- Open Access
An ontological modeling approach for abnormal states and its application in the medical domain
© Yamagata et al.; licensee BioMed Central Ltd. 2014
- Received: 17 September 2013
- Accepted: 7 May 2014
- Published: 21 May 2014
Recently, exchanging data and information has become a significant challenge in medicine. Such data include abnormal states. Establishing a unified representation framework of abnormal states can be a difficult task because of the diverse and heterogeneous nature of these states. Furthermore, in the definition of diseases found in several textbooks or dictionaries, abnormal states are not directly associated with the corresponding quantitative values of clinical test data, making the processing of such data by computers difficult.
We focused on abnormal states in the definition of diseases and proposed a unified form to describe an abnormal state as a “property,” which can be decomposed into an “attribute” and a “value” in a qualitative representation. We have developed a three-layer ontological model of abnormal states from the generic to disease-specific level. By developing an is-a hierarchy and combining causal chains of diseases, 21,000 abnormal states from 6000 diseases have been captured as generic causal relations and commonalities have been found among diseases across 13 medical departments.
Our results showed that our representation framework promotes interoperability and flexibility of the quantitative raw data, qualitative information, and generic/conceptual knowledge of abnormal states. In addition, the results showed that our ontological model have found commonalities in abnormal states among diseases across 13 medical departments.
- Abnormal state
With the development of newer technologies, data and information exchange have been required for several applications such as electronic health records (EHR) in medicine. Such data and information include abnormal states. However, abnormal states are difficult to share because of their heterogeneity, caused by the variety of grain sizes, from the level of cells, tissue, and organs to that of the entire human body. This results in diverse representations with little uniformity. BFO [1, 2] and DOLCE  have contributed to the formalization of the quality description of entities. BFO provides E (Entity), P (Property) (e.g., <Eye (E), red (P)>) and DOLCE provides E (Entity), A (Attribute), V (Value) triple (e.g., <esophagus (E), length (A), short (V)>). However, we found that there are more complicated forms of quality representations in medicine. For example, “hypertension” is a compound concept, which has three elements: blood, pressure, and high joined to form one concept/word. Another example is “hyperglycemia,” composed of four concepts: blood, glucose, concentration, and high. Furthermore, in the case of “intestinal polyposis,” it is unclear whether “intestine” or “polyp” should be considered as the entity.
This motivated us to establish a common framework for the representation of abnormal states supported by sound theories. In this study, we investigate the representation of abnormal states from the content-oriented view, which focuses on how to capture the content to be represented, on the basis YAMATO .
Quality-related concepts (dependent continuant entities) are divided into “Property,” “Generic quality,” and “Quality value.” “Quality” in BFO is identical to “Property” in YAMATO.
“Quality value” is classified in a manner identical to the classification in scales of measurement.
The context dependency of “Ordinal value” is represented by using the theory of “Role”.
Multiple kinds of informational entities are symbolically represented.
For the quality description, representations with both EAV and EP formalisms are defined. Furthermore, PATO2YAMATO aims to integrate phenotype >descriptions that exist in different structured comparison contexts . It allows (1) the classification of quality values, in which scales of measurements are properly represented; (2) strict modeling of the context dependency of ordinal values; and (3) clear distinction between “true values” and “measured data.” It provides the mapping of ontology terms of PATO  to YAMATO’s framework and enables the interoperability of the quality framework between different top-level ontologies such as BFO and DOLCE. For example, in the YAMATO framework, PATO:0000582 (increased weight) is defined as a Property that is a combination of Generic quality (Attribute), weight, and a context-dependent Quality value (Attribute Value), heavy. The context-independent value is defined as a class “Weight quality value.”
Another issue is that in several medical textbooks or medical dictionaries, abnormal states in the definitions of diseases have not been directly associated with the corresponding quantitative values of clinical test data (e.g., “ischemia” in ischemic heart disease or “muscular weakness” in muscular dystrophy), which makes their processing by a computer difficult.
Furthermore, clinicians often deal with abnormal states specific to each disease only in a particular medical division, which makes it difficult to spread awareness regarding the common nature of abnormal states. To address these issues, we have been developing abnormality ontology for the systematization of knowledge regarding abnormal states, using ontological engineering, which represents a unified framework . We focus on abnormal states in the definitions of diseases, which should be referred to in several applications. In addition, we discuss the representation of the various abnormal states on the basis of ontological theories in a consistent manner.
Our claim in this study is not isolated to adopting one of the representational forms used in the existing resources. The aim of our work is to formalize and organize different representations used in clinical medicine on the basis of ontological theories, and to realize the interoperability between them. It's not a simple matter of the use of existing resources such as PATO, LOINC , and others. Unified theoretical considerations make the various representation forms interoperable, which enables the establishment of a consistent and computer understandable model for abnormal states that are used in the definitions of diseases and medical data.
In this study, we first define abnormal states and explain our representation model. Then, we introduce our ontology of abnormal states and demonstrate an application of our work. We have constructed a disease ontology and captured a disease as one or more causal chains of the abnormal states in the human body . Till date, clinicians have described the causal chains of approximately 21,000 abnormal states for approximately 6,000 diseases across 13 medical departments. Thus, we believe that the use of our ontology will contribute to various clinical applications.
Definition of abnormal states
In the human body, abnormal states are highly diverse and involve various grain sizes, from the level of cells, tissue, and organs to that of the whole organism. Therefore, to systematize the knowledge about abnormal states, it is important to clarify the essential characteristics of the abnormal states, and to conceptualize them in a consistent manner.
In this section, we focus on the abnormal states that appear in the definition of diseases rather than in reality. A “state” is modeled as a time-indexed propertya that is associated with an entity, and has the value of an attribute that changes with time . For example, imagine the state of hunger. It is represented by “being hungry” or not at some time point in time. We define “Property” as a characteristic that is inherent in an entity, having an attribute along with its value, such as “being red”: <color, red>. Properties are distinct from attribute values; for example, the Property “hypertension” is differentiated from an Attribute Value such as “high” as in “blood pressure is high.” An Attribute Value has three subclasses: categorical value (e.g., viviparous/oviparous), quantitative value (e.g., 160 mmHg), and qualitative value (e.g., high/low, large/small, much/few). On one hand, the Attribute Value “high” can be used for several attributes such as temperature, density, and velocity. On the other hand, the Property “hypertension” cannot be used for the values of the abovementioned attributes and it has a set of attributes and values like < pressure, high >.
In several textbooks and dictionaries, diseases have been defined in terms of abnormal states. For example, the definition of diabetes is “Diabetes mellitus is characterized by chronic hyperglycemia with disturbances of…” . Another disease myocardial ischemia is presented as “The term acute myocardial infarction should be used when there is evidence of myocardial necrosis with acute myocardial ischemia” . Therefore, we can say that a disease can be defined in terms of an assertion about the patient “being in an abnormal state or not”.
Quantitative representation (e.g., blood pressure is 180 mmHg, blood glucose concentration is 135 mg/dL).
Qualitative representation (e.g., blood pressure is high, blood glucose concentration is high).
Property representation (e.g., hypertension, hyperglycemia).
Because the upper ontology YAMATO  has been carefully designed to cover the property, quality, and quantity ontologies, it supports our work on abnormal states.
A quantitative representation is important for diagnosis because a concrete value should be identified by clinical examination for each patient. However, in the definition of a disease, a property such as “being hypertensive” or “being hyperglycemic” is essential instead of quantitative data. Thus, as our basic policy, we first capture the abnormal states as properties, represented by a tuple like < Property (P), Property Value (Vp)>. The Property Value takes a binary value, i.e., <true/false>. For example, if the state “stenosis” exists, it is described as < stenosis, true>. In addition, when necessary, a Degree Value (Vd) can be used for describing the degree of the Property Value, such as < stenosis, severe >.
Some readers may think that a property represented in the above manner is extremely conceptual to be of practical use because of the lack of a representation, which would give a more concrete meaning to data. Therefore, we specify a property by decomposing it into a tuple: <Attribute (A), Attribute Value (V)>. The Attribute Value can be either a Qualitative Value (Vql) or a Quantitative Value (Vqt). For example, in a case of a qualitative representation, stenosis (P) is decomposed into < cross sectional area (A), small (Vql)>, and in another case of a quantitative representation, stenosis (P) is described as a concrete value, e.g., <cross sectional area (A), 5 mm2 (Vql)>. This approach contributes to promoting consistency in representation, as well as the interoperability between the quantitative raw data and the generic/conceptual knowledge regarding abnormal states (see after the section Interoperability between properties and attributes).
