eDEM-CONNECT: Agitation ontology for the chatbot-based support of informal caregivers of people with dementia

eDEM-CONNECT: Agitation ontology for the chatbot-based support of informal caregivers of people with dementia | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article eDEM-CONNECT: Agitation ontology for the chatbot-based support of informal caregivers of people with dementia sumaiya suravee, Christiane Pinkert, Iris Hochgraeber, Margareta Halek, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6289849/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Persons with dementia (PwD) face cognitive decline, placing added stress on family caregivers. Challenging behaviour like agitation is one of the prominent behaviors exhibited in PwD. We envision a chatbot that provides information to unprofessional caregivers to ease the stress factor while dealing with agitation in PwD. To develop the chatbot, one needs domain knowledge, including types of agitated behavior, living and socio-economic conditions of the PwD. We call this structured knowledge an ontology. This study focuses on developing the eDEM-Connect Ontology: Ontology of Dementia-related Agitation and Relationship between Informal Caregivers and Persons with Dementia (EDEM-CONNECTONTO) as the domain knowledge that chatbot needs to use for providing adequate information. We perform a systematic literature review, analyze existing ontologies, hold workshops with experts, and interview informal caregivers. We then develop and validate the EDEM-CONNECTONTO with the Protégé software. EDEM-CONNECTONTO consists of 241 Concepts, 8 relations, and 240 individuals. The results from the evaluation show that it meets the standard for biomedical ontologies. The EDEM-CONNECTONTO addresses agitation in PwD, prioritizing support for family caregivers and incorporating non-pharmacological interventions. It fills gaps in nursing science by formalizing knowledge relevant to dementia care and agitation, guiding future research in the field. Health sciences/Health care Biological sciences/Computational biology and bioinformatics/Classification and taxonomy Biological sciences/Computational biology and bioinformatics/Data integration Biological sciences/Computational biology and bioinformatics/Databases Biological sciences/Computational biology and bioinformatics/Literature mining Dementia agitation ontology non-clinical interventions Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1 Introduction Dementia affects over 55 million people worldwide [1]. It leads to limitations in cognitive abilities, such as declining memory as well as changes in behavior or emotional control [1]. The most common form of behavioral change in PwD is agitation [2]. Agitation is persistent or frequently recurring behavior that is primarily manifested in excessive motor activity, leading to significant impairments in interpersonal relationships and the ability to accomplish everyday tasks [3]. In Germany, about 70% of very old PwD live at home and are mainly cared for by relatives and outpatient care services [4]. Dealing with agitated behavior is a major challenge for formal and informal carers. Agitation can lead to early placement of the PwD in a nursing home [5], cause stress to the family caregivers [6], and threaten the stability of home-based care arrangements. Behavioral problems also hurt the quality of the dyadic relationship [7]. Disturbed communication is a big challenge for family caregivers and leads to a negative perception of the relationship [8]. Therefore, the dyadic relationship may influence the development of agitation, which can impact the quality of life of the individuals concerned [9]. To support families in managing agitated behavior, needs-based and situation-specific information and support services are required. Although both formal and informal services are available, their utilization is low and the reasons are diverse [10, 11, 12]. Family caregivers often cannot recognize which information is important for them or have difficulties in applying it to their situation. Family caregivers primarily contact help services via an online search engine, which often finds irrelevant, confusing or outdated content. Thus, there is a strong urge for user-centered and structured information. Our research is taking place within the eDEM-CONNECT project (see [13]), which aimed at developing a chatbot-based communication system that provides structured and comprehensible knowledge to family caregivers. The first step in developing such system is the collection and formalization of the relevant domain knowledge. The EDEM-CONNECTONTO allows incorporating reliable and validated knowledge in modern chatbot systems such as ChatGPT [14] thus potentially improving the performance of such systems in high-risk applications. This paper presents a systematically developed ontology as a knowledge base for the domain of agitation. The ontology could have different applications, for example, as a codebook for data annotation [15], as a knowledge base for classical rule-based and probabilistic chatbots [13], or as a knowledge base for modern chatbots that rely on large language models (LLMs) and retrieval augmented generation (RAG) [16] to provide more reliable information to the user. An ontology is a detailed specification of a conceptualization where the term conceptualization directs to a simplified domain-specific view of the world that we wish to express [17]. It consists of concepts, relations, and instances. The concepts are defined as the class of entities within a specific domain whereas the instances are the concrete situation-specific "things" defined by the concepts. A relationship is drawn by the interconnections of the concepts. In this paper, we present a systematically developed ontology and its example application as a knowledge base for a chatbot. In general, ontologies provide a systematic and structured way of defining knowledge about a particular domain in a way that can be comprehended and conveyed by different agents. It can be employed in applications such as information retrieval where machines can assert about and comprehend the concepts and relationships within a certain domain. There is a few research on the dementia domain in the form of ontologies. Roantree et al. [18] published the In-MINDD ontology, which models the risk factors that can cause dementia. In [19], the Dem@Care Lab ontology represents the experimentation protocol for dementia assessment. It contributes to the early diagnosis of dementia, inspection of the disease progression, and maintenance of the independent life of PwD. However, there are a few non-pharmacological ontologies available in this domain. In [20], the authors presented an ontology-based context model that can handle agitation behavior in PwD in a context-aware manner. Zhang [21] developed a comprehensive ontology on non-pharmacological interventions, suitable for agitated behaviors of PwD. The Alzheimer’s Disease Ontology (ADO) [22] contains extensive knowledge about alzheimer’s disease, the cause of the disease, medical interventions, medical implications, and any other medical knowledge. The User Profile ontology [23] captures the person-related factors influencing one’s behavior and their changes with the progression of dementia. It addresses the issues on outdoor mobility of PwD. The DREAMDNPTO [24] has been recently published consisting of 1258 unique classes which are focused only on dementia-related emotional and mood disturbance non-pharmacological interventions. The Situation-Aware Navigation Assistance for Dementia Patients using Causal Behavior Models (SinDem) ontology [25] contains knowledge relevant to supporting the outdoor mobility of people in the early stages of dementia. It has been used to annotate the outdoor behavior of PwD during a guided walk recorded with a camera and a variety of sensors. Unfortunately, no non-pharmacological ontology is available with concepts addressing a detailed analysis of the agitation behavior of PwD, the relevant causes that trigger the agitated behavior of the PwD, consequences that the PwD faces along with their social environment and the dyadic relationship between the PwD and the family carer. Therefore, this study presents EDEM-CONNECTONTO, which provides structured and peer-reviewed knowledge on the agitation behaviour of PwD, relevant causes which trigger the exhibit of agitated behavior of the PwD, consequences that the PwD faces in their social environment and the non-pharmacological interventions that can be taken to mitigate the effect of the agitation of the PwD. The contribution of this work is as follows: 1) we have developed an ontology based on domain analysis, literature reviews, interviews with caregivers, brainstorming with domain experts and conducting expert workshops; 2) we have validated the ontology through automated reasoning and employed as a part of an annotation task; 3) the ontology is publicly available and could be used in various applications addressing the domain of PwD. 2 Methods In this section we describe the methods and materials used for the development of the ontology. Figure 1 provides an overview of this process. 2.1 The process of knowledge extraction First, we determined the requirements that the EDEM-CONNECTONTO needs to satisfy. The ontology requirements specification includes the information about the objective, the scope, the target group, the implementing tool, and the intended use cases of the ontology (see details in the supplementary materials) and then we followed a general ontology development process proposed in [26]. We developed a domain-specific ontology, as existing ontologies only cover some of the topics, we are interested in. An initial ontology was developed by performing a systematic literature review following the process proposed in [25], which incorporates domain analysis, conceptualization, implementation, and evaluation (see Figure 1). 