This paper furthers the River Flow Model (RFM) of diseases based on the functional view of causation that we have elaborated for the sake of formal ontology. The clarification of RFM enables us to offer a clear comparison between RFM and the dispositional model of disease given by the Ontology for General Medical Science (OGMS) and to see the core ideas of disease that RFM and OGMS have in common. This work will be an initial step towards the development of an ontological module for generic disease representation.
A disease ontology aims to meet a high demand for a
common framework in which an increasing amount of medical
information and data are shareable among different
information systems. A viable definition of disease is thus
indispensable for the robust construction of disease ontologies.
Designed to represent the entities that are involved in a
clinical encounter in compliance with the framework of the
Open Biomedical Ontologies (OBO) Foundry [
The River Flow Model (RFM) of diseases was initially
presented in [
A remaining problem with the RFM of diseases resides in
its explicit reference to the notoriously difficult notion to
model: causation. RFM will not be articulated clearly until
* To whom correspondence should be addressed: {fumiaki.toyoshima;
mizo; ikeda}@jaist.ac.jp
1 The OWL file of the RFM ontology is freely available at the website:
http://rfm.hozo.jp/ (Last accessed on July 3, 2017). For a navigation system
for the definitions of diseases that is built based on the RFM of diseases
(especially on its core notion of causal chains of abnormal states), refer to
Disease Compass available at the website: http://lodc.med-ontology.jp/
(Last accessed on July 3, 2017).
the notion of causation is specified enough, much less
RFM's relationship with OGMS's dispositional account of
disease. Although it has been long investigated among
philosophers, causation still remains an enigma despite its
enormous importance for ontological modelling in a wide
range of domains. [
In this paper we provide further development of RFM on
the basis of the theory of causation that we have recently
tailored for formal ontology, based on YAMATO: the
functional view of causation [
The paper is organized as follows. Section 2 presents an overview of YAMATO and the functional view of causation. Section 3 offers a basic idea of RFM, then providing a developed version of RFM's definition of disease. Section 4 compares the RFM of diseases with OGMS's dispositional model of disease from the viewpoint of causation. Section 5 gives concluding remarks and outlines future work.3 2
Causation has been explored in formal ontology directly by
[
We have recently developed the functional view of
causation [
YAMATO is fundamentally based on [
Two prominent subcategories of occurrents are processes (e.g., walking and the flowing of water) and events (e.g., a walk). Most importantly, processes are intrinsically 'ongoing', whereas events are in nature 'completed'. Among subcategories of occurrents are states: time-indexed qualitative occurrents (e.g., being hungry at time t1, sitting at time t2, and speeding at time t3).
Concerning relations in YAMATO, an object participates in a process, an event, or a state. A process is enacted by (made possible by) an object that participates in that process. Any process constitutes a (unique) event. For instance, John enacts a walking process, which constitutes the walk event.
Finally, YAMATO exploits the model of roles (a subtype
of dependent continuants) [
The functional grounding of causation is based on the
device ontology view of reality. The device ontology [
According to the device ontology, an object O plays an agent role as a device Od in a given context and a behavior B of Od is the change from one state S1 to the other state S2 of the operand OP that is processed by Od. For instance, a behavior of the human heart (as a device) is the process of pressuring blood: the process of change from the low state to the high state of the pressure (operand) of the blood that has travelled through the human heart.
Given the intimate relationship between change and
causal connections, there must be a causal connection behind B
as a change in the operand OP.4 Aiming to explain causation
in terms of an inherently causal entity, the functional view
of causation consists in saying that B of Od is an inherently
causal process in virtue of its own functional nature whose
combination of how to achieve and what to achieve [
To illustrate this, suppose for instance that a cutting machine cuts a fish into pieces. A cutting machine plays an agent role as a cutting device in the fish-cutting context and a behavior of the cutting device is the change from the state of being one piece to the state of some pieces of the fish that is processed by the cutting machine. The causal connection behind the behavior of the cutting device is the relation between what occurs inside the cutting device (say the cutting motion with a knife) and the state of being some pieces in which the fish participates.
