Automating Quality Assessment of Medical Evidence in Systematic Reviews: Model Development and Validation Study.

BACKGROUND: Assessment of the quality of medical evidence available on the web is a critical step in the preparation of systematic reviews. Existing tools that automate parts of this task validate the quality of individual studies but not of entire bodies of evidence and focus on a restricted set of quality criteria. OBJECTIVE: We proposed a quality assessment task that provides an overall quality rating for each body of evidence (BoE), as well as finer-grained justification for different quality criteria according to the Grading of Recommendation, Assessment, Development, and Evaluation formalization framework. For this purpose, we constructed a new data set and developed a machine learning baseline system (EvidenceGRADEr). METHODS: We algorithmically extracted quality-related data from all summaries of findings found in the Cochrane Database of Systematic Reviews. Each BoE was defined by a set of population, intervention, comparison, and outcome criteria and assigned a quality grade (high, moderate, low, or very low) together with quality criteria (justification) that influenced that decision. Different statistical data, metadata about the review, and parts of the review text were extracted as support for grading each BoE. After pruning the resulting data set with various quality checks, we used it to train several neural-model variants. The predictions were compared against the labels originally assigned by the authors of the systematic reviews. RESULTS: Our quality assessment data set, Cochrane Database of Systematic Reviews Quality of Evidence, contains 13,440 instances, or BoEs labeled for quality, originating from 2252 systematic reviews published on the internet from 2002 to 2020. On the basis of a 10-fold cross-validation, the best neural binary classifiers for quality criteria detected risk of bias at 0.78 F1 (P=.68; R=0.92) and imprecision at 0.75 F1 (P=.66; R=0.86), while the performance on inconsistency, indirectness, and publication bias criteria was lower (F1 in the range of 0.3-0.4). The prediction of the overall quality grade into 1 of the 4 levels resulted in 0.5 F1. When casting the task as a binary problem by merging the Grading of Recommendation, Assessment, Development, and Evaluation classes (high+moderate vs low+very low-quality evidence), we attained 0.74 F1. We also found that the results varied depending on the supporting information that is provided as an input to the models. CONCLUSIONS: Different factors affect the quality of evidence in the context of systematic reviews of medical evidence. Some of these (risk of bias and imprecision) can be automated with reasonable accuracy. Other quality dimensions such as indirectness, inconsistency, and publication bias prove more challenging for machine learning, largely because they are much rarer. This technology could substantially reduce reviewer workload in the future and expedite quality assessment as part of evidence synthesis.

A representation model for biological entities by fusing structured axioms with unstructured texts.

MOTIVATION: Structured semantic resources, for example, biological knowledge bases and ontologies, formally define biological concepts, entities and their semantic relationships, manifested as structured axioms and unstructured texts (e.g. textual definitions). The resources contain accurate expressions of biological reality and have been used by machine-learning models to assist intelligent applications like knowledge discovery. The current methods use both the axioms and definitions as plain texts in representation learning (RL). However, since the axioms are machine-readable while the natural language is human-understandable, difference in meaning of token and structure impedes the representations to encode desirable biological knowledge. RESULTS: We propose ERBK, a RL model of bio-entities. Instead of using the axioms and definitions as a textual corpus, our method uses knowledge graph embedding method and deep convolutional neural models to encode the axioms and definitions respectively. The representations could not only encode more underlying biological knowledge but also be further applied to zero-shot circumstance where existing approaches fall short. Experimental evaluations show that ERBK outperforms the existing methods for predicting protein-protein interactions and gene-disease associations. Moreover, it shows that ERBK still maintains promising performance under the zero-shot circumstance. We believe the representations and the method have certain generality and could extend to other types of bio-relation. AVAILABILITY AND IMPLEMENTATION: The source code is available at the gitlab repository https://gitlab.com/BioAI/erbk. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Development and validation of retrospective electronic frailty index using operational data of aged care homes.

BACKGROUND: Although elderly population is generally frail, it is important to closely monitor their health deterioration to improve the care and support in residential aged care homes (RACs). Currently, the best identification approach is through time-consuming regular geriatric assessments. This study aimed to develop and validate a retrospective electronic frailty index (reFI) to track the health status of people staying at RACs using the daily routine operational data records. METHODS: We have access to patient records from the Royal Freemasons Benevolent Institution RACs (Australia) over the age of 65, spanning 2010 to 2021. The reFI was developed using the cumulative deficit frailty model whose value was calculated as the ratio of number of present frailty deficits to the total possible frailty indicators (32). Frailty categories were defined using population quartiles. 1, 3 and 5-year mortality were used for validation. Survival analysis was performed using Kaplan-Meier estimate. Hazard ratios (HRs) were estimated using Cox regression analyses and the association was assessed using receiver operating characteristic (ROC) curves. RESULTS: Two thousand five hundred eighty-eight residents were assessed, with an average length of stay of 1.2 ± 2.2 years. The RAC cohort was generally frail with an average reFI of 0.21 ± 0.11. According to the Kaplan-Meier estimate, survival varied significantly across different frailty categories (p < 0.01). The estimated hazard ratios (HRs) were 1.12 (95% CI 1.09-1.15), 1.11 (95% CI 1.07-1.14), and 1.1 (95% CI 1.04-1.17) at 1, 3 and 5 years. The ROC analysis of the reFI for mortality outcome showed an area under the curve (AUC) of ≥0.60 for 1, 3 and 5-year mortality. CONCLUSION: A novel reFI was developed using the routine data recorded at RACs. reFI can identify changes in the frailty index over time for elderly people, that could potentially help in creating personalised care plans for addressing their health deterioration.

BioNorm: deep learning-based event normalization for the curation of reaction databases.

MOTIVATION: A biochemical reaction, bio-event, depicts the relationships between participating entities. Current text mining research has been focusing on identifying bio-events from scientific literature. However, rare efforts have been dedicated to normalize bio-events extracted from scientific literature with the entries in the curated reaction databases, which could disambiguate the events and further support interconnecting events into biologically meaningful and complete networks. RESULTS: In this paper, we propose BioNorm, a novel method of normalizing bio-events extracted from scientific literature to entries in the bio-molecular reaction database, e.g. IntAct. BioNorm considers event normalization as a paraphrase identification problem. It represents an entry as a natural language statement by combining multiple types of information contained in it. Then, it predicts the semantic similarity between the natural language statement and the statements mentioning events in scientific literature using a long short-term memory recurrent neural network (LSTM). An event will be normalized to the entry if the two statements are paraphrase. To the best of our knowledge, this is the first attempt of event normalization in the biomedical text mining. The experiments have been conducted using the molecular interaction data from IntAct. The results demonstrate that the method could achieve F-score of 0.87 in normalizing event-containing statements. AVAILABILITY AND IMPLEMENTATION: The source code is available at the gitlab repository https://gitlab.com/BioAI/leen and BioASQvec Plus is available on figshare https://figshare.com/s/45896c31d10c3f6d857a.

Word embeddings and recurrent neural networks based on Long-Short Term Memory nodes in supervised biomedical word sense disambiguation.

Word sense disambiguation helps identifying the proper sense of ambiguous words in text. With large terminologies such as the UMLS Metathesaurus ambiguities appear and highly effective disambiguation methods are required. Supervised learning algorithm methods are used as one of the approaches to perform disambiguation. Features extracted from the context of an ambiguous word are used to identify the proper sense of such a word. The type of features have an impact on machine learning methods, thus affect disambiguation performance. In this work, we have evaluated several types of features derived from the context of the ambiguous word and we have explored as well more global features derived from MEDLINE using word embeddings. Results show that word embeddings improve the performance of more traditional features and allow as well using recurrent neural network classifiers based on Long-Short Term Memory (LSTM) nodes. The combination of unigrams and word embeddings with an SVM sets a new state of the art performance with a macro accuracy of 95.97 in the MSH WSD data set.

Feature engineering for MEDLINE citation categorization with MeSH.