2.1.1. Domain analysis Domain analysis involves analyzing different scenarios, requirement analysis, identification of existing ontologies that can be partially integrated into the EDEM-CONNECTONTO, discussion of the domain-specific issues, observation of use cases, identification of domain-specific knowledge through questionnaires and interviews conducted by the domain experts, and review of existing solutions. To gather domain knowledge on agitation of PwD and on strategies for dealing with agitation as well as on aspects promoting the stability of home-based care arrangements (dyadic relationship), several activities have been performed: (1) literature review of scientific literature, (2) qualitative interviews with family-caregivers, (3) expert workshop. The collection and analysis of data were carried out by experts in the domain of dementia. In the conceptualization phase, domain experts identify relationships between relevant concepts. The initial ontology is implemented in web ontology language (OWL) [27] and then automatically validated by logical inference. As the initial ontology is relatively general and incomplete, the ontology is further developed in the second phase. The second phase addressed the further development of the ontology on new knowledge resulting from user questionnaires and interviews, new system requirements, and expert knowledge not identified in the first phase. Then, this knowledge is integrated into the initial ontology. The last step of the 2nd phase of the development process is further model implementation and validation by logical interference in the Protégé 5.5.0 software [28]. Parts of the existing ontology are additionally integrated into the ontology to ensure reusability, and the ontology model is validated by the domain expert, who checks for completeness and clarity. The ontology from the second phase provides a relatively complete and validated ontology (see Section 3). 2.1.1.1 Literature In the literature research, scientific papers were evaluated without any discrimination based on race, sex, migration or ethnical background, or the geographical origin of the involved papers. At the beginning and continuously in the course of the project, literature was searched in the databases Cinahl, Scopus, Pubmed, and Google Scholar using the search terms "agitation", "dementia", "challenging behavior", "psychosocial interventions" and “dyadic relationship”. The deductive analysis of the literature was based on approaches to managing agitation of PwD, in particular the DICE approach [29] and the innovative dementia orientated Assessment system (IdA-assessment) [30]. After analyzing the DICE approach [29], 55 sub-concepts were added as the sub-concepts of "Causes". Based on the effects of the agitation of the PwDs, 12 concepts were categorized as the sub-concepts of "Consequence" extracted from the [31], [29] study. Related to dyadic relationships, an adapted approach combining deductive and inductive coding was used with the SoCA-Dem theory [32] serving as an analytical framework [33]. "Job/employment" was classified as a sub-concept of "Family-carer" [34], extracted from the SoCA-Dem theory [32] study. These approaches provide different steps of describing and analyzing the behavior and lead to developing a proposal for managing the behavior. Through carefully reading the literature, the key concepts and their inter-relationships related to the PwDs domain were extracted and added to the EDEM-CONNECTONTO. We reused knowledge resources for the 37 sub-concepts "Agitation" from the Cohen-Mansfield agitation inventory (CMAI) [35] study. From the IdA-assessment [30], we extracted topographical knowledge on the agitation of the PwD and the background knowledge of the PwDs. Subsequently, we identified 34 concepts related to the topographical knowledge of the agitation of the PwD, which were categorized as sub-concepts of "Agitation". Additionally, we pinpointed 18 distinctive concepts as sub-concepts of "Person with dementia". 2.1.1.2 Interviews with family-caregivers Qualitative interviews were conducted with 12 family caregivers of PwD to validate the ontology concepts. The interviews, structured by an interview guide (see interview questions in the supplementary materials), have been digitally recorded and analyzed using content analysis. The study participants were recruited without any differentiation between race, sex, migration or ethnical background. It is to be noted that all research was performed in accordance with relevant guidelines and informed consent was obtained from all the study participants. The ethical approval was given by the Ethic Committee of the German Society of Nursing Science on the 3rd of November 2020 and the registration number is 20-022. These were the criteria for family caregivers to be involved in this study: anyone who defines themselves as a family carer of a PwD should be involved in developing and evaluating the platform in this project and agitation should be present in the PwD. According to studies, almost every PwD develops so-called behavioral and psychological symptoms of dementia in the course of the disease. We have not carried out any tests to determine whether PwD actually have or has had agitation. It was much more important to us to gather the relatives' experiences, which they describe as agitation or restlessness. This would give us a better picture of the reality of life in the home. We, therefore, included all relatives who were able and willing to report on difficult situations in care associated with agitation and restlessness. There were no criteria defined for measuring dementia severity. The concepts extracted from the literature were used as a framework for analysis. Additional concepts were developed inductively on the data. In addition, 18 qualitative interviews with family caregivers of PwD from the UK MARQUE study [36] have been analyzed using secondary analysis. The MARQUE project consisted of six work streams, the results of which should contribute to a better understanding of agitation and dementia and the workstreams explored how agitation affects the relationship between relatives and PwD and what helps in supporting the PwD. Subsequently, 27 concepts were added as the sub-concepts of "Interventions" based on the data, collected from the interviews. For the secondary analysis of these interviews, the same analytical framework was used in the evaluation of the five interviews with family caregivers. 7 additional concepts have been developed inductively from the data. The interviews were conducted during the COVID-19 pandemic, which made the process of recruitment very difficult especially for the vulnerable group of family caregivers. 2.1.2 Workshops with experts In an expert workshop with 6 experts in counseling relatives of the PwD, the preliminary ontology has been presented and discussed. About completeness, hints were given on the linguistic adaptation of terms to the target group. Fourteen concepts mainly related to the causes of agitation were also added and prioritized to complete the ontology. 2.1.3 Integration of existing ontology models The eDEM-CONNECT ontology reused concepts from the DRANPTO ontology [21], which focuses on the nonpharmacological management of agitation in dementia. We mapped 65 concepts describing a range of agitated behavior from this ontology. 2.2 Ontology conceptualization Extracted knowledge in dementia was structured into the conceptual model. At first, the important terms representing the nature of agitated behavior and the causes of being agitated were elicited from the selected knowledge source and coded in the ontology development tool. These elicited terms were defined as nouns and verbs. Here, we identified nouns as concepts and individuals, and then added concepts that include definitions, labels, abbreviations, and synonyms. The class hierarchy was constructed following the subsumption relationship (e.g., is-a-superclass-of). For instance, the terms "Verbal_agrressive" and "Verbal_nonagrressive" are the sub-concepts of "nature" whereas "nature" is the sub-concept of the most-hierarchical concept of "Agitation". The elicited verbs were defined as object properties for determining inter-relationships between the concepts within the domain and range. For instance, the verb "exhibits" was elicited as an object property for defining the relationships between the classes "Person with dementia" and "Agitation", where the class "Person with dementia" is denoted as domain, and the class "Agitation" is denoted as the range of the object property. 2.3 Ontology development Generally, an ontology consists of concepts, relations, and instances. The concepts are defined as the class of entities within a specific domain, whereas the instances are the concrete situation-specific "things" defined by the concepts. A relationship is drawn by the interconnections of the concepts. An ontology and a set of individual instances constitute a knowledge base [37]. The ontology is designed in OWL [27] using the Protégé 5.5.0 software [28] where the OWL is a semantic web language developed to introduce information about the concepts, individuals, and relations between the concepts [28]. In our initial version of the ontology, we defined concepts such as “Person”, “Agitation”, and “Causes” as shown in Figure 2. We used a top-down hierarchical approach, which starts with the most general concepts, e.g., “Person” and “Agitation”, and subsequent sub-concepts, e.g., “Family-Carer” and “Nature”. The EDEM-CONNECTONTO includes types of agitation behaviors, the causes that trigger such behaviors, the consequences that a PwD might experience, and the possible interventions. Here, we also investigated types of family caregivers (e.g., friends, family carers, neighbors) of the PwD and their social activities. 2.4 Ontology evaluation The EDEM-CONNECTONTO was assessed in 5 steps. It started with the automatic logical reasoner, which evaluates the consistency of the model. The Protégé software and its reasoner were used to validate the ontology model during the implementing phase. In our case, Pellet [38] and HermiT [39] reasoners were used to identify redundant links in the ontology. Besides, the EDEM-CONNECTONTO was assessed to check whether it could answer the competency question (CQ) using the SPARQL query language. The EDEM-CONNECTONTO was evaluated by the standards of the biomedical ontology in terms of accuracy, clarity, completeness, conciseness, and consistency [40]. At last, two dementia care experts, an experienced professor of the Nursing Department at the University of Witten and an experienced manager of the eDEM-Connect project, were asked to manually assess the EDEM-CONNECTONTO in terms of accuracy, clarity, and completeness. Two annotators were also asked to annotate dementia-related data using an annotation codebook developed from the EDEM-CONNECTONTO. The annotated corpus was employed to assess whether the concepts defined in the EDEM-CONNECTONTO are recognizable by the annotator in dementia-related informal texts. 3 Results 3.1 The EDEM-CONNECTONTO The EDEM-CONNECTONTO has 241 concepts, 8 relations (see Table 1), and 240 individuals based on the agitation of PwD, social and physical environment, family members, causes, consequences, health issues of the PwD, and intervention strategies. The analysis of the literature has produced 142 concepts and 4 relationships between concepts. From the interviews with the family caregivers, 27 new concepts and 4 new relationships were added. 7 newly added concepts were also developed from the 18 qualitative interviews with family caregivers of PwD from the UK MARQUE study. The oval shapes in Figure 2 illustrate the concepts; the relationship (object property) is marked as a solid line that connects two concepts, and the subsumption relationship (data property) is denoted as a dotted line. The EDEM-CONNECTONTO is now publicly available at the NCBO BioPortal [41]. The 5 most abstract concepts in the EDEM-CONNECTONTO are “Person”, “Agitation”, “Causes”, “Consequences” and “Interventions”. As it is sensible to generate deeper knowledge and further specify the problem with additional diagnostic questions as child concepts, we added concepts in a top-down hierarchical way. For instance, the "Agitation" concept has the sub-concept "nature", which is then divided into many sub-sub concepts e.g. "verbal_aggressive" and "verbal_nonaggressive". In Figure 2, we created "exhibits" relations between the concept "Person with dementia" and "Agitation", marked with a solid line’s arrow. The last step was creating individual instances of concepts in the hierarchy. As no single “correct” way or methodology for developing ontologies exists, we debugged and evaluated it with the domain experts. This iterative design process continued through the entire ontology development cycle. The most abstract concept “Person” is categorized by considering the living conditions, activities, emotions, health conditions, and abilities of the PwD and the non-professional family carer who are responsible for providing long-term support to the PwD, shown in Figure 2 and 3. “Family-Carer” refers to the people who stay with the PwD or unprofessionally take care of the PwD. Generally, family caregivers and informal caregivers are unpaid family members, colleagues, and relatives who support the person undergoing acute or chronic conditions and want assistance to conduct various tasks, like bathing or dressing. An overview of the sub-subconcepts of the concept “Person with dementia” is depicted in Figure 3. Table 1: Relationships between the concepts and their definitions Relationship Definition exhibits Indicates relation between ‘PwD’ and ‘Agitation’, ‘Verbal-aggressive’, ‘Verbal-nonaggressive’, ‘Physical-aggressive’, ‘Physical-nonaggressive’ concepts. triggers Indicates the relation between the ‘Causes’ and ‘Agitation’ concepts leads Indicates the relation between the ‘Agitation’ and ‘Consequences’ concepts accepts Indicates the relation between the ‘Acceptance’ and ‘Agitation’ concepts prevents Indicates the relation between the ‘Prevention’ and ‘Agitation’ concepts reduces Indicates the relation between the ‘Reduction of negative consequences’ and ‘Consequences’ concepts resolves Indicates the relation between the ‘Problem-solving’ and ‘Causes’ concepts is-a Denotes that the concept is a sub-concept of another concept Agitation can also be defined as any inappropriate physical or verbal, aggressive or non-aggressive act that is not an obvious expression of need or disorientation. It can be defined as emotional distress such as anger, mood swings, negativism, or outbursts. During the challenging behavior analysis of the PwD, four types of agitation-based challenging behaviors were identified. These are “Physical_aggressive”, “Physical_nonaggressive”, “Verbal_aggressive” and “Verbal_nonaggressive”. Apart from them, the location, atmosphere, and topography of the PwD were added to the ontology (see Figure 4). Our domain experts analyzed the general causes that lead to the above-mentioned challenging behavior for the PwD. Based on the analysis of the general causes of agitated behavior in PwD, we added three top-level sub-concepts under the “Causes” category: “Factors related to the person with dementia”, “Environmental factors”, and “Interpersonal factors”. In the ontology, one of the most higher-up concepts of “Causes” is “environmental factors” which is categorized as the monotony of the nursing home environment, location of the PwD, the color of the room, lack of natural material, dark places, noise (e.g., fan noise, television noise, water noise), temperature, loneliness, activities, weather. For example, if a person lives in an isolated place, it can be responsible for the agitated behavior of the patient. “Factors related to the person with dementia” and “Interpersonal factors” are also added as the sub-concepts of the concept “Causes” that could be the reasons for the challenging behavior of PwD. All the concepts of the EDEM-CONNECTONTO with the definitions are published in BioPortal [41] at https://bioportal.bioontology.org/ontologies/EDEM-CONNECTONTO. Agitated behavior of the PwD might lead to physical and psychological stress on the caregivers taking care of the PwD, such as heavy workload, disruptions to daily care routines, depression, and poor quality of life ( see [42], [43],[44], [45]). There are a few other adverse outcomes that may include the PwD’s premature institutionalization [46]. Our domain experts identified these outcomes that the PwD might experience and added these outcomes as concepts in the EDEM-CONNECTONTO ( see the sub-concepts of "Consequences" concepts in BioPortal at https://bioportal.bioontology.org/ontologies/EDEM-CONNECTONTO). In general, family-caregivers support the PwD to manage agitation in order to minimize the likelihood of adverse outcomes. Probable interventions are studied that could help the PwD to improve their conditions. Thus, we included interventions based concepts such as “Prevention”, “Problem solving”, “Acceptance”, “Reduction of negative consequences” (see Figure 5). 3.2 Results of evaluating the ontology model by the competency questions using SPARQL queries The capability of the EDEM-CONNECTONTO to answer the competency questions (CQ) was evaluated using SPARQL query [47] which is defined as a semantic query language for retrieving and manipulating data stored in the ontology that are described in resource description framework (RDF) format. By using SPARQL query language, we can retrieve data from the EDEM-CONNECTONTO based on the 12 CQ. The retrieved data were checked to see if the EDEM-CONNECTONTO could deliver the correct response for each CQ. To assess the capability of the developed model to answer the CQ, each CQ was represented by SPARQL queries to retrieve data from the ontology. For instance, CQ 1: “What are the types of the agitated behaviors, described for the PwD?” in SPARQL language was prefix rdf: prefix rdfs: prefix owl: SELECT ?subclass ?subclassLabel ?subclassDefinition WHERE { ?subclass rdfs:subClassOf myont:Nature . ?subclass rdfs:label ?subclassLabel . OPTIONAL { ?subclass rdfs:comment ?subclassDefinition } ?individual rdf:type myont:Nature . } We can retrieve information on the types of agitated behavior along with the definitions from CQ 1. The retrieved data are the 5 types of agitated behavior. The retrieved outputs are the name of the concepts (“Strong_emotion”, “Verbal_aggressive”, “Verbal_nonaggressive”, “Physical_aggressive”, “Physical_nonaggressive”) along with the “rdfs:label,” “rdfs:comment,” and “definitions”. The 12 CQ, along with the retrieved output and definition of one of the concepts, are presented in Table 2. We calculated accuracy as the primary evaluation metric to measure the overall correctness of the model. This involved calculating the ratio of the number of accurate predictions to the absolute number of predictions. We specifically evaluated the model’s accuracy with the 12 CQ (see Section 3.3). The consistency between the retrieved data and the actual developed concepts was a key indicator that EDEM-CONNECTONTO can deliver the correct response for each query. As a result, we achieved 100% accuracy for each query. Table 2: We retrieved the output from the ontology model based on the competency question (CQ). The consistency between the retrieved outcomes and the developed concepts indicates that the EDEM-CONNECTONTO provides the correct response for each CQ. CQ Retrieved output Definition of the concepts What are the types of agitated behaviors described for the person with dementa? ‘Strong_emotion’, ‘Verbal_aggressive’, ‘Verbal_nonaggressive’, ‘Physical_aggressive’, and ‘Physical_nonaggressive’ Definition of ‘Physical_aggressive’: ‘It refers to the physically-aggressiveness behaviors of the people with dementia. For example: biting, kicking, hurting and others’ (see [35]). What are the causal factors of agitation in person with dementia? ‘Environmental factors’, ‘Interpersonal factors’, and ‘Factors related to the person with dementia’ Definition of ‘Environmental factors’: ‘It refers to the environmental factors that can cause agitation’ (see [35]). What are the causal environmental factors of agitation in person with dementia? ‘Lack of activity’, ‘Lack of established routines’, ‘Safety issues’, ‘Change in routines’, ‘Light level’, and ‘Noise level’ Definition of ‘Lack of activity’: ‘It refers to lack of appropriate engaging activities for a people with dementia that can cause agitation’ (see [35]). What are verbal_aggressive behaviours? ‘Screaming’, ‘Cursing’, ‘Safety issues’, and ‘Making verbal sexual advances’ Definition of ‘Screaming’: ‘It refers to speaking loudly / shouting, high-pitched vocalisations’ (see [35]). What are physical_nonaggressive behaviours ? ‘Aimless wandering’, ‘Pacing’, ‘Hyper activity’, ‘Hiding things’, ‘Intentional falling’, ‘Handling things inappropriately’, ‘Eating inappropriate substances’and ‘Performing repetitive mannerism’ Definition of ‘Aimless wandering’: ‘It refers to a certain type of behaviour that includes walking aimlessly from place to place’ (see [35]). What is the background information of person with dementia? ‘Mood and emotion’, ‘Personality’, ‘Lifestyle before dementia’, ‘Age of the person with dementia’, ‘Preferences’, and ‘Resources’. Definition of ‘Resources’: ‘It refers to the resources for people with dementia to assist individuals in managing their condition and enhancing their quality of life (e.g., support systems, services, and tools available to assist the people with dementia)’ (see [35]). Who are the informal caregivers of person with dementia? ‘Family_Carer’, ‘Neighbours’, ‘Friends’, and ‘Further_family_members’ Definition of ‘Family_Carer’ : ‘It refers to the informal caregiver who takes care of the person suspected/diagnosed with dementia’ (see [35]). What are the consequences for person with dementia experiencing agitation due to dementia? ‘Disability’, ‘Institutionalisation early placement in a nursing home’, ‘Decreased quality of life’, ‘Anorexia weight loss’, ‘Hospital admission’ and ‘Poor caregiver outcomes’ Definition of ‘Decreased quality of life’: ‘It refers to the concept that the agitation has an effect on the individual's wellbeing and the ability to enjoy life’ (see [35]). What non-pharmacological interventions are employed to mitigate the effect of agitation in dementia? ‘Maintain routines’, ‘provide for social contacts’, ‘Promote well-being’, ‘Improve the quality of relationship between the PWD and the relative’, and ‘Identify unmet needs and meet them when possibles’ Definition of ‘Maintain routines’: ‘It refers to the action of the family carer that maintains the care routine of a people with dementia according to his/her needs’ (see [35]). What intervention can be used for verbal_aggressive behaviour of the person with dementia? ‘Distract’, ‘Screening for physical/medical causes’, ‘Environmental design’, ‘Recognize and avoid triggers’ and ‘Screening for physical/medical causes’ Definition of ‘Distract’: ‘It refers to the intervention taken by the family carer which is to distract the person with dementia from problematic situations to another task, such as, to offer a snack or put on some familiar music to interrupt behaviours that are becoming difficult’ (see [35]). What are the interpersonal factors that are responsible for verbally non-agressive behavior? ‘Lack of caring skills’, ‘Language of the caregiver’, ‘verbal interaction’ and ‘Low mental capabilities’ Definition of ‘Lack of caring skills’: ‘It refers to the lack of caring skills of the caregivers to deal with person with dementia’ (see [35]). What intervention can be used to cope with the screaming of the PwD? ‘Communication: controlling the emotional tone of your own language’ and ‘Distract’ Definition of ‘Distract’: ‘It refers to the intervention taken by the family carer which is to distract the person with dementia from problematic situations to another task, such as, to offer a snack or put on some familiar music to interrupt behaviors that are becoming difficult (see [35]). 3.3 Results of evaluating EDEM-CONNECTONTO against biomedical ontology criteria EDEM-CONNECTONTO was assessed with respect to consistency, clarity, completeness, and accuracy following the biomedical ontology standards [40]. • Accuracy: We checked the definitions, descriptions, and properties of the concepts. They meet the biomedical ontology standard (see [40]) as the concepts and their definitions were extracted from the scientific papers and assessed by the domain experts. • Completeness: It measure the ontology's domain knowledge coverage. Two domain experts and an experienced manager of the eDEM-CONNECT project evaluated it manually. The 12 CQs were also used to assess the completeness of the EDEM-CONNECTONTO. • Consistency: It means that the ontology does not keep or permit any contradictions. We used Pellet and Hermit Reasoners, and the results indicate that the ontology model is consistent. • Clarity: It ensures that the EDEM-CONNECTONTO is unambiguous, with well-defined meanings for all concepts and relationships. Names of concepts and their definitions should b e comprehensible. Clarity of this ontology is accomplished by assigning a specific label with a relevant explanation to the concepts individually using “rdfs: label,” “rdfs:comment,” and “definitions” (see an illustration about the annotation of the “Person with dementia” in Figure 6). It also ensures that the ontology can communicate properly to the readers with a relevant explanation of the concepts and the relationships between these concepts. 3.4 Evaluation by domain experts Our domain experts reviewed and assessed the ontology model manually for measuring accuracy and completeness. Their revision on EDEM-CONNECTONTO is viewed as a genuine assessment. Based on their suggestions, we added four new concepts as the ontology's sub-subconcepts of the "Interventions" concept. They were: “Self-care: taking time for yourself”, “Looking for distractions”, “Safety for PWD: provide a safe environment” and “Safety for relatives: leave the room, leave PWD alone” as a subclass of “Reduction of negative consequences”. In addition, following the domain expert’s recommendation about redundancy, the term “Talking about feelings and experiences of the PwD” was removed from the subclass of “Reduction of negative consequences”. The concepts “distress”, “reduced income from employment” and “depression” were also removed from the subconcepts of the “Consequences” concept. The domain experts and the project manager then evaluated the modified ontology. They verified its accuracy and completeness without additional revision. 3.5 Data annotation We proposed an annotation scheme that is based on the 8 most general concepts and relations from the EDEM-CONNECTONTO for annotating textual data from online dementia forums. "Agitation", "Verbal_aggressive", "Verbal_nonaggressive", "Physical_aggressive", "Physical_nonaggressive", "Cause", "PwD", "Family-carer" are the labels used for the annotation task (see details in [15]). Two annotators were employed to label instances independently of the entities (concepts) and the relationships in the text corpus. We used Cohen’s kappa score [48] as a measure of interrater reliability to evaluate the quality of the annotated data. We observed that the annotators mainly annotated entities containing noun phrases (e.g., father, he) based on “PwD” and “Family-carer” labels compared to the labels describing types of agitation of PwD (e.g., verbal_nonaggressive, physical_aggressive) (see [15]). We found that the domain of PwD is difficult to conceptualize. Our domain experts and the annotators struggled with identifying the concepts and relationships that define the domain. Nonetheless, the outcomes demonstrated that both annotators could observe the labels from the concepts in EDEM-CONNECTONTO with moderate agreement. 4 Discussion The eDEM-CONNECT project has succeeded in developing a comprehensive ontology for managing the agitation of PwD. This ontology consists of 241 concepts that can be helpful in various aspects of the care of PwD. It is incorporated as a knowledge base for the eDEM-Connect chatbot (see [13]) where SPARQL query is used to search associated questions in the ontology and is forwarded to the user as the response. The eDEM-CONNECT chatbot dedicated to support the family caregivers of PwD on agitation topics, empowering them by providing structured knowledge which can ease their daily difficulties with their relatives with dementia. Both domain experts and family caregivers have been involved from the beginning of the ontology development phase. This has resulted in scientific findings and practical knowledge being incorporated into the ontology. A unique characteristic of this ontology is its focus on family caregivers. This is also reflected in the fact that aspects of the dyadic relationship have been elaborated in detail, both in terms of the possible causes of agitation and especially in terms of interventions. This distinguishes the ontology from comparable ontologies, such as the DRANPTO [21] ontology. The first challenge was identifying the relevant resources to extract knowledge in the dementia domain. We focused on the types of agitation, non-pharmacological interventions, family caregivers, and the socio-economic condition of the PwD, as no standardized document supplies vast coverage of evidence-based knowledge. We observed that it was