Medical Concept Representation Research Articles

Learning medical concept representation based on semantic information in medical textural data

Electric Health Records (EHR) have been widely adopted by many hospitals to improve clinical decision making and re-admission prediction. Each patient admission usually contains both multiple medical codes as well as clinical notes. Accurate learning of the representations (also called embeddings) of medical concepts from EHR is a key strategy to improve prediction performance in healthcare. Existing works employ medical ontologies to improve the quality of representations but focus solely on the relationships amongst the medical codes, ignoring textual data such medical code descriptions, clinical notes, and patient demographics. In this paper, we propose a new model called Semantic-based Attention model using Textual data for Medical Concept Embedding (SATexMCE). SATexMCE consists of three parts: medical codes embedding, admission embedding, and prediction model. First, we generate representations of medical codes by using both the textual description of medical codes and also the relationships among medical codes. Then, we generate the admission representation based on the representation of medical codes in the admission, the clinical notes and the demographic information associated with the admission via several attention mechanisms. The admission embeddings are used to construct a Recurrent neural network model which is used to predict patients’ readmission and a disease in the next admission based on patient admission data. Experimental results show that the proposed SATexMCE model improves not only the performance of readmission prediction but also the quality of medical concept representations. Attention mechanism helps us measure the importance of different medical codes and understand the meaning of the words in clinical notes for predictions.

Aceso: PICO-guided Evidence Summarization on Medical Literature.

Evidence-Based Medicine (EBM) aims to apply the best available evidence gained from scientific methods to clinical decision making. A generally accepted criterion to formulate evidence is to use the PICO framework, where PICO stands for Problem/Population, Intervention, Comparison, and Outcome. Automatic extraction of PICO-related sentences from medical literature is crucial to the success of many EBM applications. In this work, we present our Aceso system, which automatically generates PICO-based evidence summaries from medical literature. In Aceso 1, we adopt an active learning paradigm, which helps to minimize the cost of manual labeling and to optimize the quality of summarization with limited labeled data. An UMLS2Vec model is proposed to learn a vector representation of medical concepts in UMLS 2, and we fuse the embedding of medical knowledge with textual features in summarization. The evaluation shows that our approach is better on identifying PICO sentences against state-of-the-art studies and outperforms baseline methods on producing high-quality evidence summaries.

Medical Concept Representation Learning from Multi-source Data.

Representing words as low dimensional vectors is very useful in many natural language processing tasks. This idea has been extended to medical domain where medical codes listed in medical claims are represented as vectors to facilitate exploratory analysis and predictive modeling. However, depending on a type of a medical provider, medical claims can use medical codes from different ontologies or from a combination of ontologies, which complicates learning of the representations. To be able to properly utilize such multi-source medical claim data, we propose an approach that represents medical codes from different ontologies in the same vector space. We first modify the Pointwise Mutual Information (PMI) measure of similarity between the codes. We then develop a new negative sampling method for word2vec model that implicitly factorizes the modified PMI matrix. The new approach was evaluated on the code cross-reference problem, which aims at identifying similar codes across different ontologies. In our experiments, we evaluated cross-referencing between ICD-9 and CPT medical code ontologies. Our results indicate that vector representations of codes learned by the proposed approach provide superior cross-referencing when compared to several existing approaches.

DWE-Med

Recent advances in unsupervised language processing methods have created an opportunity to exploit massive text corpora for developing high-quality vector space representation (also known as word embeddings) of words. Towards this direction, practitioners have developed and applied several data driven embedding models with quite good rate of success. However, a drawback of these models lies in their premise of static context; wherein, the meaning of a word is assumed to remain the same over the period of time. This is limiting because it is known that the semantic meaning of a concept evolves over time. While such semantic drifts are routinely observed in almost all the domains; their effect is acute in domain such as biomedicine, where the semantic meaning of a concept changes relatively fast. To address this, in this study, we aim to learn temporally aware vector representation of medical concepts from the timestamped text data, and in doing so provide a systematic approach to formalize the problem. More specifically, a dynamic word embedding based model that jointly learns the temporal characteristics of medical concepts and performs across time-alignment is proposed. Apart from capturing the evolutionary characteristics in an optimal manner, the model also factors in the implicit medical properties useful for a variety of bio-medical applications. Empirical studies conducted on two important bio-medical use cases validates the effectiveness of the proposed approach and suggests that the model not only learns quality embeddings but also facilitates intuitive trajectory visualizations.