BACKGROUND: Research in biomedical text categorization has mostly used the bag-of-words representation. Other more sophisticated representations of text based on syntactic, semantic and argumentative properties have been less studied. In this paper, we evaluate the impact of different text representations of biomedical texts as features for reproducing the MeSH annotations of some of the most frequent MeSH headings. In addition to unigrams and bigrams, these features include noun phrases, citation meta-data, citation structure, and semantic annotation of the citations. RESULTS: Traditional features like unigrams and bigrams exhibit strong performance compared to other feature sets. Little or no improvement is obtained when using meta-data or citation structure. Noun phrases are too sparse and thus have lower performance compared to more traditional features. Conceptual annotation of the texts by MetaMap shows similar performance compared to unigrams, but adding concepts from the UMLS taxonomy does not improve the performance of using only mapped concepts. The combination of all the features performs largely better than any individual feature set considered. In addition, this combination improves the performance of a state-of-the-art MeSH indexer. Concerning the machine learning algorithms, we find that those that are more resilient to class imbalance largely obtain better performance. CONCLUSIONS: We conclude that even though traditional features such as unigrams and bigrams have strong performance compared to other features, it is possible to combine them to effectively improve the performance of the bag-of-words representation. We have also found that the combination of the learning algorithm and feature sets has an influence in the overall performance of the system. Moreover, using learning algorithms resilient to class imbalance largely improves performance. However, when using a large set of features, consideration needs to be taken with algorithms due to the risk of over-fitting. Specific combinations of learning algorithms and features for individual MeSH headings could further increase the performance of an indexing system.

Knowledge based word-concept model estimation and refinement for biomedical text mining.

Text mining of scientific literature has been essential for setting up large public biomedical databases, which are being widely used by the research community. In the biomedical domain, the existence of a large number of terminological resources and knowledge bases (KB) has enabled a myriad of machine learning methods for different text mining related tasks. Unfortunately, KBs have not been devised for text mining tasks but for human interpretation, thus performance of KB-based methods is usually lower when compared to supervised machine learning methods. The disadvantage of supervised methods though is they require labeled training data and therefore not useful for large scale biomedical text mining systems. KB-based methods do not have this limitation. In this paper, we describe a novel method to generate word-concept probabilities from a KB, which can serve as a basis for several text mining tasks. This method not only takes into account the underlying patterns within the descriptions contained in the KB but also those in texts available from large unlabeled corpora such as MEDLINE. The parameters of the model have been estimated without training data. Patterns from MEDLINE have been built using MetaMap for entity recognition and related using co-occurrences. The word-concept probabilities were evaluated on the task of word sense disambiguation (WSD). The results showed that our method obtained a higher degree of accuracy than other state-of-the-art approaches when evaluated on the MSH WSD data set. We also evaluated our method on the task of document ranking using MEDLINE citations. These results also showed an increase in performance over existing baseline retrieval approaches.

GeneRIF indexing: sentence selection based on machine learning.

BACKGROUND: A Gene Reference Into Function (GeneRIF) describes novel functionality of genes. GeneRIFs are available from the National Center for Biotechnology Information (NCBI) Gene database. GeneRIF indexing is performed manually, and the intention of our work is to provide methods to support creating the GeneRIF entries. The creation of GeneRIF entries involves the identification of the genes mentioned in MEDLINE®; citations and the sentences describing a novel function. RESULTS: We have compared several learning algorithms and several features extracted or derived from MEDLINE sentences to determine if a sentence should be selected for GeneRIF indexing. Features are derived from the sentences or using mechanisms to augment the information provided by them: assigning a discourse label using a previously trained model, for example. We show that machine learning approaches with specific feature combinations achieve results close to one of the annotators. We have evaluated different feature sets and learning algorithms. In particular, Naïve Bayes achieves better performance with a selection of features similar to one used in related work, which considers the location of the sentence, the discourse of the sentence and the functional terminology in it. CONCLUSIONS: The current performance is at a level similar to human annotation and it shows that machine learning can be used to automate the task of sentence selection for GeneRIF annotation. The current experiments are limited to the human species. We would like to see how the methodology can be extended to other species, specifically the normalization of gene mentions in other species.

MeSH indexing based on automatically generated summaries.