In clinical medicine, decomposition of some properties cannot be achieved, because the precise mechanisms in the human body have not yet been completely uncovered. For example, in the case of nausea, property representation could be nondecomposable. Whether such abnormal states represented in terms of properties defined above can be decomposed into a known attribute and its value will depend on advances in medicine.
Representation of abnormal states
In this section, we introduce our representation model for clinical abnormal states and show that we can appropriately represent them in a consistent manner.
Representations of abnormal states
Abnormal states Property (P)
Property Value (Vp)
Attribute Value (V)
Gasrtectasia (gastric dilation)
(b) Variant of Ratio
Abnormal states Property (P)
Property Value (Vp)
Attribute Value (V)
m/n (no unit)
High m ratio
m/n (focused on m of same object)
High m ratio
High Albumin ratio
m/n (focused on the ratio of same object)
High m/n ratio
Increased A/G ratio
We recognize that some properties may be difficult to decompose into the basic triple representation, such as a ratio and what we call a meta-attribute, discussed below. Accordingly, we introduce a “Sub-object” (SO) to represent a focused object (see next paragraph) as an extended representation, so that a property can be decomposed into a quadruple: <Object (O), Sub-Object (SO), Attribute (A), Attribute Value (V) >.
In the case of a ratio, in addition to identifying the target object with the ratio, it should represent for what will be focused on (“focused object”). Therefore, we introduce a Sub-Object (SO) to represent a focused object. For example, the representation of “hyperglycemia” is a quadruple, <blood (O), glucose (SO), concentration (A), high (V)>, where the Object is blood and the Sub-Object is glucose [Table 1(a), row 3].
There appear to be different kinds of ratios depending on what is focused on. As a result, the Object and Sub-Object vary according to the type of ratio. Our representation model can represent all of them, as shown in Table 1(b). A detailed discussion can be found in a report by Yamagata Y et al. .
Furthermore, we show the representation of a meta-attribute. In the case of the property “gastric polyposis,” although color and size are attributes of polyps, “many polyps” is not an attribute of “polyps” because it is not inherent in each polyp. Following the meta-attribute approach in YAMATO, where, in the case of “the road is curvy,” “number of curves” is identified as a meta-attribute of curves, and the road, which has many curves, can be represented in terms of it (the number curves). Accordingly, we regard “the number of polyps” as a meta-attribute of polyps and the stomach can be described in the same manner as a road. By introducing “Sub-object,” the property “gastric polyposis” can be decomposed into a quadruple < stomach (O), polyps (SO), number (A), many (V)>, where the stomach is identified as the Object, and the polyps as the Sub-object [Table 1(a), row 4].
Interoperability between properties and attributes
Our claim in this clause is the interoperability between the abnormal states and data. Considering the interoperability between the abnormal states and clinical test data, the OP form itself may not be compatible with the observational data.
A large amount of clinical test data is stored in hospitals. To ensure the cross-compatibility between those data and the abnormal states (described in EHR), the unified form should be required for computer processing, so that an exchange mechanism between the OP and OAV is indispensable.
The OAV form can deal with both the quantitative and qualitative representation of values. Most clinical test data are quantitative, e.g., the arterial cross sectional area of 24 mm2 and can be represented by OAVqt as < artery (O), cross sectional area (A), 24 mm2 (Vqt) > in our model. Notably, the quantitative value can be converted to a qualitative value such as “small (Vql),” with the threshold given by each hospital.
In the case of an abnormal state “arterial stenosis,” we can guarantee the interoperability between the quantitative data and abnormal state by decomposing “stenosis” into “cross sectional area is small”, which is represented as < artery (O), cross sectional area (A), small (Vql) >.
Another example is the quantitative data “blood glucose concentration level is 260 mg/dL”. It is represented by < blood (O), glucose (SO), concentration (A), 260 mg/dL (Vqt)>, and the interoperability between the quantitative test data and the abnormal states “hyperglycemia” used in the definition of diabetes is realized via qualitative representation such as < blood (O), glucose (SO), concentration (A), high (Vql) > in the extended OSoAV form.
Furthermore, another issue is the requirement of the degree value of the property. Clinicians usually need to transform test data into abnormal states in the case report. Imagine a case where a highly elevated value is observed in the clinical test. A simple OP < O, P, true > does not satisfactorily capture such data, and thus the degree of abnormal states may be needed. Therefore, we introduce the Degree Value (Vd) like “severe”. Therefore, we can describe such data in terms of the degree value in the OPVd form < blood (O), hyperglycemia, severe (Vd) > in a triplet like in the OAV form.
