In recent years, the rate of diagnoses of Autism Spectrum Disorder (ASD) has increased significantly. As a result, ASD has become one of the most studied psychological disorders in various editions of the main literature source for clinical diagnosis, the Diagnostic and Statistical Manual of Mental Disorders (DSM-5). Although it is one of the main problems faced by the public and overall health, there is still a lack of support for the diagnosis, providing a subjective approach, which varies among health professionals. The present work aims to verify the characteristics of autism through an analysis of the DSM-5, to assist health professionals in obtaining answers during the process of identifying factors that allow the diagnosis, in addition to promoting accurate information to researchers regarding people diagnosed with Autism Spectrum Disorder (ASD). Furthermore, this paper describes the project and development of a domain ontology for describing this domain, allowing a more objective way to close an early diagnosis, enhancing the chances of better social insertion of the individual. Adopting the Systematic Approach for Building Ontologies (SABiO), proposed by Falbo [1] we present a domain ontology containing information characteristic of diagnostic criteria A and B to demonstrate clear and structured data, using the DSM-5 and other data sources, providing a reference ontology (for graphical aid) and an operational ontology (for machine-readability). The study presents characteristics according to communication, social interaction, and motor behavior, as well as the level of support indicated for each diagnostic criterion.
Autism is a neurological condition characterized by afecting social communication, as well as causing
repetitive behaviors and restricted interests, often influenced by environmental and genetic factors,
resulting in an increase in its prevalence.[
CEUR
ceur-ws.org
communication is another characteristic area of autism. Some diagnosed children may develop this
skill without impairment, speaking clearly and establishing excellent communication. However, some
individuals have greater dificulty and tend to produce unilateral language. This type of communication
is an interaction ”without social reciprocity, seldom used for requesting or labeling rather than for
commenting, sharing feelings, or conversing” [
In light of these aspects, according to the Diagnostic and Statistical Manual of Mental Disorders,
5th edition [
To systematically portray the ontology development, we are grounded by the Systematic Approach for
Building Ontologies (SABiO) [
Ontology in computer science can be understood as the explicit representation of a set of concepts
(classes), attributes (properties), and the relationships that can exist within a specific domain, playing a
fundamental role in the semantic web by facilitating interoperability between systems. Antoniou &
Harmelen (2004) [
In the initial phase of gathering relevant information for eficient ontological development, creating the Ontology Requirements Specification Document (ORSD) stands out as one of the most important and 1https://ontouml.readthedocs.io/en/latest/intro/ontouml.html 2https://www.w3.org/OWL/
Role Providing the relationships stions and approaches toward the characteristics, present on the criteria A and B diagnosis. Those characteristics are aligned with the DSM-5.
Clarifying concepts and meaning toward the characteristics of ASD.
Validating and Evaluating the acquired information.
promising stages. According to Matrazo Fernando (2022) [
Therefore, the ORSD was based on collecting the necessary information for documentation creation. Thus, an attempt was made to verify the information in the Diagnostic and Statistical Manual of Mental Disorders, as it is a document used by healthcare professionals such as psychologists and psychoeducators in their studies regarding ASD. The documentation aims to raise questions that characterize diagnostic criteria A and B according to the DSM-5 manual, to deepen the understanding of these criteria, as indicated in Table 2. 1. Purpose The development of the hereby proposed ontology aims at the precise and embracing representation of the factors and particularities related to autism, to provide a solid foundation for research about ASD, as well as providing parental support on the early diagnosis. 2. Scope The ontology focuses on identifying the level of support, relating to the social condition provided by the DSM-5, employed to criteria A and B. 3. Language OntoUML with further translation to OWL 4. Intended Users User 1. Psychologists Psychiatrists, Psychopedagogists, and Physiotherapists provide support in research and facilitate the diagnosis.
User 2. Family members of autistics looking for relevant information about the characteristics of ASD. 5. Intended Uses Use 1. Graphically represent the ASD domain - Relatives may use this reference in order to better understand autistic behavior.
Use 2. Discovering tendencies and patterns - Researchers may foster on the provided ontology to answer questions regarding criteria A and B.
Use 3. Provide information about the characteristics - Psychologists may use the ontology to find information regarding the level of support.
Use 4. Exhibit the types of language, behavior, social interaction, and characterization that define ASD - Relatives
may refer to the ontology to enhance knowledge about the domain.
6. Non-Functional Requirements
NFR1 - The ontology must be grounded by the Diagnostic and Statistical Manual of Mental Disorders [
Even though we closed our scope into 4 CQs (once we adopt a CQ-driven development) [
According to Guizzardi et al. (2022) [
For the hereupon introduced reference ontology process of development, we provide Table 3 containing the involved agents, and their roles regarding the development, testing, and validation stages.
Furthermore, through the acquired information based on the DSM-5, and after being validated by domain experts, we developed a reference ontology focusing on the accessibility and endorsement of knowledge to the general users. Figure 1 showcases the graphical representation of the reference ontology.
According to the American Psychiatric Association (2014), the criteria are used with the aim of
facilitating the accurate diagnosis of disorders, through the introduction of an objective assessment
of symptoms in various clinical contexts. In this way, these criteria are fundamental elements for
obtaining the diagnosis, as they allow standardization in the study of diagnosis by providing concepts
in a common language for specialized professionals in the field of mental health. Furthermore, they are
organized to cover a wide range of related aspects, containing a detailed list of characteristics specific
to a particular disorder, as well as duration, functional impact, diagnostic exclusions, and specifiers.
Although the explanation of the concepts encompassing the characteristics of the Autism Spectrum is
beyond the scope of the present research, some of them will be shortly introduced. It was found that
Criterion A is directly related to communication and social interaction, and Criterion B is related to
restricted and repetitive patterns of behavior [
Furthermore, we introduce the concept of Characteristic Intensity, classified as a quality. The nature of the stereotype quality is correlational to the definition of moment, which in UFO is something that is existentially dependent from its bearer, and it must inhere in from it. In this case, the Characteristic Intensity is only enabled by the previous existence of Characteristc. In addition, the Characteristic Intensity defines the Support, which is a kind, and it is characterized (implies) a Support Level, which is defined as a mode. A mode stereotype, in UFO, is similarly defined as quality, however, it must provide a framework for how something should happen, in other words, a way something should behave. Finally, the support level is divided into three subkinds, i.e., Demanded Support, which is the minimal support level, Substantial Support, and as a higher level of support, Highly Substantial Support. The DSM-5 provides the definitions.
Finally, we can characterize each criterion by the three members of characteristics, and the support level of each. The Criterion A is linked through a characterization relation to Communication (hasDificultyIn , similarly,hasDificultyIn Social Interaction , and requires either a Demanded Support or Substantial Support level of support (CQ1). Whereas Criterion B is solely dependent on Behavior, requiring either a Substantial Support or a Higly Substantial Support level. Those classifications were also provided by the DSM-5 (CQ2). As proposed by the NFR3, the reference ontology, operational ontology, and all supplementary material are available in an open-access GitHub repository 3.
According to the American Psychiatric Association (2014), the criteria are used with the aim of
facilitating the accurate diagnosis of disorders, through the introduction of an objective assessment
of symptoms in various clinical contexts. In this way, these criteria are fundamental elements for
obtaining the diagnosis, as they allow standardization in the study of diagnosis by providing concepts
in a common language for specialized professionals in the field of mental health. Furthermore, they are
organized to cover a wide range of related aspects, containing a detailed list of characteristics specific
to a particular disorder, as well as duration, functional impact, diagnostic exclusions, and specifiers.
Although the explanation of the concepts encompassing the characteristics of the Autism Spectrum is
beyond the scope of the present research, some of them will be shortly introduced. It was found that
Criterion A is directly related to communication and social interaction, and Criterion B is related to
restricted and repetitive patterns of behavior [
Furthermore, we introduce the concept of Characteristic Intensity, classified as a quality. The nature of the stereotype quality is correlational to the definition of moment, which in UFO is something that is existentially dependent from its bearer, and it must inhere in from it. In this case, the Characteristic Intensity is only enabled by the previous existence of Characteristc. In addition, the Characteristic Intensity defines the Support, which is a kind, and it is characterized (implies) a Support Level, which is defined as a mode. A mode stereotype, in UFO, is similarly defined as quality, however, it must provide a 3https://github.com/jessicaruana/Autism-ontology framework for how something should happen, in other words, a way something should behave. Finally, the support level is divided into three subkinds, i.e., Demanded Support, which is the minimal support level, Substantial Support, and as a higher level of support, Highly Substantial Support. The DSM-5 provides the definitions.