BACKGROUND: MEDLINE citations are manually indexed at the U.S. National Library of Medicine (NLM) using as reference the Medical Subject Headings (MeSH) controlled vocabulary. For this task, the human indexers read the full text of the article. Due to the growth of MEDLINE, the NLM Indexing Initiative explores indexing methodologies that can support the task of the indexers. Medical Text Indexer (MTI) is a tool developed by the NLM Indexing Initiative to provide MeSH indexing recommendations to indexers. Currently, the input to MTI is MEDLINE citations, title and abstract only. Previous work has shown that using full text as input to MTI increases recall, but decreases precision sharply. We propose using summaries generated automatically from the full text for the input to MTI to use in the task of suggesting MeSH headings to indexers. Summaries distill the most salient information from the full text, which might increase the coverage of automatic indexing approaches based on MEDLINE. We hypothesize that if the results were good enough, manual indexers could possibly use automatic summaries instead of the full texts, along with the recommendations of MTI, to speed up the process while maintaining high quality of indexing results. RESULTS: We have generated summaries of different lengths using two different summarizers, and evaluated the MTI indexing on the summaries using different algorithms: MTI, individual MTI components, and machine learning. The results are compared to those of full text articles and MEDLINE citations. Our results show that automatically generated summaries achieve similar recall but higher precision compared to full text articles. Compared to MEDLINE citations, summaries achieve higher recall but lower precision. CONCLUSIONS: Our results show that automatic summaries produce better indexing than full text articles. Summaries produce similar recall to full text but much better precision, which seems to indicate that automatic summaries can efficiently capture the most important contents within the original articles. The combination of MEDLINE citations and automatically generated summaries could improve the recommendations suggested by MTI. On the other hand, indexing performance might be dependent on the MeSH heading being indexed. Summarization techniques could thus be considered as a feature selection algorithm that might have to be tuned individually for each MeSH heading.

Classifying literature mentions of biological pathogens as experimentally studied using natural language processing.

BACKGROUND: Information pertaining to mechanisms, management and treatment of disease-causing pathogens including viruses and bacteria is readily available from research publications indexed in MEDLINE. However, identifying the literature that specifically characterises these pathogens and their properties based on experimental research, important for understanding of the molecular basis of diseases caused by these agents, requires sifting through a large number of articles to exclude incidental mentions of the pathogens, or references to pathogens in other non-experimental contexts such as public health. OBJECTIVE: In this work, we lay the foundations for the development of automatic methods for characterising mentions of pathogens in scientific literature, focusing on the task of identifying research that involves the experimental study of a pathogen in an experimental context. There are no manually annotated pathogen corpora available for this purpose, while such resources are necessary to support the development of machine learning-based models. We therefore aim to fill this gap, producing a large data set automatically from MEDLINE under some simplifying assumptions for the task definition, and using it to explore automatic methods that specifically support the detection of experimentally studied pathogen mentions in research publications. METHODS: We developed a pathogen mention characterisation literature data set -READBiomed-Pathogens- automatically using NCBI resources, which we make available. Resources such as the NCBI Taxonomy, MeSH and GenBank can be used effectively to identify relevant literature about experimentally researched pathogens, more specifically using MeSH to link to MEDLINE citations including titles and abstracts with experimentally researched pathogens. We experiment with several machine learning-based natural language processing (NLP) algorithms leveraging this data set as training data, to model the task of detecting papers that specifically describe experimental study of a pathogen. RESULTS: We show that our data set READBiomed-Pathogens can be used to explore natural language processing configurations for experimental pathogen mention characterisation. READBiomed-Pathogens includes citations related to organisms including bacteria, viruses, and a small number of toxins and other disease-causing agents. CONCLUSIONS: We studied the characterisation of experimentally studied pathogens in scientific literature, developing several natural language processing methods supported by an automatically developed data set. As a core contribution of the work, we presented a methodology to automatically construct a data set for pathogen identification using existing biomedical resources. The data set and the annotation code are made publicly available. Performance of the pathogen mention identification and characterisation algorithms were additionally evaluated on a small manually annotated data set shows that the data set that we have generated allows characterising pathogens of interest. TRIAL REGISTRATION: N/A.

Exploiting MeSH indexing in MEDLINE to generate a data set for word sense disambiguation.

