, and participantsThe Department of Health and Human Services (HHS) convened public and private experts to develop a model, select measures, and complete a longitudinal pilot study to identify and prioritize opportunities for investment in biomedical product innovations that have the greatest public health benefit. Cross-sectional and longitudinal data (2012-2019) for 13 pilot medical disorders were obtained from the Institute for Health Metrics Global Burden of Disease database (IHME GBD) and the National Center for Health Statistics (NCHS).Main outcome measuresThe main outcome measure was an overall gap score reflecting high public health burden (composite measure of mortality, prevalence, years lived with disability, and health disparities), or high health care costs (composite measure of total, public, and out-of-pocket health spending) relative to low biomedical innovation. Sixteen innovation metrics were selected to reflect the pipeline of biomedical products from research and development to market approval. A higher score indicates a greater gap. Normalized composite scores were calculated for public health burden, cost, and innovation investment using the MCDM Technique for Order of Preference by Similarity to Ideal Solution method.ResultsAmong the 13 conditions tested in the pilot study, diabetes (0.61), osteoarthritis (0.46), and drug-use disorders (0.39) had the highest overall gap score reflecting high public health burden, or high health care costs relative to low biomedical innovation in these medical disorders. Chronic kidney disease (0.05), chronic obstructive pulmonary disease (0.09), and cirrhosis and other liver diseases (0.10) had the least amount of biomedical product innovation despite similar public health burden and health care cost scores.ConclusionsIn this cross-sectional pilot study, we developed and implemented a data-driven, proof-of-concept model that can help identify, quantify, and prioritize opportunities for biomedical product innovation. Quantifying the relative alignment between biomedical product innovation, public health burden, and health care cost may help identify and prioritize investments that can have the greatest public health benefit.

Project description:We designed a strategy for microarray analysis that is based on the identification of transcriptional modules formed by genes coordinately expressed in multiple disease data sets. Mapping changes in gene expression at the module level generated disease-specific transcriptional fingerprints that provide a stable framework for the visualization and functional interpretation of microarray data. The first step of the module-construction process analyzes expression patterns of transcripts across samples for individual diseases: sets of coordinately expressed transcripts were identified with an unsupervised clustering algorithm; in this case, the GeneSpring Version 7.1 (Agilent) implementation of the K-Means algorithm (k = 30). All transcripts detected in at least one sample were used as input; no screening for differential expression was performed. The second step of the module-construction process analyzed the “clustering behavior” of transcripts across diseases, taking into account the possibility that genes may cocluster in some diseases but not in others. Also, in our example, the transcripts that clustered together across all eight diseases were grouped to form a set of modules (round 1 of selection), and the stringency of the analysis was then decreased gradually to identify transcripts that belong to a similar K-means cluster in only a subset of diseases (round 2: seven out of eight diseases; round 3: six out of eight diseases). It is important to note that the module-selection process is “data-driven” and does not involve manual selection of genes by the investigator. We implemented the module-construction strategy described above, using as input a total of 239 peripheral-blood mononuclear cell (PBMC) samples obtained from individuals with one of the following conditions: systemic juvenile idiopathic arthritis (n = 47), systemic lupus erythematosus (n = 40), type I diabetes (n = 20), metastatic melanoma (n = 39), acute infections (Escherichia coli [n = 22], Staphylococcus aureus [n = 18], Influenza A [n = 16]) or liver-transplant recipients undergoing immunosuppressive therapy (n = 37). Transcriptional profiles were generated with Affymetrix U133A and U133B GeneChips (> 44,000 probe sets). A total of 4742 transcripts, distributed among 28 sets, were selected after running of the module-construction algorithm described above. Each module is assigned a unique identifier indicating the round and order of selection (i.e., M3.1 is the first module identified in the third round of selection). The stringency of this algorithm was tested statistically by implementation of the same module-construction procedure after randomization of the original data set. This process was repeated 200 times, without a single module identified. Therefore, the analysis of gene-cluster membership across multiple diseases provided a stringent means to identify PBMC transcriptional modules.

Project description:MotivationClustering methods can be useful for automatically grouping documents into meaningful clusters, improving human comprehension of a document collection. Although there are clustering algorithms that can achieve the goal for relatively large document collections, they do not always work well for small and homogenous datasets.MethodsIn this article, we present Retro-a novel clustering algorithm that extracts meaningful clusters along with concise and descriptive titles from small and homogenous document collections. Unlike common clustering approaches, our algorithm predicts cluster titles before clustering. It relies on the hypergeometric distribution model to discover key phrases, and generates candidate clusters by assigning documents to these phrases. Further, the statistical significance of candidate clusters is tested using supervised learning methods, and a multiple testing correction technique is used to control the overall quality of clustering.ResultsWe test our system on five disease datasets from OMIM(®) and evaluate the results based on MeSH(®) term assignments. We further compare our method with several baseline and state-of-the-art methods, including K-means, expectation maximization, latent Dirichlet allocation-based clustering, Lingo, OPTIMSRC and adapted GK-means. The experimental results on the 20-Newsgroup and ODP-239 collections demonstrate that our method is successful at extracting significant clusters and is superior to existing methods in terms of quality of clusters. Finally, we apply our system to a collection of 6248 topical sets from the HomoloGene(®) database, a resource in PubMed(®). Empirical evaluation confirms the method is useful for small homogenous datasets in producing meaningful clusters with descriptive titles.Availability and implementationA web-based demonstration of the algorithm applied to a collection of sets from the HomoloGene database is available at http://www.ncbi.nlm.nih.gov/CBBresearch/Wilbur/IRET/CLUSTERING_HOMOLOGENE/index.html.Contactlana.yeganova@nih.govSupplementary informationSupplementary data are available at Bioinformatics online.

Dataset Information

A modular framework for biomedical concept recognition.

Publications

A modular framework for biomedical concept recognition.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets