Supporting biomedical ontology evolution by identifying outdated concepts and the required type of change.

The consistent evolution of ontologies is a major challenge for systems using semantically enriched data, for example, for annotating, indexing, or reasoning. The biomedical domain is a typical example where ontologies, expressed with different formalisms, have been used for a long time and whose dynamic nature requires the regular revision of underlying systems. However, the automatic identification of outdated concepts and proposition of revision actions to update them are still open research questions. Solutions to these problems are of great interest to organizations that manage huge and dynamic ontologies. In this paper, we present an approach for (i) identifying the concepts of an ontology that require revision and (ii) suggesting the type of revision. Our analysis is based on three aspects: structural information encoded in the ontology, relational information gained from external source of knowledge (i.e., PubMed and UMLS) and temporal information derived from the history of the ontology. Our approach aims to evaluate different methods and parameters used by supervised learning classifiers to identify both the set of concepts that need revision, and the type of revision. We applied our approach to four well-known biomedical ontologies/terminologies (ICD-9-CM, MeSH, NCIt and SNOMED CT) and compared our results to similar approaches. Our model shows accuracy ranging from 68% (for SNOMED CT) to 91% (for MeSH), and an average of 71% when considering all datasets together.

DyKOSMap: A framework for mapping adaptation between biomedical knowledge organization systems.

BACKGROUND: Knowledge Organization Systems (KOS) and their associated mappings play a central role in several decision support systems. However, by virtue of knowledge evolution, KOS entities are modified over time, impacting mappings and potentially turning them invalid. This requires semi-automatic methods to maintain such semantic correspondences up-to-date at KOS evolution time. METHODS: We define a complete and original framework based on formal heuristics that drives the adaptation of KOS mappings. Our approach takes into account the definition of established mappings, the evolution of KOS and the possible changes that can be applied to mappings. This study experimentally evaluates the proposed heuristics and the entire framework on realistic case studies borrowed from the biomedical domain, using official mappings between several biomedical KOSs. RESULTS: We demonstrate the overall performance of the approach over biomedical datasets of different characteristics and sizes. Our findings reveal the effectiveness in terms of precision, recall and F-measure of the suggested heuristics and methods defining the framework to adapt mappings affected by KOS evolution. The obtained results contribute and improve the quality of mappings over time. CONCLUSIONS: The proposed framework can adapt mappings largely automatically, facilitating thus the maintenance task. The implemented algorithms and tools support and minimize the work of users in charge of KOS mapping maintenance.

A Semantic Big Data Platform for Integrating Heterogeneous Wearable Data in Healthcare.

Advances supported by emerging wearable technologies in healthcare promise patients a provision of high quality of care. Wearable computing systems represent one of the most thrust areas used to transform traditional healthcare systems into active systems able to continuously monitor and control the patients' health in order to manage their care at an early stage. However, their proliferation creates challenges related to data management and integration. The diversity and variety of wearable data related to healthcare, their huge volume and their distribution make data processing and analytics more difficult. In this paper, we propose a generic semantic big data architecture based on the "Knowledge as a Service" approach to cope with heterogeneity and scalability challenges. Our main contribution focuses on enriching the NIST Big Data model with semantics in order to smartly understand the collected data, and generate more accurate and valuable information by correlating scattered medical data stemming from multiple wearable devices or/and from other distributed data sources. We have implemented and evaluated a Wearable KaaS platform to smartly manage heterogeneous data coming from wearable devices in order to assist the physicians in supervising the patient health evolution and keep the patient up-to-date about his/her status.

Understanding semantic mapping evolution by observing changes in biomedical ontologies.

Knowledge Organization Systems (KOSs) are extensively used in the biomedical domain to support information sharing between software applications. KOSs are proposed covering different, but overlapping subjects, and mappings indicate the semantic relation between concepts from two KOSs. Over time, KOSs change as do the mappings between them. This can result from a new discovery or a revision of existing knowledge which includes corrections of concepts or mappings. Indeed, changes affecting KOS entities may force the underline mappings to be updated in order to ensure their reliability over time. To tackle this open research problem, we study how mappings are affected by KOS evolution. This article presents a detailed descriptive analysis of the impact that changes in KOS have on mappings. As a case study, we use the official mappings established between SNOMED CT and ICD-9-CM from 2009 to 2011. Results highlight factors according to which KOS changes in varying degrees influence the evolution of mappings.