Consider the function of a cutting machine. Analyzed in
terms of [
This conception of function justifies the idea that a cutting behavior of a cutting device is inherently causal primarily because cutting involves both to divide something to some pieces (what to achieve) and to use a sharp object in order to divide something (how to achieve). The state of being some pieces in which the fish participates and the cutting motion with a knife inside the cutting device are concretizations of what to achieve and how to achieve of a cutting behavior of a cutting device, respectively.
Suppose for another example that, when emitted by a nuclear reactor, an electron e1 impacts another electron e2. E1 plays an agent role as a device (say Device1) in the context of a quantum jump of e2 and a behavior of Device1 is the process of exciting e2: the change from the lower state to the higher state of the energy (operand).5 Being of a functional nature, the behavior of exciting e2 by Device1 involves both to bring e2 to the higher energy level (what to achieve) and to use the motion of e1 (how to achieve). 4 Consider for instance that an object travels at a constant velocity. A behavior of the object in constant motion is the change in the position (operand) corresponding to the transition between two places which the particle occupies. The causal connection behind this behavior is the relation between the motion of the object and the position change behavior of the object. 5 An electron that receives the amount of energy exceeding the threshold value is said to make a quantum jump or to be excited from the lowerenergy state to the higher-energy state.
It is interesting to note that, given the device ontology, the example of a quantum jump of an electron bears a close similarity to the scenario in which, when shot by a billiard cue, a billiard ball collides with another billiard ball. As this shows, the device ontology view of reality is versatile enough to be applicable to different granularity levels of reality. 2.3
The functional square of causal relations is the fourfold
distinction of causal relations that are in nature functional:
achievement, allowing, disallowing, and prevention (see
Table 1). The functional square claims to be an
allencompassing table of causal relations, thereby meshing
well with a counterfactual theory of causation [
Achievement is a direct and positive causal relation: a
relation between occurrents (not only events but also processes
and states) that we ordinarily take to be causal. The
functional view of causation [
Example: regarding thrombosis, the growing process of a blood clot in a blood vessel achieves the state of being small of the cross-sectional area of the blood vessel, which in turn achieves the state of being in short supply of oxygen. Note that thrombosis is cured only when the cross-sectional area of the blood vessel is of a clinically normal size.
Allowing holds between an occurrent and the state of being a necessary condition for achievement. In this respect, allowing is indirect with respect to achievement. It is also positive because its holds both between actual occurrents both ontologically and linguistically. Example: ATP production allows active calcium transport (by achieving the state of being in operation of transporters as a necessary condition for active calcium transport).
Prevention is direct because it is ontologically the same as
achievement but it is nonetheless negative because,
linguistically speaking, it has as relatum a non-actual occurrent
(see [
Disallowing consists in 'preventing' some sufficient condition for a phenomenon, thereby preventing the phenomenon. It is therefore indirect with respect to prevention and it is also negative, just as prevention is. Example: ras inactivation disallows signal amplification of MAP kinease cascade (by 'preventing' the activation of kinease molecules such as MAPKK in the following MAP kinease cascades). 2.4
Dispositions are of high utility in overall ontological
modelling, especially in the biomedical domain [
A disposition is a dependent continuant (bearer) that exists because certain features of the physical make-up (material basis) of the independent continuant in which it inheres and whose instances can be realized in associated BFOprocesses of specific correlated types in which the bearer participates. For instance, fragility is the disposition of a glass (bearer) to break (realization) that depends on a particular physical molecule structure (material basis) of the glass.
In philosophy, there is a growing interest in a
dispositional theory of causation according to which, roughly,
causation occurs when a disposition realizes itself [
3.1
We said in Section 2.1 that a river is an object (continuant)
and the flowing of water is a process (occurrent). According
to YAMATO or rather [
Moreover, just as a river has the flowing of water as its internal process (a process that occurs inside the river), a disease has as its internal process a number of chains of causal phenomena. A disease is in this respect constituted of causal chains of phenomena that are harmful to the organism from a medical viewpoint. This constituted-of relation has a close affinity with the constitution relation of a process to the event (see Section 2.1).
