Double click on rdfs:comment at left This is a demonstration ontology reconstructing much of the OpenGALEN Upper Ontology in OWL-DL. Detailed comments are included with classes and properties along with demonstration notions preceded with the prefix "Probe_". The ontology is provided strictly "as is" and without guarantees. It is still in progress and will be updated from time to time. The ontology is free for use with acknowledgement to OpenGALEN and the University of Manchester. All material copyright University of Manchester. A discussion is available in "Patterns, Properties and Minimizing Commitment: Reconstruction of the GALEN Upper Ontology in OWL" Alan L Rector and Jeremy Rogers, to be presented at the Workshop on Core Ontologies (CORONT) in conjunction with the European Knowledge Acquisition Workshop (EKAW-2004), Northampton, UK. All material available from http://www.cs.man.ac.uk/~rector/ontologies/sample-top-bio/ Defined here as a Biological Entity which is the outcome of a pathological process. Is this correct or should Disease be a primitive top level class? Disease / Problem is always a biological entity ---- but is it always an occurent or a continuant ? Generally disease should be a process with some outcome which presents as signs or symptoms --- So just leave this inside Biological Entity without further sub classification Entities which can stand on their own. Primitive self-standing entities in the upper ontology come in disjoint pairs. In the domain ontology they form disjoint non-exhaustive trees. The pairs show up poorly because the windows are ordered alphabetically. However, they can be identified by the disjoint axioms (lower right hand corner of this tab). The disjoint pairs are deiberately not exhaustive - i.e. we defer the issue of whether everything must be either physical or non-physical etc. This allows defering commitment on such issues as whether sets of physical things are themselves physical, etc. This is a matter on which other ontologies differe. Also no opposite for 'Complex' is provided, again to defer decisions. 'Biological' is included for convenience in a BioMedical ontology. In general it should be possible to classify a self-standing entity by asking questions about the basic dichotomies and subdichotomies The level of a process. Used in this demonstration to define Heart Failure. This axis has been extracted from NCI and should include Organ and Organ_part to form whole organ systems. Set in FMA sense of a specified list with a specific number. NOT a mathematical set. All entities which are dependent modifiers - roughly speaking value types and values. (See SWBP note on value partitions) Place holder for any finding of clinical history or clinical examination or investigation findings --- after classification, the concepts Sign, Symptom and Investigation_finding should fall under History_examination_finding. See Physical_entity Place holder for anything that a patient complains of ..... This will include entities like pain, cough, fever ---- they are all biological entities and not going to divide them further as occurents or continuants. Patient can complain of "pain" located in the abdomen which is a symptom but "tenderness" demonstrated during clinical examination will be a finding/sign NTS : Should have a property ---has_finding maybe Normality_status is a special status for biomedical ontologies. It has a much more complex structure than most other status, but is a good demonstration of the advantages of using "value partitions" rather than simple values. See comment under Metadata tab for ontology documentation. (double click on rdfs:comment) Domain_entity is the top proper entity in the ontology. We recommend defining a top entity within the ontology because: a) It is often useful to define other entities either for debugging purposes (probes) or to simulate individuals, etc. b) Some tools treat the immediate children of owl:Thing peculirarly. Those things that are manifest physically - both occurrrents and continuants but not abstractions such as ideas, patterns, etc. GALEN is specifically silent on whether "Agents" are physical or not in order to be able to fit in with ontologies which make either choice. ProbeClass:true Continuant or Endurant. Entities that peresists in time but has parts in space. As opposed to Occurrent Modifiers are the common modifiers which change the state of a quality or other aspect of a self-standing entity. In many cases it is worth representing the "quality" as a class. Probes are tests inserted to check classification. Tools are almost ready to have them visible or invisible as required. ProbeClass:true This says headache is a pain in some part of the head....this really depends on what we put under the concept Head....Need to define so that pain in the ear is not a headache just cos ear in some classification comes under Head_and_Neck structure. Pathology of stomach, analogous to Cardiac_pathology. Again both the syndrome and the physical structure are included for