BACKGROUND: Evaluation of Word Sense Disambiguation (WSD) methods in the biomedical domain is difficult because the available resources are either too small or too focused on specific types of entities (e.g. diseases or genes). We present a method that can be used to automatically develop a WSD test collection using the Unified Medical Language System (UMLS) Metathesaurus and the manual MeSH indexing of MEDLINE. We demonstrate the use of this method by developing such a data set, called MSH WSD. METHODS: In our method, the Metathesaurus is first screened to identify ambiguous terms whose possible senses consist of two or more MeSH headings. We then use each ambiguous term and its corresponding MeSH heading to extract MEDLINE citations where the term and only one of the MeSH headings co-occur. The term found in the MEDLINE citation is automatically assigned the UMLS CUI linked to the MeSH heading. Each instance has been assigned a UMLS Concept Unique Identifier (CUI). We compare the characteristics of the MSH WSD data set to the previously existing NLM WSD data set. RESULTS: The resulting MSH WSD data set consists of 106 ambiguous abbreviations, 88 ambiguous terms and 9 which are a combination of both, for a total of 203 ambiguous entities. For each ambiguous term/abbreviation, the data set contains a maximum of 100 instances per sense obtained from MEDLINE.We evaluated the reliability of the MSH WSD data set using existing knowledge-based methods and compared their performance to that of the results previously obtained by these algorithms on the pre-existing data set, NLM WSD. We show that the knowledge-based methods achieve different results but keep their relative performance except for the Journal Descriptor Indexing (JDI) method, whose performance is below the other methods. CONCLUSIONS: The MSH WSD data set allows the evaluation of WSD algorithms in the biomedical domain. Compared to previously existing data sets, MSH WSD contains a larger number of biomedical terms/abbreviations and covers the largest set of UMLS Semantic Types. Furthermore, the MSH WSD data set has been generated automatically reusing already existing annotations and, therefore, can be regenerated from subsequent UMLS versions.

Collocation analysis for UMLS knowledge-based word sense disambiguation.

BACKGROUND: The effectiveness of knowledge-based word sense disambiguation (WSD) approaches depends in part on the information available in the reference knowledge resource. Off the shelf, these resources are not optimized for WSD and might lack terms to model the context properly. In addition, they might include noisy terms which contribute to false positives in the disambiguation results. METHODS: We analyzed some collocation types which could improve the performance of knowledge-based disambiguation methods. Collocations are obtained by extracting candidate collocations from MEDLINE and then assigning them to one of the senses of an ambiguous word. We performed this assignment either using semantic group profiles or a knowledge-based disambiguation method. In addition to collocations, we used second-order features from a previously implemented approach.Specifically, we measured the effect of these collocations in two knowledge-based WSD methods. The first method, AEC, uses the knowledge from the UMLS to collect examples from MEDLINE which are used to train a Naïve Bayes approach. The second method, MRD, builds a profile for each candidate sense based on the UMLS and compares the profile to the context of the ambiguous word.We have used two WSD test sets which contain disambiguation cases which are mapped to UMLS concepts. The first one, the NLM WSD set, was developed manually by several domain experts and contains words with high frequency occurrence in MEDLINE. The second one, the MSH WSD set, was developed automatically using the MeSH indexing in MEDLINE. It contains a larger set of words and covers a larger number of UMLS semantic types. RESULTS: The results indicate an improvement after the use of collocations, although the approaches have different performance depending on the data set. In the NLM WSD set, the improvement is larger for the MRD disambiguation method using second-order features. Assignment of collocations to a candidate sense based on UMLS semantic group profiles is more effective in the AEC method.In the MSH WSD set, the increment in performance is modest for all the methods. Collocations combined with the MRD disambiguation method have the best performance. The MRD disambiguation method and second-order features provide an insignificant change in performance. The AEC disambiguation method gives a modest improvement in performance. Assignment of collocations to a candidate sense based on knowledge-based methods has better performance. CONCLUSIONS: Collocations improve the performance of knowledge-based disambiguation methods, although results vary depending on the test set and method used. Generally, the AEC method is sensitive to query drift. Using AEC, just a few selected terms provide a large improvement in disambiguation performance. The MRD method handles noisy terms better but requires a larger set of terms to improve performance.