challenging for the domain experts to define the concepts after extracting relevant information. The second challenge was looking for an efficient ontology assessment approach for the EDEM-CONNECTONTO because of the need for standardized methods to assess ontologies. A few ontology assessment approaches have been developed to determine the quality of ontology models, such as the gold standard assessment, domain expert-based assessment, tool-based assessment approach, CQ-based assessment, and quality criteria-based assessment [40]. Generally, ontology developers use any one of the assessment approaches or a mixture of these assessment methods to measure the quality of ontology models (see [40], [49]). Therefore, we evaluated the EDEM-CONNECTONTO using five approaches that include 1) evaluation by automatic reasoner 2) answering SPARQL queries 3) applying ontology quality criteria e.g., accuracy, clarity 4) evaluating the ontology by domain experts, and 5) using concepts from the ontology for an annotation task. This combined assessment approach has generated excellent results. The EDEM-CONNECTONTO satisfies the quality standards defined in [40]. Another challenging aspect was engaging domain experts to assess the EDEM-CONNECTONTO manually. In the domain expert–based assessment method, domain experts need to manually check all entities (concepts and relationships) of the suggested ontology. As this method requires lots of effort and time from domain experts, it is considered a costly assessment approach. Two domain experts and the eDEM-CONNECT project manager manually assessed the EDEM-CONNECTONTO concerning accuracy, clarity, and completeness. We also conducted 2 workshops where the concepts and the relationships were discussed to ensure the clarity of the EDEM-CONNECTONTO. The EDEM-CONNECTONTO has many potentials. One of the prospects is that we can convert dementia-specific textual data into machine-processable data by producing semantic annotations that directly map a data element to an ontology concept. Another potential of the EDEM-CONNECTONTO is to be employed as an annotation codebook for data annotation tasks where the annotated corpus could be used for concept and relationship extraction tasks in the domain of PwD and the annotated corpus will generate more insights about agitation in dementia. This, in turn, will enable the training of machine learning models that could support PwD and their caregivers in challenging situations and potentially improve their overall wellbeing. From the beginning, we have involved both domain experts and the family caregivers of the PwD in the ontology’s development phase. However, the participation of family caregivers in fundamental decisions on the selection of knowledge for ontology could not be learned due to restrictions during the COVID-19 pandemic. The EDEM-CONNECTONTO is limited to the non-clinical intervention strategies for managing agitation in dementia as we have a considerable volume of clinical data that are publicly available for addressing agitation in dementia. The EDEM-CONNECTONTO has a particular value for nursing science. This is highly relevant because one of the existing ontologies concerning nursing, the International Classification for Nursing Practice (ICNP) Ontology, does not contain any concepts that are important for dealing with the agitation of PwD. The structured exploration of the ontology concepts clearly showed weak evidence on the individual concepts, especially on the weighting and causal relationships between the concepts. Most of the literature originates from the field of medicine. Structured nursing knowledge of the concepts analyzed needs to be improved. In particular, there is a lack of evidence on interventions that are effective for family caregivers in specific situations. In the future, we plan to conduct workshops with the family caregivers of the PwD to define more concepts and relationship entities on the intervention strategies. We plan to annotate more textual unstructured data for the concepts and relationships extraction task using a machine learning approach, which will be integrated into the EDEM-CONNECTONTO. Furthermore, EDEM-CONNECTONTO will also be extended by incorporating information from additional sources (interviews and group discussions with relatives as well as publicly available sources such as online forums). As mentioned before, the vision we have is a chatbot that relies on the domain knowledge from the ontology and through interaction with the user it could provide him or her with helpful information regarding a given agitation-related situation with a PwD. During the project we have implemented a minimal probabilistic chatbot [13] that relies on parts of the ontology. In the future we plan to use modern LLM architectures and combine them with our ontology either in the form of RAG, through fine-tuning or even by incorporating an additional logical layer on top of the LLM to ensure the validity of the provided information. Declarations Acknowledgements We recognize the efforts of our annotators in annotating the dementia forum texts. Fundings This work was supported by the German Federal Ministry of Education and Research (BMBF, reference number: 16SV8335) Conflict of Interest The authors have no conflict of interest to report. Data availability The EDEM-CONNECTONTO ontology is publicly available at the NCBO BioPortal at https://bioportal.bioontolog y. org/ontologies/EDEM-CONNECTONTO . Additional information The authors declare no competing interests. AUTHOR CONTRIBUTIONS STATEMENT S.S., C.P., I.H., and K.Y. conceived the idea. All authors discussed the results and commented on the paper. C.P., I.H., M.H., and B.H. conceptualized all the concepts of the ontology and S.S. designed, developed and evaluated the EDEM-CONNECTONTO ontology. All authors reviewed the definition of the ontology and the S.S., C.P., I.H., and K.Y. wrote the paper. References Dementia (2023) World Health Organization. Available at: https://www.who.int/news-room/fact-sheets/detail/dementia (Accessed: 11 June 2024). Thyrian JR, Eichler T, Hertel J, Wucherer D, Dreier A, Michalowsky B, Killimann I, Teipel S, Hoffmann W. Burden of behavioral and psychiatric symptoms in people screened positive for dementia in primary care: results of the DelpHi-study. Journal of Alzheimer's Disease. 2015 Jan 1;46(2):451-9. Cummings J, Mintzer J, Brodaty H, Sano M, Banerjee S, Devanand DP, Gauthier S, Howard R, Lanctôt K, Lyketsos CG, Peskind E. Agitation in cognitive disorders: International Psychogeriatric Association provisional consensus clinical and research definition. International psychogeriatrics. 2015 Jan;27(1):7-17. Brijoux T, Zank S. Auswirkungen kognitiver Einschränkungen (Demenz) auf Lebensqualität und Versorgung. InHohes Alter in Deutschland 2023 Jul 5 (pp. 173-195). Berlin, Heidelberg: Springer Berlin Heidelberg. Yaffe K, Fox P, Newcomer R, Sands L, Lindquist K, Dane K, Covinsky KE. Patient and caregiver characteristics and nursing home placement in patients with dementia. Jama. 2002 Apr 24;287(16):2090-7. George C, Ferreira N, Evans R, Honeyman V. A systematic review of the association between individual behavioural and psychological symptoms in dementia and carer burden. Working with Older People. 2020 Sep 18;24(3):181-203. De Vugt ME, Stevens F, Aalten P, Lousberg R, Jaspers N, Winkens I, Jolles J, Verhey FR. Behavioural disturbances in dementia patients and quality of the marital relationship. International journal of geriatric psychiatry. 2003 Feb;18(2):149-54. Pozzebon M, Douglas J, Ames D. Spouses’ experience of living with a partner diagnosed with a dementia: A synthesis of the qualitative research. International psychogeriatrics. 2016 Apr;28(4):537-56. Luiu AL, Favez N, Betrancourt M, Szilas N, Ehrler F. Family relationships and alzheimer’s disease: A systematic review. Journal of Alzheimer's Disease. 2020 Jan 1;76(4):1595-608. Stephan A, Bieber A, Hopper L, Joyce R, Irving K, Zanetti O, Portolani E, Kerpershoek L, Verhey F, de Vugt M, Wolfs C. Barriers and facilitators to the access to and use of formal dementia care: findings of a focus group study with people with dementia, informal carers and health and social care professionals in eight European countries. BMC geriatrics. 2018 Dec;18:1-6. Vecchio N, Fitzgerald JA, Radford K, Fisher R. The association between cognitive impairment and community service use patterns in older people living in A ustralia. Health & social care in the community. 2016 May;24(3):321-33. Michalowsky B, Kaczynski A, Hoffmann W. Ökonomische und gesellschaftliche Herausforderungen der Demenz in Deutschland–Eine Metaanalyse. Bundesgesundheitsblatt-Gesundheitsforschung-Gesundheitsschutz. 2019 Jul 11;62(8):981-92. Boiting M, Tschorn N, Suravee S, Yordanova K, Halek M, Jagoda FA, Lüdtke S, Burmann A. eDEM-CONNECT: An Ontology-Based Chatbot for Family Caregivers of People with Dementia. InInternational Conference on Wireless Mobile Communication and Healthcare 2023 Nov 29 (pp. 73-91). Cham: Springer Nature Switzerland. De Angelis L, Baglivo F, Arzilli G, Privitera GP, Ferragina P, Tozzi AE, Rizzo C. ChatGPT and the rise of large language models: the new AI-driven infodemic threat in public health. Frontiers in Public Health. 2023 Apr 25;11:1166120. Suravee S, Stoev T, Schindler D, Hochgraeber I, Pinkert C, Holle B, Halek M, Krüger F, Yordanova K. Annotation scheme for named entity recognition and relation extraction tasks in the domain of people with dementia. In2022 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops) 2022 Mar 21 (pp. 236-241). IEEE. Lewis P, Perez E, Piktus A, Petroni F, Karpukhin V, Goyal N, Küttler H, Lewis M, Yih WT, Rocktäschel T, Riedel S. Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in neural information processing systems. 2020;33:9459-74. Gruber TR. A translation approach to portable ontology specifications. Knowledge acquisition. 