MCPL-Based FT-LSTM: Medical Representation Learning-Based Clinical Prediction Model for Time Series Events

Large collections of electronic medical records (EMRs) provide us with a vast source of information on medical practice. However, the utilization of these data to support clinical decisions is still limited. Extracting useful patterns from such data is particularly challenging because the data are variable longitudinal, sparse, and heterogeneous. Therefore, in this paper, we propose the MCPL-based FT-LSTM, a clinical event prediction method based on medical concept representation learning. On one hand, inspired by FASTTEXT, we have developed an interpretative vector representation of medical events in EMRs, which enables us to capture the medical concept information effectively so that the patient's clinical data can be represented more reasonably. On the other hand, we propose a novel time-controlled long short-term memory (LSTM) prediction model, which adds time-control units to the original LSTM model. The model can describe the variable time intervals in EMRs, better capture long-term, and short-term information, and eliminate the strong dependence of clinical data on timestamps; thus, improving the model's prediction performance for clinical events. Through extensive experiments on the MIMICIII dataset, we demonstrate that the MCPL-based FT-LSTM achieves higher precision in the field of clinical event prediction, which is of great significance for the medical information research.

An Extraordinary Obituary for an Extraordinary Man: Attila Naszlady

An Extraordinary Obituary for an Extraordinary Man: Attila Naszlady

From Concept Representations to Ontologies: A Paradigm Shift in Health Informatics?

ObjectivesThis work aims at uncovering challenges in biomedical knowledge representation research by providing an understanding of what was historically called "medical concept representation" and used as the name for a working group of the International Medical Informatics Association.MethodsBibliometrics, text mining, and a social media survey compare the research done in this area between two periods, before and after 2000.ResultsBoth the opinion of socially active groups of researchers and the interpretation of bibliometric data since 1988 suggest that the focus of research has moved from "medical concept representation" to "medical ontologies".ConclusionsIt remains debatable whether the observed change amounts to a paradigm shift or whether it simply reflects changes in naming, following the natural evolution of ontology research and engineering activities in the 1990s. The availability of powerful tools to handle ontologies devoted to certain areas of biomedicine has not resulted in a large-scale breakthrough beyond advances in basic research.

Medical concept representation: the years beyond 2000.

This work aims at understanding the state of the art in the broad contextual research area of "medical concept representation". Our data support the general understanding that the focus of research has moved toward medical ontologies, which we interpret as a paradigm shift. Both the opinion of socially active groups of researchers and changes in bibliometric data since 1988 support this opinion. Socially active researchers mention the OBO foundry, SNOMED CT, and the UMLS as anchor activities.

HIR Collaborating with the CODATA Conference

HIR Collaborating with the CODATA Conference

Fuzzy logic and semiotic methods in modeling of medical concepts

The field of medicine is a quickly growing area of application for computer-based systems. However, the use of computerized methods in this knowledge-intensive and expert-based discipline brings multiple challenges. The major problem is the modeling, representing, and interpreting of diverse medical concepts. For example, some symptoms and their etiologies are described in terms of molecular biology and genetics, physiological processes are defined using models from chemistry and physics; yet mental disorders are defined in more subjective terms of feelings, behaviours, habits, and life events. Thus, the representation of medical concepts must be sufficiently expressive to model concepts which are inherently complex, context-dependent, evolving, and often imprecise. Furthermore, the representation must be formal or, at least, sufficiently rigorous in order to be processed by computers and at the same time, the representation must be human-readable in order to be validated by humans. In this paper, we describe the modeling process of medical concepts as a mapping from the real-world medical concepts into their computational models, and further into their physical implementation. First, we define the notion of a concept as a fundamental unit of knowledge and specify the fundamental principles of the computational representation of a concept. Second, we describe the characteristics of medical concepts, specifically their historical and cultural changeability, their social and cultural ambiguity, and their varied levels of precision. Third, we present a meta-modeling framework for computational representation of medical concepts. Our framework is based on fuzzy logic and semiotic methods which allow us to explicitly model two important characteristics of medical concepts: imprecision and context-dependency. We present the framework using an example of a mental disorder, specifically, the concept of clinical depression. To exemplify the changeable and evolutionary character of medical concepts, we discuss the development of the diagnostic criteria for depression. Finally, we use the example of the assessment of depression to describe the computational representation for polythetic and multi-dimensional concepts and for categorical and non-categorical concepts. We demonstrate how the proposed modeling framework utilizes (1) a fuzzy-logic approach to represent the non-categorical (continuous) nature of the symptoms and (2) a semiotic approach to represent the polythetic (contextual interpretation) and dimensional nature of the symptoms.

