Elsevier

Expert Systems with Applications

Volume 36, Issue 10, December 2009, Pages 12214-12224
Expert Systems with Applications

Medical knowledge management for specific hospital departments

https://doi.org/10.1016/j.eswa.2009.04.064Get rights and content

Abstract

Medical knowledge representation and management is concerned with how to organise the often vague clinical experience of medical staff required for computable models. However, few knowledge management and acquisition tools have entered routine use, since such tools are not perceived by physicians as part of the clinical information process. An attempt to partially solve this problem, we identify two key aspects of knowledge representation and management tasks. The first is to adopt a medical knowledge standardisation to provide a consistent terminology control and to simplify the integration between knowledge management tools and the health information system. The second is to establish an effective knowledge acquisition process in specific medical fields by adapting knowledge acquisition tools. Therefore, the main goal of this work is to define computational models and to design mechanisms for the effective acquisition and management of medical knowledge in real-life hospital departments. To this end, we analyse the representation of medical knowledge (based on deep-causal models) and the development of knowledge management tools (based on ontologies), integrated within the information processing activities of the clinical user. Finally, we illustrate its applicability in the Intensive Care Unit and Pediatry scenarios.

Introduction

Ongoing advancements in biomedical computer-based systems have opened up new perspectives on medical activity (especially in hospitals) due to the availability of large volumes of clinical data from Electronic Health Records (EHRs), medical articles and clinical trials (Bali and Dwivedi, 2005, Wyatt, 2001). Despite the huge amount of information produced by physicians in their daily activities (e.g. clinical evidence, diagnosis or the treatment provided), much key information is not registered, for instance why a physician selected a particular therapy protocol, the temporal evolution of a specific disease, or which literature was consulted by the physician to obtain the diagnosis of a patient. In other words, much knowledge generated in clinical practice is getting lost.

In Shortliffe and Barnett (2006, chap. 2), the authors state that the problem is not unlike that of identifying what the user needs to interpret, internalize and apply from the wealth of information in a report. The creation of a computer-based system that manages knowledge requires substantial modelling activity: deciding what clinical distinctions and patients are relevant, and identifying the concepts and relationships amongst them. However, most of the medical data structured in hospitals is focused towards EHR, which is not structured in a way that can be reused for decision making, and is often redundant (McKibbon et al., 1990).

As far as we are concerned, medical services are work-overloaded environments where time is often critical and the information must be available to make correct decisions. In order to improve medical decisions, it is necessary to provide the correct knowledge at the correct time. Due to the daily use of computers in medical activities, there is a large amount of potential biomedical knowledge that could be recorded to improve patients Health Care quality (Black, 1997).

Medical science fields are not unaware of this reality and some proposals have been made. In particular, Evidence-Based Medicine (EBM) (Sackett, Scott Richardson, Rosenberg, & Brian Haynes, 1997) faces this problem and identifies the need to combine the physicians own experience with the best scientific evidence from the medical community. In this sense, there are several proposals in the literature dealing with medical information-sharing by means of using terminological standards and an international description of pathologies (Regenstrief Institute, 2006, SNOMED International, 2003, World Health Organization, 1992).

Therefore, we consider that medical knowledge is double-valued due to its consequences for patients and the difficulty of its acquisition in a computable structure. This difficulty lies in the nature of medical knowledge (imprecise, incomplete, the many years needed to master such knowledge, and it is multidisciplinary nature). In particular, medical specialities also focus, apart from their clinical content, on different aspects of knowledge. For instance, paediatrics deals with young-patient general pathologies, whilst Intensive Care Units face the temporal evolution of critical patients (Palma, Juárez, Campos, & Marín, 2006).

In a more general perspective, business organizations and big companies have also coped with a similar problem in the last two decades (Carlucci and Shiuma, 2006, Hall, 1993). Knowledge management (KM) and knowledge management systems (KMS) have been demonstrated to provide an effective stimulus for organizations to structure, mobilize and reuse knowledge, stored in a knowledge base (KB), resulting in improved performance (Carlucci & Shiuma, 2006). Thus, it seems reasonable to think that, after a carefully adaptation of KM and KMS, this strategy could be valid to clinical environments.

At this point, we identify three factors considered essential for building KM and KMS integrated in medical domain: (1) the need to representing specific knowledge depending on the medical field (e.g. atemporal or temporal aspects), (2) the need to implement effective knowledge acquisition tools (KAT) adapted to the medical service in order to build the KB, and (3) the need to share a common medical terminology.

Of the wide range of artificial intelligence (AI) approaches available to represent knowledge, only the model-based reasoning (MBR) approach tackles the problem of troubleshooting systems by starting from their behaviour (structure and function) and extending their formalisms to include additional knowledge, such as the temporal dimension. Therefore, MBR provides deep-causal models to describe the medical knowledge, in contrast to the classical rule-based systems (Brusoni et al., 1998, Torasso, 2001), that is, MBR describes all related factors of a single disease using an atomic piece of knowledge (model) instead of a set of them (rules).

In order to effectively build a medical KB using the aforementioned models, knowledge acquisition tools (KATs) are required to help physicians to extract and formalise their tacit knowledge using these formal structures. Significant advances have been achieved in this field through the years, clinical decisions based on knowledge having been seen to be effective (Dojat et al., 2000, Lavrac and Mozetic, 1992, Shahsavar et al., 1991, Shortliffe, 1976). Nonetheless, few systems have so far entered routine use. In Kalogeropoulos (2001), the author identifies one main problem: KMSs and KATs are isolated from the clinical environment itself and they are perceived as experimental entities. Therefore, new efforts must be made to obtain effective KAT and KMS in medical environments to build medical KBs.