1993 Jun 1;5(2):199-220. Roantree M, O’Donoghue J, O’Kelly N, Pierce M, Irving K, Van Boxtel M, Köhler S. Mapping longitudinal studies to risk factors in an ontology for dementia. Health Informatics Journal. 2016 Jun;22(2):414-26. Meditskos G. Multimodal sensing and fusion for comprehensive monitoring and feedback: the integrated Dem@ Care system.2015. Fook VF, Tay SC, Jayachandran M, Biswas J, Zhang D. An ontology-based context model in monitoring and handling agitation behavior for persons with dementia. InFourth Annual IEEE International Conference on Pervasive Computing and Communications Workshops (PERCOMW'06) 2006 Mar 13 (pp. 5-pp). IEEE. Zhang Z, Yu P, Chang HC, Lau SK, Tao C, Wang N, Yin M, Deng C. Developing an ontology for representing the domain knowledge specific to non‐pharmacological treatment for agitation in dementia. Alzheimer's & Dementia: Translational Research & Clinical Interventions. 2020;6(1):e12061. Malhotra A, Younesi E, Gündel M, Müller B, Heneka MT, Hofmann-Apitius M. ADO: A disease ontology representing the domain knowledge specific to Alzheimer's disease. Alzheimer's & dementia. 2014 Mar 1;10(2):238-46. Skillen KL, Chen L, Nugent CD, Donnelly MP, Solheim I. A user profile ontology based approach for assisting people with dementia in mobile environments. In2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society 2012 Aug 28 (pp. 6390-6393). IEEE. Zhang Z, Yu P, Yin M, Chang HC, Thomas SJ, Wei W, Song T, Deng C. Developing an ontology of non-pharmacological treatment for emotional and mood disturbances in dementia. Scientific Reports. 2024 Jan 22;14(1):1937. Yordanova K, Koldrack P, Heine C, Henkel R, Martin M, Teipel S, Kirste T. Situation model for situation-aware assistance of dementia patients in outdoor mobility. Journal of Alzheimer's Disease. 2017 Jan 1;60(4):1461-76. Simperl EP, Tempich C. Ontology engineering: A reality check. InOn the Move to Meaningful Internet Systems 2006: CoopIS, DOA, GADA, and ODBASE: OTM Confederated International Conferences, CoopIS, DOA, GADA, and ODBASE 2006, Montpellier, France, October 29-November 3, 2006. Proceedings, Part I 2006 (pp. 836-854). Springer Berlin Heidelberg. Antoniou G, Harmelen FV. Web ontology language: Owl. Handbook on ontologies. 2009:91-110. Musen MA. The protégé project: a look back and a look forward. AI matters. 2015 Jun 16;1(4):4-12. Kales HC, Gitlin LN, Lyketsos CG. Assessment and management of behavioral and psychological symptoms of dementia. Bmj. 2015 Mar 2;350. Bartholomeyczik S, Halek M, editors. Assessmentinstrumente in der Pflege: Möglichkeiten und Grenzen. Schlütersche; 2010 Dec 6. Kong EH. Agitation in dementia: concept clarification. Journal of advanced nursing. 2005 Dec;52(5):526-36. Köhler K, Dreyer J, Hochgraeber I, von Kutzleben M, Pinkert C, Roes M, Holle B. Towards a middle-range theory of ‘Stability of home-based care arrangements for people living with dementia’(SoCA-Dem): findings from a meta-study on mixed research. BMJ open. 2021 Apr 1;11(4):e042515. Thomas J, Harden A. Methods for the thematic synthesis of qualitative research in systematic reviews. BMC medical research methodology. 2008 Dec;8:1-0. Ward-Griffin C, Oudshoorn A, Clark K, Bol N. Mother-adult daughter relationships within dementia care: A critical analysis. Journal of Family Nursing. 2007 Feb;13(1):13-32. Cohen-Mansfield J. Cohen-Mansfield Agitation Inventory. International Journal of Geriatric Psychiatry. 2012. Laybourne, Anne (2019), “Managing agitation and raising quality of life: semi structured interviews with family carers of people living with dementia”, Mendeley Data, V1, doi: 10.17632/s8wtptnhyc.1 Noy NF, McGuinness DL. Ontology development 101: A guide to creating your first ontology. Sirin E, Parsia B, Grau BC, Kalyanpur A, Katz Y. Pellet: A practical owl-dl reasoner. Journal of Web Semantics. 2007 Jun 1;5(2):51-3. Motik B, Shearer R, Horrocks I. Optimized reasoning in description logics using hypertableaux. InAutomated Deduction–CADE-21: 21st International Conference on Automated Deduction Bremen, Germany, July 17-20, 2007 Proceedings 21 2007 (pp. 67-83). Springer Berlin Heidelberg. Amith M, He Z, Bian J, Lossio-Ventura JA, Tao C. Assessing the practice of biomedical ontology evaluation: Gaps and opportunities. Journal of biomedical informatics. 2018 Apr 1;80:1-3. Geller J, Keloth VK, Musen MA. How sustainable are biomedical ontologies?. InAMIA Annual Symposium proceedings 2018 (Vol. 2018, p. 470). American Medical Informatics Association. Cerejeira J, Lagarto L. Behavioral and psychological symptoms of dementia. Frontiers in neurology. 2012 May 7;3:23573. Cubit K, Farrell G, Robinson A, Myhill M. A survey of the frequency and impact of behaviours of concern in dementia on residential aged care staff. Australasian Journal on Ageing. 2007 Jun;26(2):64-70. Feast A, Orrell M, Charlesworth G, Melunsky N, Poland F, Moniz-Cook E. Behavioural and psychological symptoms in dementia and the challenges for family carers: systematic review. The British Journal of Psychiatry. 2016 May;208(5):429-34. Pérez J, Arenas M, Gutierrez C. Semantics and complexity of SPARQL. ACM Transactions on Database Systems (TODS). 2009 Sep 3;34(3):1-45. Kales HC, Gitlin LN, Lyketsos CG. Assessment and management of behavioral and psychological symptoms of dementia. Bmj. 2015 Mar 2;350. Pérez J, Arenas M, Gutierrez C. Semantics and complexity of SPARQL. ACM Transactions on Database Systems (TODS). 2009 Sep 3;34(3):1-45. Kvålseth TO. Note on Cohen's kappa. Psychological reports. 1989 Aug;65(1):223-6. Raad J, Cruz C. A survey on ontology evaluation methods. In Proceedings of the International Conference on Knowledge Engineering and Ontology Development, part of the 7th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management 2015 Nov 12. Additional Declarations No competing interests reported. Supplementary Files EDEMCONNECTONTOSupplementarymaterial.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6289849","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":440634696,"identity":"94c70285-ff0a-4910-8df6-50017b1cf5a0","order_by":0,"name":"sumaiya suravee","email":"data:image/png;base64,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","orcid":"","institution":"Universität Greifswald","correspondingAuthor":true,"prefix":"","firstName":"sumaiya","middleName":"","lastName":"suravee","suffix":""},{"id":440634702,"identity":"83945e87-7c63-43c5-b001-e4ad147ccf3b","order_by":1,"name":"Christiane Pinkert","email":"","orcid":"","institution":"Universität Witten/Herdecke","correspondingAuthor":false,"prefix":"","firstName":"Christiane","middleName":"","lastName":"Pinkert","suffix":""},{"id":440634703,"identity":"b2118339-9849-48ca-9270-0488bf2a6126","order_by":2,"name":"Iris Hochgraeber","email":"","orcid":"","institution":"Seniorenhaus GmbH der Cellitinnen zur heiligen Maria","correspondingAuthor":false,"prefix":"","firstName":"Iris","middleName":"","lastName":"Hochgraeber","suffix":""},{"id":440634707,"identity":"534a5fe6-b5f8-460f-9f8a-38528d5cff28","order_by":3,"name":"Margareta Halek","email":"","orcid":"","institution":"Universität Witten/Herdecke","correspondingAuthor":false,"prefix":"","firstName":"Margareta","middleName":"","lastName":"Halek","suffix":""},{"id":440634710,"identity":"ece751e4-baee-4d8d-99c8-787eea8c37dd","order_by":4,"name":"Bernhard Holle","email":"","orcid":"","institution":"German Centre for Neurodegenerative Diseases (DZNE)","correspondingAuthor":false,"prefix":"","firstName":"Bernhard","middleName":"","lastName":"Holle","suffix":""},{"id":440634713,"identity":"8d46a704-c5cb-4337-9aa6-b8335e8be15e","order_by":5,"name":"Kristina Yordanova","email":"","orcid":"","institution":"Universität Greifswald","correspondingAuthor":false,"prefix":"","firstName":"Kristina","middleName":"","lastName":"Yordanova","suffix":""}],"badges":[],"createdAt":"2025-03-23 18:53:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6289849/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6289849/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":81117224,"identity":"4761fa5f-b37b-4000-b366-2da60507e331","added_by":"auto","created_at":"2025-04-22 11:54:26","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":129215,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic flow of the ontology development process. The development consists of two phases: (1) Initial ontology's development on the collected domain knowledge, (2) further development of the ontology and integration of existing ontologies and new requirements, and then final validation of the ontology by domain experts and automatic inference.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6289849/v1/271618fa4a37ef02a707ea6e.png"},{"id":81118117,"identity":"b1a1c738-05b3-44c0-9556-6342250933dc","added_by":"auto","created_at":"2025-04-22 12:10:26","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":303330,"visible":true,"origin":"","legend":"\u003cp\u003eThe 5 most hierarchical concepts of the EDEM-CONNECTONTO consist of the types of agitated behavior of the PwD as well as the causes that trigger such behaviors, the consequences that PwD might experience, and the possible interventions a carer might select in response to these behaviors.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6289849/v1/ea3677993bed2c1515dcb7ed.png"},{"id":81117226,"identity":"f237cbf2-9777-49aa-8523-e033075d18ea","added_by":"auto","created_at":"2025-04-22 11:54:26","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":356673,"visible":true,"origin":"","legend":"\u003cp\u003eSubconcepts and sub-subconcepts of the most abstract concept “Person with dementia” in the ontology\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6289849/v1/e372c8ce7f5defaca655272d.png"},{"id":81118304,"identity":"6e9a0985-4ab7-4bdf-b199-e26e301ef4f3","added_by":"auto","created_at":"2025-04-22 12:18:26","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":239321,"visible":true,"origin":"","legend":"\u003cp\u003eSubconcept and sub-subconcepts of the most hierarchical concept “Agitation” in the ontology\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6289849/v1/504de56ead04692c40fd21b5.png"},{"id":81117227,"identity":"aa85b163-e8b0-4646-97d9-c14514ab567b","added_by":"auto","created_at":"2025-04-22 11:54:26","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":269984,"visible":true,"origin":"","legend":"\u003cp\u003eIdentified concepts of the EDEM-CONNECTONTO concerning the possible interventions as their subconcepts.