Semantic Integration of Cervical Cancer Data Repositories to Facilitate Multicenter Association Studies: The ASSIST Approach

The current work addresses the unification of Electronic Health Records related to cervical cancer into a single medical knowledge source, in the context of the EU-funded ASSIST research project. The project aims to facilitate the research for cervical precancer and cancer through a system that virtually unifies multiple patient record repositories, physically located in different medical centers/hospitals, thus, increasing flexibility by allowing the formation of study groups "on demand" and by recycling patient records in new studies. To this end, ASSIST uses semantic technologies to translate all medical entities (such as patient examination results, history, habits, genetic profile) and represent them in a common form, encoded in the ASSIST Cervical Cancer Ontology. The current paper presents the knowledge elicitation approach followed, towards the definition and representation of the disease's medical concepts and rules that constitute the basis for the ASSIST Cervical Cancer Ontology. The proposed approach constitutes a paradigm for semantic integration of heterogeneous clinical data that may be applicable to other biomedical application domains.

An iconic language for the graphical representation of medical concepts

BackgroundMany medication errors are encountered in drug prescriptions, which would not occur if practitioners could remember the drug properties. They can refer to drug monographs to find these properties, however drug monographs are long and tedious to read during consultation. We propose a two-step approach for facilitating access to drug monographs. The first step, presented here, is the design of a graphical language, called VCM.MethodsThe VCM graphical language was designed using a small number of graphical primitives and combinatory rules. VCM was evaluated over 11 volunteer general practitioners to assess if the language is easy to learn, to understand and to use. Evaluators were asked to register their VCM training time, to indicate the meaning of VCM icons and sentences, and to answer clinical questions related to randomly generated drug monograph-like documents, supplied in text or VCM format.ResultsVCM can represent the various signs, diseases, physiological states, life habits, drugs and tests described in drug monographs. Grammatical rules make it possible to generate many icons by combining a small number of primitives and reusing simple icons to build more complex ones. Icons can be organized into simple sentences to express drug recommendations. Evaluation showed that VCM was learnt in 2 to 7 hours, that physicians understood 89% of the tested VCM icons, and that they answered correctly to 94% of questions using VCM (versus 88% using text, p = 0.003) and 1.8 times faster (p < 0.001).ConclusionVCM can be learnt in a few hours and appears to be easy to read. It can now be used in a second step: the design of graphical interfaces facilitating access to drug monographs. It could also be used for broader applications, including the design of interfaces for consulting other types of medical document or medical data, or, very simply, to enrich medical texts.

Natural language generation of surgical procedures

A number of compositional Medical Concept Representation systems are being developed. Although these provide for a detailed conceptual representation of the underlying information, they have to be translated back to natural language for used by end-users and applications. The GALEN programme has been developing one such representation and we report here on a tool developed to generate natural language phrases from the GALEN conceptual representations. This tool can be adapted to different source modelling schemes and to different destination languages or sublanguages of a domain. It is based on a multilingual approach to natural language generation, realised through a clean separation of the domain model from the linguistic model and their link by well defined structures. Specific knowledge structures and operations have been developed for bridging between the modelling ‘style’ of the conceptual representation and natural language. Using the example of the scheme developed for modelling surgical operative procedures within the GALEN-IN-USE project, we show how the generator is adapted to such a scheme. The basic characteristics of the surgical procedures scheme are presented together with the basic principles of the generation tool. Using worked examples, we discuss the transformation operations which change the initial source representation into a form which can more directly be translated to a given natural language. In particular, the linguistic knowledge which has to be introduced—such as definitions of concepts and relationships—is described. We explain the overall generator strategy and how particular transformation operations are triggered by language-dependent and conceptual parameters. Results are shown for generated French phrases corresponding to surgical procedures from the urology domain.

Modeling Concepts in Medicine for Medical Language Understanding

Over the past two decades, the construction of models for medical concept representation and for understanding of the deep meaning of medical narrative texts have been challenging areas of medical informatics research. This review highlights how these two inter-related domains have evolved, emphasizing aspects of medical modeling as a tool for medical language understanding. A representation schema, which balances partially but accurately with complete but complex representations of domain-specific knowledge, must be developed to facilitate language understanding. Representative examples are drawn from two major independent efforts undertaken by the authors: the elaboration and the subsequent adjustment of the RECIT multilingual analyzer to include a robust medical concept model, and the recasting of a frame-based interlingua system, originally developed to map equivalent concepts between controlled clinical vocabularies, to invoke a similar concept model.