All these considerations lead to the above-mentioned main idea of the RFM of diseases: a disease is a dependent continuant constituted of causal chains of abnormal states. Type I diabetes, for instance, may have inside it the causal relation between the state of being broken of pancreatic beta cells and the state of being little of insulin in the blood. 3.2
The imbalance model, initially presented in [
To illustrate the imbalance model, consider diabetes in general. In a patient with diabetes (whether Type I diabetes or steroid diabetes), the required amount of insulin (demand) exceeds to a clinically abnormal degree the amount available for working insulin (supply) and this imbalance state causes the state of being at an elevated level of glucose in the blood, which may result in the loss of sight of the patient over a long period of time. The difference between Type I diabetes and steroid diabetes resides partly in the fact that, in the case of Type I diabetes and steroid diabetes, causal chains that lead to the imbalance state described above include the state of having depleted pancreatic beta cells and the state of having large quantity of steroids, respectively.
We have the following definition of a clinical imbalance state: Clinical imbalance state =def. a state of an organism such that, given a parameter p that is relevant to the organism, the mismatch between the supply and the demand that are specified with respect to p falls outside a clinically normal range for the organism's homeostasis. 3.3
An existing RFM definition of disease [
Note also that an RFM-disease has at least one clinical imbalance state (see Section 3.2).
The reader should keep in mind that RFM's notion of
clinical abnormality is virtually primitive. This is justified
through the sharp distinction between the domain-neutral
notion of clinical abnormality and the domain-specific task
of the identification of clinical abnormality. What counts as
clinically abnormal would vary from clinical to clinician in
a broad biomedical field. RFM purports to be an ontological
(domain-neutral) model of diseases and RFM's notion of
clinical abnormality refers to the existence (rather than the
content) of criteria for observing a state from a clinical
perspective. For an application of RFM's notion of abnormal
states, see [
To investigate the commonalities and differences between
RFM and OGMS's dispositional model of disease, we
briefly present some core terms of OGMS. A disorder basically
refers to a material entity which is clinically abnormal and
part of an organism, although the precise definition of a
disorder has been repeatedly changed and seems to be under
development (see [
An OGMS-disease is a disposition (i) to undergo pathological processes that (ii) exists in an organism because of one or more disorders in that organism. The material basis of a disease as a disposition is a disorder of the disease and a disease comes into existence when its corresponding disorder does, i.e. when the organism disposes towards its relevant pathological processes. A disease as a disposition may go unrealized, e.g., when it lies dormant over a long period of time. A disease course is the totality of all BFOprocesses through which a given disease instance is realized. A disease course of a disease ranges widely from potentially asymptomatic early stages of the disease to its recognizable, pathological processes.
For instance, epilepsy as a disease is a disposition to undergo the occurrence of seizures (pathological processes) that exists owing to some clinically abnormal, neuronal circuitry of the brain (disorder). 4.2
RFM and OGMS share some common views on disease.
First of all, an RFM-disease and an OGMS-disease both say
that a disease is in nature a 'causal pattern'. An RFM-disease
is characterized by the regular way in which abnormal states
are causally connected and the pattern nature of an
RFMdisease may be represented in terms of a directed graph in
graph theory [
Similarly, an OGMS-disease reasonably qualifies as a causal pattern. For one thing, an OGMS-disease is causal, since a disposition is an inherently causal property. For another, an OGMS-disease is of a pattern nature because a disposition which has a 'specific' material basis realizes its 'corresponding' BFO-processes when exercised under some 'appropriate' circumstances.
It is important to emphasize that a disease is a dependent continuant, but not an occurrent, although some existing ontologies classify a disease a subtype of occurrent. A disease is some entity with which a patient is affected and which medical practitioners identify, diagnose, and cure. A disease is something that comes into existence, grows, and finally disappears in the patient's body. All these observations imply that a disease is an entity that persists in time, i.e. a continuant. Additionally, a disease is a dependent continuant that inheres in an organism. 4.3
Furthermore, both an RFM-disease and an OGMS-disease involve what we may call a 'clinical threshold': the level at which symptoms of a disease begin to develop. RFM characterizes a clinical threshold employing the imbalance model. Symptoms of a disease show themselves when the clinical imbalance state that is relevant to the disease has occurred. Likewise, OGMS explicates a clinical threshold in terms of the emergence of a disease as a disposition (and of its corresponding disorder as a material basis of the disposition). Having reached a clinical threshold, an organism disposes towards pathological processes, which are recognizable through symptoms.