Analysis and implementation of the DynDiff tool when comparing versions of ontology.

BACKGROUND: Ontologies play a key role in the management of medical knowledge because they have the properties to support a wide range of knowledge-intensive tasks. The dynamic nature of knowledge requires frequent changes to the ontologies to keep them up-to-date. The challenge is to understand and manage these changes and their impact on depending systems well in order to handle the growing volume of data annotated with ontologies and the limited documentation describing the changes. METHODS: We present a method to detect and characterize the changes occurring between different versions of an ontology together with an ontology of changes entitled DynDiffOnto, designed according to Semantic Web best practices and FAIR principles. We further describe the implementation of the method and the evaluation of the tool with different ontologies from the biomedical domain (i.e. ICD9-CM, MeSH, NCIt, SNOMEDCT, GO, IOBC and CIDO), showing its performance in terms of time execution and capacity to classify ontological changes, compared with other state-of-the-art approaches. RESULTS: The experiments show a top-level performance of DynDiff for large ontologies and a good performance for smaller ones, with respect to execution time and capability to identify complex changes. In this paper, we further highlight the impact of ontology matchers on the diff computation and the possibility to parameterize the matcher in DynDiff, enabling the possibility of benefits from state-of-the-art matchers. CONCLUSION: DynDiff is an efficient tool to compute differences between ontology versions and classify these differences according to DynDiffOnto concepts. This work also contributes to a better understanding of ontological changes through DynDiffOnto, which was designed to express the semantics of the changes between versions of an ontology and can be used to document the evolution of an ontology.

Combining rules, background knowledge and change patterns to maintain semantic annotations.

Knowledge Organization Systems (KOS) play a key role in enriching biomedical information in order to make it machine-understandable and shareable. This is done by annotating medical documents, or more specifically, associating concept labels from KOS with pieces of digital information, e.g., images or texts. However, the dynamic nature of KOS may impact the annotations, thus creating a mismatch between the evolved concept and the associated information. To solve this problem, methods to maintain the quality of the annotations are required. In this paper, we define a framework based on rules, background knowledge and change patterns to drive the annotation adaption process. We evaluate experimentally the proposed approach in realistic cases-studies and demonstrate the overall performance of our approach in different KOS considering the precision, recall, F1-score and AUC value of the system.

Analyzing interactions on combining multiple clinical guidelines.

Accounting for patients with multiple health conditions is a complex task that requires analysing potential interactions among recommendations meant to address each condition. Although some approaches have been proposed to address this issue, important features still require more investigation, such as (re)usability and scalability. To this end, this paper presents an approach that relies on reusable rules for detecting interactions among recommendations coming from various guidelines. It extends a previously proposed knowledge representation model (TMR) to enhance the detection of interactions and it provides a systematic analysis of relevant interactions in the context of multimorbidity. The approach is evaluated in a case study on rehabilitation of breast cancer patients, developed in collaboration with experts. The results are considered promising to support the experts in this task.

Towards natural language question generation for the validation of ontologies and mappings.

BACKGROUND: The increasing number of open-access ontologies and their key role in several applications such as decision-support systems highlight the importance of their validation. Human expertise is crucial for the validation of ontologies from a domain point-of-view. However, the growing number of ontologies and their fast evolution over time make manual validation challenging. METHODS: We propose a novel semi-automatic approach based on the generation of natural language (NL) questions to support the validation of ontologies and their evolution. The proposed approach includes the automatic generation, factorization and ordering of NL questions from medical ontologies. The final validation and correction is performed by submitting these questions to domain experts and automatically analyzing their feedback. We also propose a second approach for the validation of mappings impacted by ontology changes. The method exploits the context of the changes to propose correction alternatives presented as Multiple Choice Questions. RESULTS: This research provides a question optimization strategy to maximize the validation of ontology entities with a reduced number of questions. We evaluate our approach for the validation of three medical ontologies. We also evaluate the feasibility and efficiency of our mappings validation approach in the context of ontology evolution. These experiments are performed with different versions of SNOMED-CT and ICD9. CONCLUSIONS: The obtained experimental results suggest the feasibility and adequacy of our approach to support the validation of interconnected and evolving ontologies. Results also suggest that taking into account RDFS and OWL entailment helps reducing the number of questions and validation time. The application of our approach to validate mapping evolution also shows the difficulty of adapting mapping evolution over time and highlights the importance of semi-automatic validation.