We recommend that the degree value should only have minimum variations such as mild/moderate/severe for representation because numerous degree values would lead to dispersion and destruction of the unified representation. Taking the interoperability into account, it would be considered preferable to decide how to set a threshold for determining the degree value of “severe”. However, because such concrete threshold values tend to change with time, such threshold values are left undetermined in this study.
Here we introduce Property Value (Vp). A Property has a meaning of < hyperglycemia, existence > or < hyperglycemia, true>. The value as “existence/nonexistence” or “true/false” should be independent of the degree value. However, because adding another form to the degree value would make the system of representation forms more complex, we deal with the degree value (e.g., mild/moderate/severe) as the specialization of the state of hyperglycemia; we treat these values of “existence/nonexistence” or “true/ false” in the same manner as the degree value. Consequently, we use Vp to represent both Vd and Vp.
The use of the value is e.g., a representation of the condition of “not being hyperglycemia,” < hyperglycemia, false>, in latent diabetes. For this reason, we consider that the true/false value is needed for computer processing.
The < OAV > form as clinical test data
The < OPV > form as abnormal states and
The extended form < OSoAV > as clinical test data and < OSoPV > as abnormal states
Consequently, the OPV form is the same as the OAV form. Therefore, the OPV can be compliant with the unified representation (OAV triplet), which realizes the interoperability between the data and abnormal states.
Is-a Hierarchy of abnormality ontology
Ontologically, the structural abnormality, dysfunction, pathological conditions, pathological processes, etc. are kinds of abnormal states. We propose abnormality ontology, which is a comprehensive ontology that covers all of the abovementioned concepts, and abnormal states are defined as the top level.
Clinicians work with strongly domain-specific knowledge, which causes difficulties in finding common and generic knowledge across domains. A clear distinction between the basic/generic and specific concepts is required to be made. To this end, we propose the following three levels of abnormal states (Figure 1):
Level 1: Generic abnormal states
Level 2: Object-dependent abnormal states
Level 3: Specific context-dependent abnormal states
Level 1: generic abnormal states
Level 1 defines very basic (or generic) concepts, which do not depend on any structural entity, i.e., object-independent states. Examples include deformation, additional/missing anatomical structures, translocation, and dysfunction, which are commonly found in several objects, and can be usable in several domains besides medicine, such as machinery, materials, and aviation.
A functional abnormality is defined as an abnormal state that is related to an impaired function and is classified into hyperfunction and malfunction. Malfunction is subcategorized into dysfunction, function arrest, and hypofunction.
Other abnormal states include parametric abnormalities, which are classified into increased or decreased parameters, depending on whether or not the attribute has a higher or lower value than a threshold level. Examples included increased/decreased pressure or increased/decreased weight.
Our model has a recursive structure, in which the generic abnormal states at Level 1 are referred to by Level 2 object-dependent abnormal states.
Level 2: Object-dependent abnormal states
Level 2 defines object-dependent abnormal states. The top level concepts at Level 2 are dependent on generic structures, such as “wall-type structure,” “tubular structure,” and “bursiform structure,” which are common and are used in several domains. Level 2 has been developed by identifying the target object and specializing generic abnormal states at Level 1 with consistency. For example, by specializing “small in area” at Level 1, “narrowing tube,” where the cross-sectional area has become narrow, is defined at Level 2 this is further specialized in the definitions “oil pipe narrowing” or “tracheal stenosis.”
In the lower layer of Level 2, abnormal states that are dependent on medical domain-specific objects, such as human anatomical structures, are defined and designed to represent concepts at all required granularities in the medical domain. Here in general, one problem arises in how fine the level of granularity needs to be supported in our ontology. In the case of “stenosis,” the term, “coronary artery stenosis” in a specific organ (the coronary artery) may be redundant. However, it is noteworthy that the abnormal states in one anatomical object can influence the adjacent objects, which causes other abnormal states. For example, although both are types of stenosis, coronary artery stenosis is different from rhinostenosis because the former causes myocardial ischemia and ischemic heart disease, whereas the latter causes sleep apnea. Therefore, there is a need for distinct abnormal states at specific organ levels.