It is nonetheless vital to clarify the difference between an RFM-disease and an OGMS-disease from the viewpoint of a clinical threshold. In the case of an RFM-disease, a clinical imbalance state of a disease is not always an initial state of the disease. As said in Section 3.2, the imbalance model abstracts from a disease its generality and eliminates its specificity. For instance, Type I diabetes and steroid diabetes fall into a group of diabetes, since they have the same kind of clinical imbalance state (i.e. the deficiency of insulin), but those two diseases still differ from each other because they have different causal chains of abnormal states.
By comparison, it is clear that the emergence of an
OGMS-disease as a disposition is always at the beginning of
the disease. To do justice to the specificity as well as the
generality of disease, OGMS covers a predisposition to
disease of type X: a disposition in an organism that constitutes
an increased risk of the organism's subsequently developing
the disease X [
Roughly speaking, the generality and the specificity of an OGMS-disease are to be captured by a disease as a disposition and a predisposition to disease of type X, respectively. For instance, Type I diabetes and steroid diabetes belong to the same diabetes category because they are essentially the 'diabetes disposition'. These two diseases are nevertheless different because a predisposition to have diabetes that is involved in Type I diabetes is of a different nature from that involved in steroid diabetes. A realization of the former predisposition, but not of the latter predisposition, may have as part the BFO-process of destruction of pancreatic beta cells; conversely, a realization of the latter predisposition, but not of the former predisposition, may have as part the BFOprocess of the increase of steroids. 4.4
Though conceived as a causal pattern in common, an OGMS-disease and an RFM-disease are significantly difference in the sense that OGMS's basic ''unit of thought'' is a disposition (dependent continuant) but RFM's is a state (occurrent). This fundamental difference between an OGMSdisease and an RFM-disease is largely, if not totally, explicable in terms of causation.
For OGMS, causation is dispositional: causation occurs when a disposition as an inherently causal property realizes itself. The dispositional conception of causation leads directly to an OGMS-disease as a disposition. For RFM, causation is functional: causation occurs when an inherently causal behavior of an object (a device) achieves some occurrent. An RFM-disease inherits its causal nature from the linkages, brought about by functional causal power, among abnormal states. 4.5
Clinically abnormal: continuant vs. occurrent Another indicator of the contrast between OGMS's and RFM's conceptions of causation is the difference in what is clinically abnormal between OGMS and RFM. Clinically abnormal are a disorder (continuant) in OGMS and a state (occurrent) in RFM. In OGMS, a disease as a disposition inherits its clinical abnormality from a disorder as its material basis; and therefore, a disorder (a material entity) is defined as clinically abnormal. In contrast, RFM's notion of clinically abnormal state reflects well the idea of an inherently causal occurrent (process) embraced by the functional view of causation.
Consider for instance inflammation as a disease. OGMS would say that inflammation as a disposition is clinically abnormal primarily because so is its material basis, e.g., the cells in the relevant part of the organism. RFM says however that the clinical abnormality of the cells there, if any, is a contributor to that of states that inflammation has inside it. This marks a close analogy with the argument for the functional view of causation: a realization of a disposition contributes to functional causation (see Section 2.4). 5
We have furthered the River Flow Model (RFM) of diseases
employing the functional view of causation. By so doing,
we have offered a clear comparison between an
RFMdisease and an OGMS-disease in terms of causation, in
particular the contrast between functional causation and
dispositional causation. The work done here will contribute to the
formalization of a general disease module for foundational
ontologies (whether BFO, YAMATO, or others) and also to
the methodological supplementation of OGMS [
Future work includes further development of RFM using
the functional view of causation. For instance, the
exploration of the formal relationships among the causal relations in
the functional square would enable us to have a closer
examination of the nature of causal chains of abnormal states
in a disease. Along another line of research lies the question
of whether RFM can be extended to mental disease, just as
OGMS's disease model was in [
We are deeply grateful to Yuki Yamagata for her valuable comments on a variety of causal phenomena in biomedicine.