Studying the correlation between different word sense disambiguation methods and summarization effectiveness in biomedical texts.

BACKGROUND: Word sense disambiguation (WSD) attempts to solve lexical ambiguities by identifying the correct meaning of a word based on its context. WSD has been demonstrated to be an important step in knowledge-based approaches to automatic summarization. However, the correlation between the accuracy of the WSD methods and the summarization performance has never been studied. RESULTS: We present three existing knowledge-based WSD approaches and a graph-based summarizer. Both the WSD approaches and the summarizer employ the Unified Medical Language System (UMLS) Metathesaurus as the knowledge source. We first evaluate WSD directly, by comparing the prediction of the WSD methods to two reference sets: the NLM WSD dataset and the MSH WSD collection. We next apply the different WSD methods as part of the summarizer, to map documents onto concepts in the UMLS Metathesaurus, and evaluate the summaries that are generated. The results obtained by the different methods in both evaluations are studied and compared. CONCLUSIONS: It has been found that the use of WSD techniques has a positive impact on the results of our graph-based summarizer, and that, when both the WSD and summarization tasks are assessed over large and homogeneous evaluation collections, there exists a correlation between the overall results of the WSD and summarization tasks. Furthermore, the best WSD algorithm in the first task tends to be also the best one in the second. However, we also found that the improvement achieved by the summarizer is not directly correlated with the WSD performance. The most likely reason is that the errors in disambiguation are not equally important but depend on the relative salience of the different concepts in the document to be summarized.

Predictive models for cochlear implant outcomes: Performance, generalizability, and the impact of cohort size.

While cochlear implants have helped hundreds of thousands of individuals, it remains difficult to predict the extent to which an individual's hearing will benefit from implantation. Several publications indicate that machine learning may improve predictive accuracy of cochlear implant outcomes compared to classical statistical methods. However, existing studies are limited in terms of model validation and evaluating factors like sample size on predictive performance. We conduct a thorough examination of machine learning approaches to predict word recognition scores (WRS) measured approximately 12 months after implantation in adults with post-lingual hearing loss. This is the largest retrospective study of cochlear implant outcomes to date, evaluating 2,489 cochlear implant recipients from three clinics. We demonstrate that while machine learning models significantly outperform linear models in prediction of WRS, their overall accuracy remains limited (mean absolute error: 17.9-21.8). The models are robust across clinical cohorts, with predictive error increasing by at most 16% when evaluated on a clinic excluded from the training set. We show that predictive improvement is unlikely to be improved by increasing sample size alone, with doubling of sample size estimated to only increasing performance by 3% on the combined dataset. Finally, we demonstrate how the current models could support clinical decision making, highlighting that subsets of individuals can be identified that have a 94% chance of improving WRS by at least 10% points after implantation, which is likely to be clinically meaningful. We discuss several implications of this analysis, focusing on the need to improve and standardize data collection.

A MultiCenter Analysis of Factors Associated with Hearing Outcome for 2,735 Adults with Cochlear Implants.

While the majority of cochlear implant recipients benefit from the device, it remains difficult to estimate the degree of benefit for a specific patient prior to implantation. Using data from 2,735 cochlear-implant recipients from across three clinics, the largest retrospective study of cochlear-implant outcomes to date, we investigate the association between 21 preoperative factors and speech recognition approximately one year after implantation and explore the consistency of their effects across the three constituent datasets. We provide evidence of 17 statistically significant associations, in either univariate or multivariate analysis, including confirmation of associations for several predictive factors, which have only been examined in prior smaller studies. Despite the large sample size, a multivariate analysis shows that the variance explained by our models remains modest across the datasets (R2=0.12-0.21). Finally, we report a novel statistical interaction indicating that the duration of deafness in the implanted ear has a stronger impact on hearing outcome when considered relative to a candidate's age. Our multicenter study highlights several real-world complexities that impact the clinical translation of predictive factors for cochlear implantation outcome. We suggest several directions to overcome these challenges and further improve our ability to model patient outcomes with increased accuracy.

Knowledge-based biomedical word sense disambiguation: comparison of approaches.