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-6289849/v1/b56bd85ceaecdc5176ae6bcd.png"},{"id":81117547,"identity":"806cc58a-b0be-4cc4-8ec8-b8c49333adb1","added_by":"auto","created_at":"2025-04-22 12:02:26","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":161711,"visible":true,"origin":"","legend":"\u003cp\u003eAnnotation of the concept “Person with dementia” in the EDEM-CONNECTONTO.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-6289849/v1/be3492cd95f6f3adc1f67967.png"},{"id":82793181,"identity":"2ea4d0ee-4f35-459a-9c5f-e1780d636443","added_by":"auto","created_at":"2025-05-15 10:24:04","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2435218,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6289849/v1/0953ac71-aed8-4fc4-8625-3934b1063c4e.pdf"},{"id":81117225,"identity":"1c44d01f-00a7-419b-8995-72803fe42651","added_by":"auto","created_at":"2025-04-22 11:54:26","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":23280,"visible":true,"origin":"","legend":"","description":"","filename":"EDEMCONNECTONTOSupplementarymaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-6289849/v1/1200e3ef6dd53f7c70f44aa3.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"eDEM-CONNECT: Agitation ontology for the chatbot-based support of informal caregivers of people with dementia","fulltext":[{"header":"1 Introduction","content":"\u003cp\u003eDementia affects over 55\u0026nbsp;million people worldwide [1]. It leads to limitations in cognitive abilities, such as declining memory as well as changes in behavior or emotional control [1]. The most common form of behavioral change in PwD is agitation [2]. Agitation is persistent or frequently recurring behavior that is primarily manifested in excessive motor activity, leading to significant impairments in interpersonal relationships and the ability to accomplish everyday tasks [3]. In Germany, about 70% of very old PwD live at home and are mainly cared for by relatives and outpatient care services [4]. Dealing with agitated behavior is a major challenge for formal and informal carers. Agitation can lead to early placement of the PwD in a nursing home [5], cause stress to the family caregivers [6], and threaten the stability of home-based care arrangements. Behavioral problems also hurt the quality of the dyadic relationship [7]. Disturbed communication is a big challenge for family caregivers and leads to a negative perception of the relationship [8]. Therefore, the dyadic relationship may influence the development of agitation, which can impact the quality of life of the individuals concerned [9]. To support families in managing agitated behavior, needs-based and situation-specific information and support services are required. Although both formal and informal services are available, their utilization is low and the reasons are diverse [10, 11, 12]. Family caregivers often cannot recognize which information is important for them or have difficulties in applying it to their situation. Family caregivers primarily contact help services via an online search engine, which often finds irrelevant, confusing or outdated content. Thus, there is a strong urge for user-centered and structured information. Our research is taking place within the eDEM-CONNECT project (see [13]), which aimed at developing a chatbot-based communication system that provides structured and comprehensible knowledge to family caregivers. The first step in developing such system is the collection and formalization of the relevant domain knowledge. The EDEM-CONNECTONTO allows incorporating reliable and validated knowledge in modern chatbot systems such as ChatGPT [14] thus potentially improving the performance of such systems in high-risk applications. This paper presents a systematically developed ontology as a knowledge base for the domain of agitation. The ontology could have different applications, for example, as a codebook for data annotation [15], as a knowledge base for classical rule-based and probabilistic chatbots [13], or as a knowledge base for modern chatbots that rely on large language models (LLMs) and retrieval augmented generation (RAG) [16] to provide more reliable information to the user.\u003c/p\u003e \u003cp\u003eAn ontology is a detailed specification of a conceptualization where the term conceptualization directs to a simplified domain-specific view of the world that we wish to express [17]. It consists of concepts, relations, and instances. The concepts are defined as the class of entities within a specific domain whereas the instances are the concrete situation-specific \"things\" defined by the concepts. A relationship is drawn by the interconnections of the concepts. In this paper, we present a systematically developed ontology and its example application as a knowledge base for a chatbot.\u003c/p\u003e \u003cp\u003eIn general, ontologies provide a systematic and structured way of defining knowledge about a particular domain in a way that can be comprehended and conveyed by different agents. It can be employed in applications such as information retrieval where machines can assert about and comprehend the concepts and relationships within a certain domain. There is a few research on the dementia domain in the form of ontologies. Roantree et al. [18] published the In-MINDD ontology, which models the risk factors that can cause dementia. In [19], the Dem@Care Lab ontology represents the experimentation protocol for dementia assessment. It contributes to the early diagnosis of dementia, inspection of the disease progression, and maintenance of the independent life of PwD. However, there are a few non-pharmacological ontologies available in this domain. In [20], the authors presented an ontology-based context model that can handle agitation behavior in PwD in a context-aware manner. Zhang [21] developed a comprehensive ontology on non-pharmacological interventions, suitable for agitated behaviors of PwD. The Alzheimer\u0026rsquo;s Disease Ontology (ADO) [22] contains extensive knowledge about alzheimer\u0026rsquo;s disease, the cause of the disease, medical interventions, medical implications, and any other medical knowledge. The User Profile ontology [23] captures the person-related factors influencing one\u0026rsquo;s behavior and their changes with the progression of dementia. It addresses the issues on outdoor mobility of PwD. The DREAMDNPTO [24] has been recently published consisting of 1258 unique classes which are focused only on dementia-related emotional and mood disturbance non-pharmacological interventions. The Situation-Aware Navigation Assistance for Dementia Patients using Causal Behavior Models (SinDem) ontology [25] contains knowledge relevant to supporting the outdoor mobility of people in the early stages of dementia. It has been used to annotate the outdoor behavior of PwD during a guided walk recorded with a camera and a variety of sensors. Unfortunately, no non-pharmacological ontology is available with concepts addressing a detailed analysis of the agitation behavior of PwD, the relevant causes that trigger the agitated behavior of the PwD, consequences that the PwD faces along with their social environment and the dyadic relationship between the PwD and the family carer. Therefore, this study presents EDEM-CONNECTONTO, which provides structured and peer-reviewed knowledge on the agitation behaviour of PwD, relevant causes which trigger the exhibit of agitated behavior of the PwD, consequences that the PwD faces in their social environment and the non-pharmacological interventions that can be taken to mitigate the effect of the agitation of the PwD.\u003c/p\u003e \u003cp\u003eThe contribution of this work is as follows: 1) we have developed an ontology based on domain analysis, literature reviews, interviews with caregivers, brainstorming with domain experts and conducting expert workshops; 2) we have validated the ontology through automated reasoning and employed as a part of an annotation task; 3) the ontology is publicly available and could be used in various applications addressing the domain of PwD.\u003c/p\u003e"},{"header":"2 Methods","content":"\u003cp\u003eIn this section we describe the methods and materials used for the development of the ontology. Figure 1 provides an overview of this process.