From an ontological engineering point of view, our framework for modeling abnormal states is intended to capture the abnormal states from generic to specific levels, so as to provide abnormal states at the required granularity of specific organ/tissue/cell layers in the medical domain.
Here, the abnormal states of a specific object defined at Level 2 should be distinct from the disease-dependent concepts at Level 3. For example, hyperglycemia is defined in a context-independent manner at Level 2, and this is referred to in Level 3 concepts in various diseases, such as diabetes, metabolic syndrome, and lipodystrophy.
Currently, we are developing and enriching Level 2 concepts to link each Level 3 concept to the upper common level concept.
Level 3: specific context-dependent abnormal states
Level 3 consists of context-dependent abnormal states, which refer to the Level 2 abnormal states, and are specialized into specific disease-dependent ones. For example, “rectal stenosis,” which is dependent on the rectum at Level 2, is defined as a constituent of Crohn’s disease at Level 3; this is also defined as a cause or an effect of other diseases, such as rectal cancer, Hirschsprung disease, or intestinal tuberculosis.
Causal chains of disease
We have been developing a disease ontology, in which a disease is defined as a causal chain of abnormal states . We divided the diseases into two major kinds: (1) ones where the etiological and pathological processes are well understood and (2) otherwise. Case (2) includes the so-called syndromes, typically represented in terms of the criteria for diagnosis. Diseases of type (1) is identified by its inherent etiological/pathological process(es). In this paper, we deal only with type (1) diseases. After careful examination of several diseases, we believe that every disease of type (1) should have a cue for identification. This means that we should be able to find the so-called main pathological/etiological condition(s), which theoretically characterize the disease to identify it. We know that diseases of type (2) necessarily employ criteria for diagnosis to identify the disease because of the lack of knowledge regarding etiological/pathological processes.
In addition, we believe that we need a formulation for organizing diseases in an is-a hierarchy in a disease model. According to the definition of a disease, this would consist of a causal chain(s), which consisted of nodes and links; a disease would be represented as a directed acyclic graph (DAG). We can introduce the is-a relation between diseases using the inclusion relationship between the causal chains as noted below.
Is-a relation between diseases
Disease A is a super class of disease B if all the causal chains at the class level of disease A are included in those of disease B. The inclusion of nodes (disorders) is judged by considering the is-a relation between the nodes into account as well as the sameness of the nodes.
Core causal chain of a disease
Causal chain(s) of a disease included in the chains of all its subclass diseases is called the core causal chain of a disease.
Derived causal chains of a disease
Causal chains of a disease defined as possible causal chains of abnormal states are called derived causal chains.
Till date, clinicians have described the causal chains of diseases and abnormal states. We have been using these abnormal states to develop an is-a hierarchy of abnormalities. Abnormal states used in disease definitions in the ontology are defined as abnormal states at Level 3, where clinicians defined diseases in the respective medical departments. We collected all causal relationships from all disease concepts defined in the 13 medical departments and combined the causal chains, including the same abnormal states. As a result, the generic causal chains that contain all causal relationships, including approximately 21,000 abnormal states from 13 medical departments have been generated . For example, we assume that a cardiovascular specialist in the division of cardiovascular medicine describes “coronary artery stenosis” and its causal chain as < coronary artery stenosis → myocardial ischemia → myocardial hypoxia > in ischemic cardiac disease. This can be linked with “coronary artery stenosis” in other diseases (e.g., hyperlipidemia) in other departments (metabolic medicine). As a result, a generic causal chain < accumulation of cholesterol → coronary artery stenosis → myocardial ischemia → myocardial hypoxia > of hyperlipidemia can be obtained as a possible causal relationship of abnormal states in the disease.
We proposed a representation model of abnormal states designed in a unified manner. Our medical ontology project was started seven years ago. Since then, it has been refined and revised several times through discussion with both ontologists and clinicians. Till date, we have applied this model to approximately 21,000 abnormal states from approximately 6000 diseases.
We have demonstrated that our model has interoperability between quantitative and qualitative data and the conceptual knowledge of abnormal states in the definition of diseases. With this model, we have been developing an ontology of abnormal states from generic to specific levels. In the application we considered, we built disease chains consisting of causal relationships of abnormal states. By combining the disease chains and the ontology, we have captured all causal relations of the 21,000 abnormal states in the 6,000 diseases across 13 medical departments.