BACKGROUND: Word sense disambiguation (WSD) algorithms attempt to select the proper sense of ambiguous terms in text. Resources like the UMLS provide a reference thesaurus to be used to annotate the biomedical literature. Statistical learning approaches have produced good results, but the size of the UMLS makes the production of training data infeasible to cover all the domain. METHODS: We present research on existing WSD approaches based on knowledge bases, which complement the studies performed on statistical learning. We compare four approaches which rely on the UMLS Metathesaurus as the source of knowledge. The first approach compares the overlap of the context of the ambiguous word to the candidate senses based on a representation built out of the definitions, synonyms and related terms. The second approach collects training data for each of the candidate senses to perform WSD based on queries built using monosemous synonyms and related terms. These queries are used to retrieve MEDLINE citations. Then, a machine learning approach is trained on this corpus. The third approach is a graph-based method which exploits the structure of the Metathesaurus network of relations to perform unsupervised WSD. This approach ranks nodes in the graph according to their relative structural importance. The last approach uses the semantic types assigned to the concepts in the Metathesaurus to perform WSD. The context of the ambiguous word and semantic types of the candidate concepts are mapped to Journal Descriptors. These mappings are compared to decide among the candidate concepts. Results are provided estimating accuracy of the different methods on the WSD test collection available from the NLM. CONCLUSIONS: We have found that the last approach achieves better results compared to the other methods. The graph-based approach, using the structure of the Metathesaurus network to estimate the relevance of the Metathesaurus concepts, does not perform well compared to the first two methods. In addition, the combination of methods improves the performance over the individual approaches. On the other hand, the performance is still below statistical learning trained on manually produced data and below the maximum frequency sense baseline. Finally, we propose several directions to improve the existing methods and to improve the Metathesaurus to be more effective in WSD.

Measuring prediction capacity of individual verbs for the identification of protein interactions.

MOTIVATION: The identification of events such as protein-protein interactions (PPIs) from the scientific literature is a complex task. One of the reasons is that there is no formal syntax to denote such relations in the scientific literature. Nonetheless, it is important to understand such relational event representations to improve information extraction solutions (e.g., for gene regulatory events). In this study, we analyze publicly available protein interaction corpora (AIMed, BioInfer, BioCreAtIve II) to determine the scope of verbs used to denote protein interactions and to measure their predictive capacity for the identification of PPI events. Our analysis is based on syntactical language patterns. This restriction has the advantage that the verb mention is used as the independent variable in the experiments enabling comparability of results in the usage of the verbs. The initial selection of verbs has been generated from a systematic analysis of the scientific literature and existing corpora for PPIs. We distinguish modifying interactions (MIs) such as posttranslational modifications (PTMs) from non-modifying interactions (NMIs) and assumed that MIs have a higher predictive capacity due to stronger scientific evidence proving the interaction. We found that MIs are less frequent in the corpus but can be extracted at the same precision levels as PPIs. A significant portion of correct PPI reportings in the BioCreAtIve II corpus use the verb "associate", which semantically does not prove a relation. The performance of every monitored verb is listed and allows the selection of specific verbs to improve the performance of PPI extraction solutions. Programmatic access to the text processing modules is available online (www.ebi.ac.uk/webservices/whatizit/info.jsf) and the full analysis of Medline abstracts will be made through the Web pages of the Rebholz group.

Combining Social Media and FDA Adverse Event Reporting System to Detect Adverse Drug Reactions.