Formal Classification of Medical Concept Descriptions: Graph-Oriented Operators

A crucial component of a medical concept representation system is the classifier. It requires features that are not sufficiently supported by current logic based formalisms like description logics and conceptual graphs. Those features are, for instance, the representation of partitive and spatial relations and their impact on subsumption. This paper introduces graph oriented classification operators for a concept representation language with normal forms. Emphasis is on the separation of generic and partitive relations and on the mutual interdependence of subsumption and part-whole. For that purpose operators are given for formal subsumption, formal part-whole, subsumptive part-whole and part-sensitive subsumption. These operators are based on the formal structure of concept descriptions and on explicitly introduced generic and partitive relationships between their constituents.

The guideline interchange format: a model for representing guidelines.

To allow exchange of clinical practice guidelines among institutions and computer-based applications. The GuideLine Interchange Format (GLIF) specification consists of GLIF model and the GLIF syntax. The GLIF model is an object-oriented representation that consists of a set of classes for guideline entities, attributes for those classes, and data types for the attribute values. The GLIF syntax specifies the format of the test file that contains the encoding. Researchers from the InterMed Collaboratory at Columbia University, Harvard University (Brigham and Women's Hospital and Massachusetts General Hospital), and Stanford University analyzed four existing guideline systems to derive a set of requirements for guideline representation. The GLIF specification is a consensus representation developed through a brainstorming process. Four clinical guidelines were encoded in GLIF to assess its expressivity and to study the variability that occurs when two people from different sites encode the same guideline. The encoders reported that GLIF was adequately expressive. A comparison of the encodings revealed substantial variability. GLIF was sufficient to model the guidelines for the four conditions that were examined. GLIF needs improvement in standard representation of medical concepts, criterion logic, temporal information, and uncertainty.

Dependency parsing for medical language and concept representation

The theory of conceptual structures serves as a common basis for natural language processing and medical concept representation. We present a P rolog-based formalization of dependency grammar that can accommodate conceptual structures in its dependency rules. First results indicate that this formalization provides an operational basis for the implementation of medical language parsers and for the design of medical concept representation languages.

Informatics in the Service of Health, a Look to the Future

The paper attempts a balanced look at the directions of health informatics required in the future. In a high level review of impediments to health and health care a number of issues are identified that may be amenable to improvement by contributions from health informatics. Attention is drawn to the improvement of the collection and dissemination of knowledge in addition to the analysis of morbid conditions as a focus for health informatics. On this basis a review of the current state of health information systems is undertaken. The importance of adaptable user interfaces for end users and systems personnel, privacy and confidentiality protection, and linkage among clinical support systems and knowledge repositories is stressed. These improvements hinge on advancements in medical concept representation. Canadian contributions to these developments, particularly the instigation of Evidence Based Medicine (EBM) are briefly reviewed.

Making the Conceptual Connections: The UMLS after a Decade of Research and Development

The UMLS Section in this issue of JAMIA is dedicated to the memory of Marsden Scott Blois, Jr., a pioneer in medical concept representation. Dr. Blois, an internationally recognized physician-investigator and medical informatician, made important contributions to melanoma research and to biomedical informatics during his distinguished and varied career. His keen insight in the early days of the UMLS project contributed significantly to the future direction and success of the project. This issue of JAMIA helps mark the tenth anniversary of the Unified Medical Language System (UMLS) project. From the beginning, the project has focused on overcoming the barriers users face when attempting to interact with computerized health information systems. The UMLS developers and their collaborators envisioned and then created a set of knowledge sources designed to support the development of sophisticated and accessible information systems. The UMLS knowledge sources have grown to …

Restructuring Routinely Collected Patient Data: ORCA Applied to Andrology

Hospital information systems do not always cover all required detail per specialty. This may lead to scattering of data over disparate systems and the paper record. The ORCA (Open Record for CAre) CPR offers a generic structure for record sharing, and record keeping tailored to specific needs. We studied whether a semantic integration of existing and new data was possible, using the ORCA structure. Existing andrology data, originating from separate sources, were utilized for this purpose. During normalization, validation and explication steps, latent problems in the source data were exposed and removed, followed by a merge with new data items. By conversion of source data to ORCA, a unique representation of medical concepts in the database was attained, facilitating retrieval of univocal data for multiple purposes. We conclude that the expansion to the andrology domain, including transparent integration of existing data, provides support for the generality of ORCA.