Although abnormal states have traditionally been considered to be specific to each disease in a particular medical department, our approach has found commonalities among abnormal states across medical departments.
Data sources for representation and ontology development of abnormal states
Medical doctors of The University of Tokyo Hospital described the disease ontology, and definitions of diseases were determined. Medical dictionaries [10, 27] and textbooks [28, 29] were used as references. Using them as resources, the definitions of the abnormal states in the definition of diseases were decided through repeated discussions by ontology engineers and medical experts. The Level 1 generic concepts were based on YAMATO ; the top level concepts at Level 2, i.e., generic object-dependent concepts, have been generated by ontologists. In our Japanese medical ontology project, anatomical entities have been defined , and their anatomical structure-dependent abnormal states have been defined by the medical experts. Level 3 disease-specific abnormal states were collected and generated by clinicians in 13 special fields.
Ontology editing tool
Visual input/editing tool
Although it has been implemented as a client application using HOZO’s ontology API, we have published the part of disease ontology as Linked Open Data with SPARQL endpoint to get their causal chains . The Disease Chain LOD is available at the URL http://lodc.med-ontology.jp/.
aOntologically, a state corresponds to a time-indexed property. However, in the definition of diseases, time-indexed abnormal states are rarely used (e.g., ischemia in the definition of ischemic heart disease)” because the time-dependence of properties of diseases is out of the scope of our project.
bIn this paper, although we mainly deal with a simple parameter, additional parameters can be used when necessary. For example, in the definition of neonatal anemia, we associate an additional attribute “age” and the value “neonatal” with it.
A part of this research is supported by the Ministry of Health, Labor and Welfare, Japan, through its “Research and development of medical knowledge base databases for medical information systems” and by the Japan Society for the Promotion of Science (JSPS) through its “Funding Program for World-Leading Innovative R&D on Science and Technology (FIRST Program).” We are deeply grateful to medical doctors (Natsuko Ohtomo, Aki Hayashi, Takayoshi Matsumura, Ryota Sakurai, Satomi Terada, Kayo Waki, et al.), of The University of Tokyo Hospital for describing disease ontology and providing us with broad clinical knowledge. We also would like to thank other team members, Drs. Yoshimasa Kawazoe, Masayuki Kajino, and Emiko Shinohara from The University of Tokyo for useful discussions related to biomedicine.
- Grenon P, Smith B, Goldberg L: Biodynamic ontology: applying BFO in the biomedical domain. Stud Health Technol Inform. 2004, 102: 20-38.Google Scholar
- Basic formal ontology. [http://www.ifomis.uni-saarland.de/bfo/]
- Guarino N: Some ontological principles for designing upper level lexical resources. Proc. of International Conference on Lexical Resources and Evaluation 1998, Granada, Spain. Edited by: Rubio A, Gallardo, Castro, Tejada. 1998, 527-534.Google Scholar
- Mizoguchi R: YAMATO: Yet another more advanced top-level ontology. Proceedings of the Sixth Australasian Ontology Workshop, December 7, 2010. 2010, 1-16.Google Scholar
- Masuya H, Mizoguchi R: An advanced strategy for integration of biological measurement data. The 2nd International Conference on Biomedical Ontology (ICBO2011): 28–30 July 2011; Buffalo, NY, USA. 2011, 79-86.Google Scholar
- Gkoutos GV, Green EC, Mallon AM, Blake A, Greenaway S, Hancock JM, Davidson D: Ontologies for the description of mouse phenotypes. Comp Funct Genomics. 2004, 5: 545-51. 10.1002/cfg.430.View ArticleGoogle Scholar