INTRODUCTION: Adverse drug reactions (ADRs) are unintended reactions caused by a drug or combination of drugs taken by a patient. The current safety surveillance system relies on spontaneous reporting systems (SRSs) and more recently on observational health data; however, ADR detection may be delayed and lack geographic diversity. The broad scope of social media conversations, such as those on Twitter, can include health-related topics. Consequently, these data could be used to detect potentially novel ADRs with less latency. Although research regarding ADR detection using social media has made progress, findings are based on single information sources, and no study has yet integrated drug safety evidence from both an SRS and Twitter. OBJECTIVE: The aim of this study was to combine signals from an SRS and Twitter to facilitate the detection of safety signals and compare the performance of the combined system with signals generated by individual data sources. METHODS: We extracted potential drug-ADR posts from Twitter, used Monte Carlo expectation maximization to generate drug safety signals from both the US FDA Adverse Event Reporting System and posts from Twitter, and then integrated these signals using a Bayesian hierarchical model. The results from the integrated system and two individual sources were evaluated using a reference standard derived from drug labels. Area under the receiver operating characteristics curve (AUC) was computed to measure performance. RESULTS: We observed a significant improvement in the AUC of the combined system when comparing it with Twitter alone, and no improvement when comparing with the SRS alone. The AUCs ranged from 0.587 to 0.637 for the combined SRS and Twitter, from 0.525 to 0.534 for Twitter alone, and from 0.612 to 0.642 for the SRS alone. The results varied because different preprocessing procedures were applied to Twitter. CONCLUSION: The accuracy of signal detection using social media can be improved by combining signals with those from SRSs. However, the combined system cannot achieve better AUC performance than data from FAERS alone, which may indicate that Twitter data are not ready to be integrated into a purely data-driven combination system.

Prediction of secondary structure population and intrinsic disorder of proteins using multitask deep learning.

Recent research in predicting protein secondary structure populations (SSP) based on Nuclear Magnetic Resonance (NMR) chemical shifts has helped quantitatively characterise the structural conformational properties of intrinsically disordered proteins and regions (IDP/IDR). Different from protein secondary structure (SS) prediction, the SSP prediction assumes a dynamic assignment of secondary structures that seem correlate with disordered states. In this study, we designed a single-task deep learning framework to predict IDP/IDR and SSP respectively; and multitask deep learning frameworks to allow quantitative predictions of IDP/IDR evidenced by the simultaneously predicted SSP. According to independent test results, single-task deep learning models improve the prediction performance of shallow models for SSP and IDP/IDR. Also, the prediction performance was further improved for IDP/IDR prediction when SSP prediction was simultaneously predicted in multitask models. With p53 as a use case, we demonstrate how predicted SSP is used to explain the IDP/IDR predictions for each functional region.

Evaluation of Combined Artificial Intelligence and Radiologist Assessment to Interpret Screening Mammograms.

Importance: Mammography screening currently relies on subjective human interpretation. Artificial intelligence (AI) advances could be used to increase mammography screening accuracy by reducing missed cancers and false positives. Objective: To evaluate whether AI can overcome human mammography interpretation limitations with a rigorous, unbiased evaluation of machine learning algorithms. Design, Setting, and Participants: In this diagnostic accuracy study conducted between September 2016 and November 2017, an international, crowdsourced challenge was hosted to foster AI algorithm development focused on interpreting screening mammography. More than 1100 participants comprising 126 teams from 44 countries participated. Analysis began November 18, 2016. Main Outcomes and Measurements: Algorithms used images alone (challenge 1) or combined images, previous examinations (if available), and clinical and demographic risk factor data (challenge 2) and output a score that translated to cancer yes/no within 12 months. Algorithm accuracy for breast cancer detection was evaluated using area under the curve and algorithm specificity compared with radiologists' specificity with radiologists' sensitivity set at 85.9% (United States) and 83.9% (Sweden). An ensemble method aggregating top-performing AI algorithms and radiologists' recall assessment was developed and evaluated. Results: Overall, 144 231 screening mammograms from 85 580 US women (952 cancer positive ≤12 months from screening) were used for algorithm training and validation. A second independent validation cohort included 166 578 examinations from 68 008 Swedish women (780 cancer positive). The top-performing algorithm achieved an area under the curve of 0.858 (United States) and 0.903 (Sweden) and 66.2% (United States) and 81.2% (Sweden) specificity at the radiologists' sensitivity, lower than community-practice radiologists' specificity of 90.5% (United States) and 98.5% (Sweden). Combining top-performing algorithms and US radiologist assessments resulted in a higher area under the curve of 0.942 and achieved a significantly improved specificity (92.0%) at the same sensitivity. Conclusions and Relevance: While no single AI algorithm outperformed radiologists, an ensemble of AI algorithms combined with radiologist assessment in a single-reader screening environment improved overall accuracy. This study underscores the potential of using machine learning methods for enhancing mammography screening interpretation.