demonstration although a more practical query would include only one or the other. ProbeClass:true See biological Any of the various independent bodies in the cell - mitihorndria, chloroplasts etc. There is some disagreement as to whether the nucleus should be classified as an organelle. Here it is not. Probes for the clinical parts of the heart. This includes both functional and structural parthood. This is what a doctor would expect to see, including Pericarditis - i.e. an inflammation of the covering of the heart. Strictly speaking, the Pericardium is, however, a spearate organ, so according to the FMA view, it does not form part of the heart. ProbeClass:true Any pathology in the heart. Includes both the processes and the syndromes. A sensible interface would ask for one or the other, but they are both included here for demonstration purposes. Note that Circulatory_failure_due_to_heart_failure is included. This is the reason for having has_actor imply has_locus. This is a mild over-generalisation and will lead to some over inclusion. ProbeClass:true Tried to differentiate behaviour of cancers into Benign and Malignant. This should probably cover the behaviours of cancer and both these partitions are mutually exclusive. NTS : Is this definition of "has_indication" the right way of saying it can have an indication either "disease" or "clinical_finding" ClinicalSituation "Clinical Situation" in GALEN Effectively "Disease" as opposed to the simple process or lesion at the heart of the disease. Hence we have both "Pericarditis_syndrome" and "Pericarditis_process" Likewise, both "Ulcer of stomach" and "Ulcer_of_stomach_syndrome" which includes the process and might include the pain and other things that go with it. Note thatt his simplified version just has a single level of "includes" property for syndromes. By rights, there should be two levels - one for the defining conditions and one for everything else. See 'Nexus' Is a clinical finding always a biological entity ?? Where do things like calculus, foreign body , etc fit if its always a biological entity..... A solution would be to leave these as Primitive classes and then define only the concepts that the reasoner will have to manipulate. Patterns are general forms and scripts including 'oeuvres' - the abstractions for intellectual outputs. These were not developed in GALEN and so are left undeveloped here. There is also an argument, according to Barry Smith, for using pattern rather than complexes for Diseases/Clinical situations. However, this leaves us with three entities at least for each disease. Furthermore, we talk of the duration etc. of a disaease which seems to apply more sensibly to the process. Fl.ocks of geese, schools of fish, etc. Reified relations. Primarily in biology "Qualities" but might also be other forms of relationship Is the outcome of any Investigation ---- clinical examination, investigation_procedure, intervention_procedure. Clinical examination findings will include all concepts from Sign. A discrete collection of things - flocks etc. Also includes The major parts of the skull including structural and functional - in this case in demo only structural parts. Excludes the Brain because the Brain is contained in rather than part of the skull. ProbeClass:true 1 This is just a place holder for classifying commonly encountered diseases of interest in patient records. Specific complex lijnked by some fundamental principle as in Syndromes. Also dualities such as ulcer and ulceration which are here not distinguished from other syndromes but might be in a more pedantic version. Also this version does not distinguish arbitrary syndromes such as Tetrology of Falot from systematic syndromes, such as all the things that fit together around Diabetes. Ad hoc grouping - taken from FMA The membrane enclosing the cell A primitive in the FMA. Here defined. There may still be case for distinguishing the primitive Cell_part_NAMED as in GALEN from the defined "Part of Cell". ProbeClass:true ProbeClass:true place holder for all examination findings ---- refer Symotom ----should be classified as a subset of Investigation_findings. ---- identified by flag "is_sign". This flag can be further used to associate sign to diseases/locus if needed. List of all normal cells in the body----extracted from NCI meta-thesaurus. Functional system ProbeClass:true Everything located in the skull Includes the brain which is contained in the cranial cavity but not a part of the skull. The stuff that fills the cell defined as a neoplasia structure with malignant behaviour status and being an outcome of a neoplastic separately. How does this biological definition differ from the other definition where they been defined as intersection 1 1 1 ProbeClass:true Axiom - A Biological Entity Annotated as a Probe to make it disappear if needed An axiom that a Syndrome that includes anything pathological is itself pathological A good example of why we need to be able to make axioms either anonymous or at least invisible. ProbeClass:true