The main purpose of this work is to demonstrate in practice how deep-causal models (MBR) could be used to describe medical knowledge effectively through the development of KAT integrated within the information processing activities of the clinical user. These tools have been adapted to two different real medical environments: an Intensive Care Unit (ICU) and Paediatrics. The reminder of this paper is organised as follows. In Section 2, previous research works on medical knowledge management are reviewed. Our experiences on knowledge acquisition in an ICU are described in Section 3. In Section 4, we propose a deep-causal model and a knowledge management tool for the paediatric service. Finally, Section 5 presents our conclusions and suggests future works.

Section snippets

Medical knowledge representation: background

Medical knowledge representation is concerned with how to organise the often vague and unanalysed ideas of clinical staff required for computable models. Knowledge modelling should not be regarded as a process of mapping expert knowledge for computational representation, but as a model-building process (Taylor, 2006, chap. 8).

In medical domains, we consider that this issue (modelling) should cover two key aspects: (1) standardising medical knowledge and (2) constructing computable models using

ICU knowledge acquisition tool

The Intensive Care Unit (ICU) is a hospital department that provides critical attention to medically recoverable patients. One of the fundamental characteristics of the ICU is that patients require a permanent availability of monitoring equipment and specialist care. Thus, the temporal evolution of patients is permanently recorded and analysed by physicians, who must tackle a wide range of patient pathological problems (e.g. cardiovascular, renal, infections, neurological, etc.). The temporal

Knowledge acquisition in paediatrics: WOMKA

Paediatrics is the branch of medicine concerned with the development, care, and diseases of infants and children. Unlike adult medicine, young bodies are involved in different maturation processes and, therefore, the neonate physiology differs from children and adolescents. From the point of view of the hospital departments, the paediatric service also differs from the rest of the departments since this paediatric service deals with the medical care of a range-limited old-age population, but

Conclusions and future works

The main objective of this work is to design mechanisms for the effective acquisition and management of medical knowledge in real-life hospital departments. In particular, we approach this problem by representing medical knowledge using deep-causal models (TBM and SBM models) for two hospital departments: ICU and paediatrics with different requirements. Firstly, to manage medical knowledge in the ICU, we designed and implemented the KAT named CATEKAT2, which is focused on building the TBM and

Acknowledgements

We would like to thank Dr. Palacios from the ICU of the Getafe Hospital (Madrid, Spain); Dr. Carmona, head of the Paediatric Service of the Virgen de la Vega Clinic (Murcia, Spain); and A. Jimenez head of the computer service of ASISA clinics.

This work has been partially supported by contributions from the contract between the University of Murcia and the Clinica Virgen de la Vega (No. 9821), the PETRI Project PET2007_0033, the Spanish MEC Project TIN2006-15460-C04-01, and the Excellence

References (47)

  • S.S. Abidi et al.

    Knowledge management in pediatric pain: Mapping on-line expert discussions to medical literature

    Medinfo

    (2004)
  • Bali, R. K., & Dwivedi, A. (2005). Clinical knowledge management: An overview of current understanding. In L. Bos, S....
  • J.B. Bard et al.

    Ontology in biology: Design applications and future challenges

    Nature Reviews Genetics

    (2004)
  • N.A. Black

    Developing high-quality clinical databases the key to a new research paradigm

    British Medical Journal

    (1997)
  • D. Carlucci et al.

    Knowledge asset value spiral: Linking knowledge assets to company’s performance

    Knowledge and Process Management

    (2006)
  • M. Casella Dos Santos et al.

    Medical natural language processing enhancing drug ordering and coding

  • DICOEMS Consortium (2007). Dicoems project....
  • M. Dojat et al.

    Clinical evaluation of a computer-controlled pressure support mode

    American Journal of Respiratory and Critical Care Medicine

    (2000)
  • Dubois, D., Lang, J., & Prade, H. (1994). Possibilistic logic. In C. J. Hogger, D. M. Gabbay, & J. A. Robinson, (Eds.),...
  • Gangemi, A., Pisanelli, D. M., & Steve G. (1998). Some requirements and experiences in engineering terminological...
  • Gene Ontology Consortium (2007). Gene ontology....
  • R. Hall

    A framework linking intangibles resources and capabilities to sustainable competitive advantage

    Strategic Management Journal

    (1993)
  • HL7 (2007). Health level 7....
  • Cited by (17)

    • Automated experiential engineering knowledge acquisition through Q&A contextualization and transformation

      2016, Advanced Engineering Informatics
      Citation Excerpt :

      Knowledge representation is important as it affects knowledge’s expressiveness, readability and machine operability. The if-then rules are the most widely adopted knowledge representation method because it makes a balance between the machine operability and the human readability [13,20–22,28,30]. Concept maps denote the knowledge concepts connected by relations with nodes and edges [2,3,20,27].

    • Collaboration-based medical knowledge recommendation

      2012, Artificial Intelligence in Medicine
      Citation Excerpt :

      Conventional medical knowledge management systems provide search functions to help clinicians to find needed knowledge. It is not convenient for them to retrieve appropriate knowledge quickly and accurately, especially within the explosion of knowledge nowadays [8,9]. Most existing clinical information systems adopt simple approaches to search medical knowledge, e.g., keyword-based knowledge search [10,11].

    • Study on intelligent home care platform based on chronic disease knowledge management

      2016, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    View all citing articles on Scopus
    View full text