Recognizing lexical and semantic change patterns in evolving life science ontologies to inform mapping adaptation.

BACKGROUND: Mappings established between life science ontologies require significant efforts to maintain them up to date due to the size and frequent evolution of these ontologies. In consequence, automatic methods for applying modifications on mappings are highly demanded. The accuracy of such methods relies on the available description about the evolution of ontologies, especially regarding concepts involved in mappings. However, from one ontology version to another, a further understanding of ontology changes relevant for supporting mapping adaptation is typically lacking. METHODS: This research work defines a set of change patterns at the level of concept attributes, and proposes original methods to automatically recognize instances of these patterns based on the similarity between attributes denoting the evolving concepts. This investigation evaluates the benefits of the proposed methods and the influence of the recognized change patterns to select the strategies for mapping adaptation. RESULTS: The summary of the findings is as follows: (1) the Precision (>60%) and Recall (>35%) achieved by comparing manually identified change patterns with the automatic ones; (2) a set of potential impact of recognized change patterns on the way mappings is adapted. We found that the detected correlations cover ∼66% of the mapping adaptation actions with a positive impact; and (3) the influence of the similarity coefficient calculated between concept attributes on the performance of the recognition algorithms. CONCLUSIONS: The experimental evaluations conducted with real life science ontologies showed the effectiveness of our approach to accurately characterize ontology evolution at the level of concept attributes. This investigation confirmed the relevance of the proposed change patterns to support decisions on mapping adaptation.

A perspective of adaptation in healthcare.

Emerging new technologies in healthcare has proven great promises for managing patient care. In recent years, the evolution of Information and Communication Technologies pushes many research studies to think about treatment plan adaptation in this area. The main goal is to accelerate the decision making by dynamically generating new treatment due to unexpected situations. This paper portrays the treatment adaptation from a new perspective inspired from the human nervous system named autonomic computing. Thus, the selected potential studies are classified according to the maturity levels of this paradigm. To guarantee optimal and accurate treatment adaptation, challenges related to medical knowledge and data are identified and future directions to be explored in healthcare systems are discussed.

The influence of similarity between concepts in evolving biomedical ontologies for mapping adaptation.

Biomedical ontologies continuously evolve which demands maintain associated mappings up-to-date. This article studies whether similarity calculated between values of concept attributes issued from successive ontology versions plays a role in deciding mapping adaptation actions. We empirically analyse the evolution of official mappings established between large biomedical ontologies. The results point out the relevance of this factor for mapping adaptation.

Characterizing semantic mappings adaptation via biomedical KOS evolution: a case study investigating SNOMED CT and ICD.

Mappings established between Knowledge Organization Systems (KOS) increase semantic interoperability between biomedical information systems. However, biomedical knowledge is highly dynamic and changes affecting KOS entities can potentially invalidate part or the totality of existing mappings. Understanding how mappings evolve and what the impacts of KOS evolution on mappings are is therefore crucial for the definition of an automatic approach to maintain mappings valid and up-to-date over time. In this article, we study variations of a specific KOS complex change (split) for two biomedical KOS (SNOMED CT and ICD-9-CM) through a rigorous method of investigation for identifying and refining complex changes, and for selecting representative cases. We empirically analyze and explain their influence on the evolution of associated mappings. Results point out the importance of considering various dimensions of the information described in KOS, like the semantic structure of concepts, the set of relevant information used to define the mappings and the change operations interfering with this set of information.