The structural parts of the heart. Essentially the FMA view of what should be part of the heart, excluding individual cardiac cells etc. which require the still broader is_part_of relation. Measurable entities - water, air,blood, tissue, etc. as opposed to Discrete_entities For now only handles weight with units Milligrams...but should be able to hold values like Rads, etc for Radiotherapy The skull is actually a set of bones. The level of activity of a process. Amongst other statuses takes Level_adequacy_status 1 1 A collection of grains at one level of granularity that make up a mass entity at the next - i.e. an entity where it is the behaviour of the collective of grains that is important rather than the behaviour of the individual grains is unimportant In general, informally "multiple of" can be read as a pluraliser, i.e. "Multiple has_grain only Red_cell" = "Red cells" Note that multiples are always formed with universal restrictions rather than existential. IE multiples are homogeneous. Quality_status_partitions are the vast majority of status partitions which go with qualities. Only a single example, "Level" is given here, but most clinical features such as body temperature, height, weight, pulse rate, etc. would be treated as "qualities". Each would have its own "Quality" defined under self-standing entities and a set of mutually exclusive statuses. Locus can be both a structural (anatomical) location or can be a functional (physiological) location ----serves as range for has_locus property. 1 A treatment can also have some goals ---- should be defined in "Status_partition" maybe ?? Countable entities - people, buildings, organs, etc. Things that happen - processes, events, and activities Occurents - have duration - have parts in time as well as space. as opposed to Continuant Anything that is in any way aggregated This definition isnt right and needs to be clarified. It assumes all Mass_strucutures have a size feature and so are physically measurable. Is this right ? 1 1 1 1 Intrinsic properties are those which are not "qualities" - laterality, topology, etc. There is usually little point in reifying such properties. Refer to defintion of Mass_structure cos this defintion is analogous to it. GALEN isFeatureOf Children for each quality to control cardinality. Links features to self-standing entities. Placed in this way around to make child of relation between syndrome and the different pathologies. The Subproperty includes_specifically is a sub of has_locus. So in general a syndrome or "Clinical Situation" can include things which it is not a disease of - e.g. Diabetes should not be a renal or eye disease, although the total syndrome includes these things, at least as possibilities. However, in this version this broader version of includes is not used. Physical location See Refining Property The usual clinical meaning of partonomy - including functional and physical partonomy but excluding membership - which effectively excludes malfufunction of individual cells but not of multiples of cells. Outcome of process. Has additional parent has_locus to get reasoning right in absence of more sophisticated rules. The actor in a process. If the Heart is the actor in a pathological process then it is a locus of the pathology - see Circulatory_failure_due_to_cardiac_failure Relation betwen occurrents and either continuants or other occurrents. TODO: Consider whether to place under affects. should has_outcome be a child of has_locus. has_outcome The general outcome of an act. Should probably be a sub of affects. Still being chased. See Relational Property Containment. In this veresion must be contained in a cavity. This is more pedantic than GALEN. Note containment implies location but not partonomy Membership - not standard partonomy and not included in clinical partonomy Properties linking self-standing entities Subvariety of acts_on which indicates actual effect on the thing acted on - creation, modification, etc. A sub of has_locus. Otherwise, everything acted on by the process would become the locus. Counterexample - heart failure. The pumping acts on blood but is adequate. It affects circulation. This is a variant of the GALEN version which achieved many of the same effects via the propagates_via axiom. GALEN 'hasLocation' - locus of pathology. NB because of differences between sanctioning and domain/range constraints, must always ask for has_normality pathological & hasLocus... The specific version of includes which points to the causative or key pathology. Follows GALEN naming conventions. Alternatives welcome. Acts on substrate Pumping acts_on blood. Properties linking self-standing entities to refining entities or refining entities to each other. Connection. Not symmetric in GALEN because GALEN handled symmetric properties badly. Generic partonomy including membership. Structural partonomy not including membership. Structural partonomy in GALEN does not cross scales. Most general notion of location. Not necessarily physical NOT GALEN "hasLocation"