- Yamagata Y, Kou H, Kozaki K, Imai T, Ohe K, Mizoguchi R: Ontological model of abnormal states and its application in the medical domain. The 4th International Conference on Biomedical Ontology (ICBO2013): 8–9 July 2013; Montreal, Qc, Canada. 2013, 28-33.Google Scholar
- LOINC. [http://loinc.org/]
- Mizoguchi R, Kozaki K, Kou H, Yamagata Y, Imai T, Waki K, Ohe K: River flow model of diseases. The 2nd International Conference on Biomedical Ontology (ICBO2011): 28–30 July 2011. 2011, 63-70.Google Scholar
- Mizoguchi R: Ontology Engineering. 2005, Tokyo: Ohmsha Ltd.Google Scholar
- Peter HB, William CK: Definition, diagnosis, and classification of diabetes mellitus and glucose homeostasis. Joslin’s Diabetes Mellitus. Edited by: Kahn CR. 2005, 0: Lipplincott, Williams & Wilkins, 19: 331-332. 14Google Scholar
- Ito M: IGAKU-SHOIN’s Medical Dictionary. 2010, Tokyo: Igaku-Shoin, 2Google Scholar
- Yamagata Y, Kou H, Kozaki K, Imai T, Ohe K, Mizoguchi R: Ontological modelling of interoperable abnormal states. Proc. of JIST2012, Nara, Japan, LNCS 7774. 2013, 33-48.Google Scholar
- Kozaki K, Kou H, Yamagata Y, Imai T, Ohe K, Mizoguchi R: Browsing causal chains in a disease ontology. Poster & Demo Notes of 11th International Semantic Web Conference, Boston, USA. 2012, The Semantic Web Science Association (SWSA)Google Scholar
- Scheuermann RH, Ceusters W, Smith B: Toward an ontological treatment of disease and diagnosis. Proceedings of the 2009 AMIA Summit on Translational Bioinformatics, San Francisco, California. 2009, 116-120. American Medical Informatics AssociationGoogle Scholar
- World Health Organization: Diagnosis and Classification of Diabetes Mellitus and its Complications. 1999, Geneva: World Health OrganizationGoogle Scholar
- Rector AL, Rogers JE, Pole P: The GALEN high level ontology. Studies in Health Technology and Informatics. 1996, 34: 174-178.Google Scholar
- Masuya H, Kozaki K, Ohe K, Mizoguchi R: Trial to develop a database of context-dependent phenotype data. Proc. of the 27th Annual conference of the Japanese Society for Artificial Intelligence. 2013, 311-312.Google Scholar
- International Classification of Diseases (ICD). [http://www.who.int/classifications/icd/en/]
- SNOMED Clinical Terms (SNOMED-CT). [http://www.ihtsdo.org/snomed-ct/]
- Schulz S, Suntisrivaraporn B, Baader F: SNOMED CT’s problem list: Ontologists' and Logicians' therapy suggestions. Stud Health Technol Inform. 2007, 129 (Pt 1): 802-806.Google Scholar
- MEDIS. [http://www.medis.or.jp/4_hyojyun/medis-master/index.html]
- Ontology for General Medical Science (OGMS). [http://code.google.com/p/ogms/]
- Schriml LM, Arze C, Nadendla S, Chang YW, Mazaitis M, Felix V, Feng G, Kibbe WA: Disease Ontology: a backbone for disease semantic integration. Nucleic Acids Re. 2012, 40 (Database issue): D940-D946.View ArticleGoogle Scholar
- Smith B, Ashburner M, Rosse C, Bard J, Bug W, Ceusters W, Goldberg LJ, Eilbeck K, Ireland A, Mungall CJ, Leontis N, Rocca-Serra P, Ruttenberg A, Sansone SA, Scheuermann RH, Shah N, Whetzel PL, Lewis S: OBI Consortium: The OBO Foundry: coordinated evolution of ontologies to support biomedical data integration. Nat Biotechnol. 25 (11): 1251-1255.Google Scholar
- Kozaki K, Yamagata Y, Imai T, Ohe K, Mizoguchi R: Publishing a disease ontologies as linked data. In Proc. of JIST2013. 2013, 0: 0-in press, http://www.jist2013.org/Google Scholar
- Stedman’s: Stedman’s Medical Dictionary. 2008, Tokyo: Medical View Co Ltd, 6Google Scholar
- Ogawa S: Standard Textbook of Internal Medicine. 2009, Tokyo: Nakayama ShotenGoogle Scholar
- Anthony S: Harrison’s Principle of Internal Medicine. 2009, New York: McGraw Hill Medical, 17Google Scholar
- Imai T, Kou H, Zhou J, Kozaki K, Mozoguchi M, Ohe K: Japan medical ontology development project for advanced clinical information Systems. Proc. of 10th International HL7 Interoperability Conference 2009 (IHIC2009); Kyoto, Japan. 2009, 42-46.Google Scholar
- HOZO. [http://www.hozo.jp/]
- Disease Chain LOD. [http://lodc.med-ontology.jp/]
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.