Finally, we can characterize each criterion by the three members of characteristics, and the support level of each. The Criterion A is linked through a characterization relation to Communication (hasDificultyIn , similarly,hasDificultyIn Social Interaction , and requires either a Demanded Support or Substantial Support level of support (CQ1). Whereas Criterion B is solely dependent on Behavior, requiring either a Substantial Support or a Higly Substantial Support level. Those classifications were also provided by the DSM-5 (CQ2). As proposed by the NFR3, the reference ontology, operational ontology, and all supplementary material are available in an open-access GitHub repository 4.
With the creation of the graphical representation and generation of classes, it is necessary to describe the
internal study in order to provide answers to the competency questions defined in the ORSD document.
Thus, after the demonstration of information through OntoUML, a translation is made for the OWL
language in which the study will be conducted with the assistance of the Protégé tool. ”An image is
worth a thousand words,” as expressed by Elkashef, et al. (2020) [
As demonstrated, the child classes are characteristics that define the parent classes, such as types of verbal language, non-verbal language, restrictive and repetitive behavior, social reciprocity, and social interactions. To represent the diferences between the attributes mentioned here as child classes, the ”disjoint with” was assigned so that it would be possible to make distinctions while searching for answers to competency questions.
The Protégé tool uses a property to find relationships between classes, the Object Properties, as
depicted by Silva et al. [
The Object Properties hasSupport received the class criterion Aas its Domain and the class characteristic as its Range. The properties hasHighlySusbtamtialSupport and hasSubstantialSupport received both the classes criterion A and criterion B as their Domain, and the class ”characteristic” as their Range. As a ifnal step, instances were created in the Individuals tab in the Protégé mechanism, demonstrating the type of support required according to each diagnostic criterion attributed to the autistic individual.
It is worth noting that this ontology is intended for identifying the characteristics contained in diagnostic criteria A and B as advocated in the DSM-5, in order to find attributes that demonstrate the type of support to be used according to the particularities that diferentiate them. Therefore, the Object Properties maintain the relationships between the instances of the Support Level class.
According to [
CQ2 (Informal): What characteristics compose criterion B? CQ2 (Formal):
The main role of the ontology aligns with the provision of structured knowledge and explanation, along with relevant answers. Furthermore, using such methodology in the autism study context becomes fundamental, supplying organized and detailed information that facilitates the clear and eficient comprehension of the characteristics that compose ASD, i.e., communication, social interaction, and behavior. This explanation allows a wider comprehension by society. Moreover, as a subjective matter, the evolution process is always open. This ontology was framed by the requirements proposed by stakeholders, and alignment with literature. The chosen methodology provides a broader application for the ontology, once we could use two artifacts for validation and refinement. For instance, we could partially answer CQ1 with Figure 4, once it provides the relationships for criterion A, allowing a user-friendly visualization of data, hence, more professionals could share their thoughts and evaluate our work.
We may state a few challenges while developing such an ontology. Work with health professionals always involves an ethical approach, especially when dealing with children observation and data sharing. Furthermore, the encounters with specialists were limited, in average, each specialist have been consulted after a major change in the reference model, hence, 2-3 times per person. Overall, ontologies, especially domain ontologies, are designed to answer specific conceptual questions, defining their granularity. For that, we propose as future work the implementation of case studies, retaining the instantiation of the ontology, along with the implementation of diferent observed properties, granting the identification and intensity of the characteristics. Finally, as a technical approach, we aim to develop diferent SPARQL queries, outgrowing the current knowledge retrieval capabilities, supplemented by SWRL rules, and SHACL constraints, thus, enhancing the supplementary diagnosis capabilities for health professionals.