The Unified Medical Language System (UMLS) and the UMLS Semantic Network (SN)


Anand Kumar and Barry Smith, “The Unified Medical Language System and the Gene Ontology: Some Critical Reflections”, in A. Günter, R. Kruse and B. Neumann (eds.), KI 2003: Advances in Artificial Intelligence (Lecture Notes in Artificial Intelligence 2821), Berlin: Springer, 2003, 135–148.

The Unified Medical Language System and the Gene Ontology are among the most widely used terminology resources in the biomedical domain. However, when we evaluate them in the light of simple principles for well-constructed ontologies we find a number of characteristic inadequacies. Employing the theory of granular partitions, a new approach to the understanding of ontologies and of the relationships ontologies bear to instances in reality, we provide an application of this theory in relation to an example drawn from the context of the pathophysiology of hypertension. This exercise is designed to demonstrate how, by taking ontological principles into account we can create more realistic biomedical ontologies which will also bring advantages in terms of efficiency and robustness of associated software applications.

Anand Kumar and Barry Smith, “Ontology for Task-Based Clinical Guidelines and the Theory of Granular Partitions”, AIME 2003: Proceedings of the 9th Conference on Artificial Intelligence in Medicine (Lecture Notes on Artificial Intelligence 2780), Berlin: Springer, 2003, 71–75.

The theory of granular partitions (TGP) is a new approach to the understanding of ontologies and other classificatory systems. The paper explores the use of this new theory in the treatment of task-based clinical guidelines as a means for better understanding the relations between different clinical tasks, both within the framework of a single guideline and between related guidelines. We used as our starting point a DAML+OIL-based ontology for the WHO guideline for hypertension management, comparing this with related guidelines and attempting to show that TGP provides a flexible and highly expressive basis for the manipulation of ontologies of a sort which might be useful in providing more adequate Computer Interpretable Guideline Models (CIGMs) in the future. 

Barry Smith, Anand Kumar, and Steffen Schulze-Kremer, “Revising the UMLS Semantic Network,” in M. Fieschi, et al. (eds.), Medinfo 2004, Amsterdam: IOS Press, 1700. Expanded version

The integration of standardized biomedical terminologies into a single, unified knowledge representation system has formed a key area of applied informatics research in recent years. The Unified Medical Language System (UMLS) is the most advanced and most prominent effort in this direction, bringing together within its Metathesaurus a large number of distinct source-terminologies. The UMLS Semantic Network, which is designed to support the integration of these source-terminologies, has proved to be a highly successful combination of formal coherence and broad scope. We argue here, however, that its organization manifests certain structural problems, and we describe revisions which we believe are needed if the network is to be maximally successful in realizing its goals of supporting terminology integration.

Anand Kumar, Matteo Piazza, Barry Smith, Silvana Quaglini, Mario Stefanelli “Formalizing UMLS Relations Using Semantic Partitions in the Context of a Task-Based Clinical Guidelines Model” (MS 2004)

An important part of the Unified Medical Language System (UMLS) is its Semantic Network, consisting of 134 Semantic Types connected to each other by edges formed by one or more of 54 distinct Relation Types. This Network is however for many purposes overcomplex, and various groups have thus made attempts at simplification. Here we take this work further by simplifying the relations which involve the three Semantic Types – Diagnostic Procedure, Laboratory Procedure and Therapeutic or Preventive Procedure. We define operators which can be used to generate terms instantiating types from this selected set when applied to terms designating certain other Semantic Types, including almost all the terms specifying clinical tasks. Usage of such operators thus provides a useful and economical way of specifying clinical tasks. The operators allow us to define a mapping between those types within the UMLS which do not represent clinical tasks and those which do. This mapping then provides a basis for an ontology of clinical tasks that can be used in the formulation of computer-interpretable clinical guideline models.

Barry Smith, “From Concepts to Clinical Reality: An Essay on the Benchmarking of Biomedical Terminologies”, Journal of Biomedical Informatics, 2006; 39(3): 288-298.

It is only by fixing on agreed meanings of terms in biomedical terminologies that we will be in a position to achieve that accumulation and integration of knowledge that is indispensable to progress at the frontiers of biomedicine. Standardly, the goal of fixing meanings is seen as being realized through the alignment of terms on what are called ‘concepts’. Part I addresses three versions of the concept-based approach – by Cimino, by Wüster, and by Campbell and associates – and surveys some of the problems to which they give rise, all of which have to do with a failure to anchor the terms in terminologies to corresponding referents in reality. Part II outlines a new, realist solution to this anchorage problem, which sees terminology construction as being motivated by the goal of alignment not on concepts but on the universals (kinds, types) in reality and thereby also on the corresponding instances (individuals, tokens). We outline the realist approach, and show how on its basis we can provide a benchmark of correctness for terminologies which will at the same time allow a new type of integration of terminologies and electronic health records. We conclude by outlining ways in which the framework thus defined might be exploited for purposes of diagnostic decision-support.