\u003c/p\u003e\n\u003ch2\u003e2.1 The process of knowledge extraction\u003c/h2\u003e\n\u003cp\u003eFirst, we determined the requirements that the EDEM-CONNECTONTO needs to satisfy. The ontology requirements specification includes the information about the objective, the scope, the target group, the implementing tool, and the intended use cases of the ontology (see details in the supplementary materials) and then we followed a general ontology development process proposed in [26]. We developed a domain-specific ontology, as existing ontologies only cover some of the topics, we are interested in. An initial ontology was developed by performing a systematic literature review following the process proposed in [25], which incorporates domain analysis, conceptualization, implementation, and evaluation (see Figure 1). \u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003e2.1.1. \u003c/strong\u003e\u003cstrong\u003eDomain analysis\u003c/strong\u003e \u003c/h3\u003e\n\u003cp\u003eDomain analysis involves analyzing different scenarios, requirement analysis, identification of existing ontologies that can be partially integrated into the EDEM-CONNECTONTO, discussion of the domain-specific issues, observation of use cases, identification of domain-specific knowledge through questionnaires and interviews conducted by the domain experts, and review of existing solutions. To gather domain knowledge on agitation of PwD and on strategies for dealing with agitation as well as on aspects promoting the stability of home-based care arrangements (dyadic relationship), several activities have been performed: (1) literature review of scientific literature, (2) qualitative interviews with family-caregivers, (3) expert workshop. The collection and analysis of data were carried out by experts in the domain of dementia. In the conceptualization phase, domain experts identify relationships between relevant concepts. The initial ontology is implemented in web ontology language (OWL) [27] and then automatically validated by logical inference. As the initial ontology is relatively general and incomplete, the ontology is further developed in the second phase. The second phase addressed the further development of the ontology on new knowledge resulting from user questionnaires and interviews, new system requirements, and expert knowledge not identified in the first phase. Then, this knowledge is integrated into the initial ontology. The last step of the 2nd phase of the development process is further model implementation and validation by logical interference in the Protégé 5.5.0 software [28]. Parts of the existing ontology are additionally integrated into the ontology to ensure reusability, and the ontology model is validated by the domain expert, who checks for completeness and clarity. The ontology from the second phase provides a relatively complete and validated ontology (see Section 3).\u003c/p\u003e\n\u003ch4\u003e\u003cstrong\u003e2.1.1.1 Literature\u003c/strong\u003e\u003c/h4\u003e\n\u003cp\u003eIn the literature research, scientific papers were evaluated without any discrimination based on race, sex, migration or ethnical background, or the geographical origin of the involved papers. At the beginning and continuously in the course of the project, literature was searched in the databases Cinahl, Scopus, Pubmed, and Google Scholar using the search terms \"agitation\", \"dementia\", \"challenging behavior\", \"psychosocial interventions\" and “dyadic relationship”. The deductive analysis of the literature was based on approaches to managing agitation of PwD, in particular the DICE approach [29] and the innovative dementia orientated Assessment system (IdA-assessment) [30]. After analyzing the DICE approach [29], 55 sub-concepts were added as the sub-concepts of \"Causes\". Based on the effects of the agitation of the PwDs, 12 concepts were categorized as the sub-concepts of \"Consequence\" extracted from the [31], [29] study. Related to dyadic relationships, an adapted approach combining deductive and inductive coding was used with the SoCA-Dem theory [32] serving as an analytical framework [33]. \"Job/employment\" was classified as a sub-concept of \"Family-carer\" [34], extracted from the SoCA-Dem theory [32] study. These approaches provide different steps of describing and analyzing the behavior and lead to developing a proposal for managing the behavior. Through carefully reading the literature, the key concepts and their inter-relationships related to the PwDs domain were extracted and added to the EDEM-CONNECTONTO. We reused knowledge resources for the 37 sub-concepts \"Agitation\" from the Cohen-Mansfield agitation inventory (CMAI) [35] study. From the IdA-assessment [30], we extracted topographical knowledge on the agitation of the PwD and the background knowledge of the PwDs. Subsequently, we identified 34 concepts related to the topographical knowledge of the agitation of the PwD, which were categorized as sub-concepts of \"Agitation\". Additionally, we pinpointed 18 distinctive concepts as sub-concepts of \"Person with dementia\".\u003c/p\u003e\n\u003ch4\u003e\u003cstrong\u003e2.1.1.2 Interviews with family-caregivers\u003c/strong\u003e\u003c/h4\u003e\n\u003cp\u003eQualitative interviews were conducted with 12 family caregivers of PwD to validate the ontology concepts. \u003c/p\u003e\n\u003cp\u003eThe interviews, structured by an interview guide (see interview questions in the supplementary materials), have been digitally recorded and analyzed using content analysis. The study participants were recruited without any differentiation between race, sex, migration or ethnical background. It is to be noted that all research was performed in accordance with relevant guidelines and informed consent was obtained from all the study participants. The ethical approval was given by the Ethic Committee of the German Society of Nursing Science on the 3rd of November 2020 and the registration number is 20-022. These were the criteria for family caregivers to be involved in this study: anyone who defines themselves as a family carer of a PwD should be involved in developing and evaluating the platform in this project and agitation should be present in the PwD. According to studies, almost every PwD develops so-called behavioral and psychological symptoms of dementia in the course of the disease. We have not carried out any tests to determine whether PwD actually have or has had agitation. It was much more important to us to gather the relatives' experiences, which they describe as agitation or restlessness. This would give us a better picture of the reality of life in the home. We, therefore, included all relatives who were able and willing to report on difficult situations in care associated with agitation and restlessness. There were no criteria defined for measuring dementia severity. \u003c/p\u003e\n\u003cp\u003eThe concepts extracted from the literature were used as a framework for analysis. Additional concepts were developed inductively on the data. In addition, 18 qualitative interviews with family caregivers of PwD from the UK MARQUE study [36] have been analyzed using secondary analysis. The MARQUE project consisted of six work streams, the results of which should contribute to a better understanding of agitation and dementia and the workstreams explored how agitation affects the relationship between relatives and PwD and what helps in supporting the PwD. Subsequently, 27 concepts were added as the sub-concepts of \"Interventions\" based on the data, collected from the interviews. For the secondary analysis of these interviews, the same analytical framework was used in the evaluation of the five interviews with family caregivers. 7 additional concepts have been developed inductively from the data. The interviews were conducted during the COVID-19 pandemic, which made the process of recruitment very difficult especially for the vulnerable group of family caregivers. \u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003e2.1.2 Workshops with experts\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eIn an expert workshop with 6 experts in counseling relatives of the PwD, the preliminary ontology has been presented and discussed. About completeness, hints were given on the linguistic adaptation of terms to the target group. Fourteen concepts mainly related to the causes of agitation were also added and prioritized to complete the ontology.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003e2.1.3 Integration of existing ontology models\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eThe eDEM-CONNECT ontology reused concepts from the DRANPTO ontology [21], which focuses on the nonpharmacological management of agitation in dementia. We mapped 65 concepts describing a range of agitated behavior from this ontology.\u003c/p\u003e\n\u003ch2\u003e2.2 Ontology conceptualization\u003c/h2\u003e\n\u003cp\u003eExtracted knowledge in dementia was structured into the conceptual model. At first, the important terms representing the nature of agitated behavior and the causes of being agitated were elicited from the selected knowledge source and coded in the ontology development tool. These elicited terms were defined as nouns and verbs. Here, we identified nouns as concepts and individuals, and then added concepts that include definitions, labels, abbreviations, and synonyms. The class hierarchy was constructed following the subsumption relationship (e.g., is-a-superclass-of). For instance, the terms \"Verbal_agrressive\" and \"Verbal_nonagrressive\" are the sub-concepts of \"nature\" whereas \"nature\" is the sub-concept of the most-hierarchical concept of \"Agitation\". The elicited verbs were defined as object properties for determining inter-relationships between the concepts within the domain and range. For instance, the verb \"exhibits\" was elicited as an object property for defining the relationships between the classes \"Person with dementia\" and \"Agitation\", where the class \"Person with dementia\" is denoted as domain, and the class \"Agitation\" is denoted as the range of the object property.\u003c/p\u003e\n\u003ch2\u003e2.3 Ontology development\u003c/h2\u003e\n\u003cp\u003eGenerally, an ontology consists of concepts, relations, and instances. The concepts are defined as the class of entities within a specific domain, whereas the instances are the concrete situation-specific \"things\" defined by the concepts. A relationship is drawn by the interconnections of the concepts. An ontology and a set of individual instances constitute a knowledge base [37]. The ontology is designed in OWL [27] using the Protégé 5.5.0 software [28] where the OWL is a semantic web language developed to introduce information about the concepts, individuals, and relations between the concepts [28]. In our initial version of the ontology, we defined concepts such as “Person”, “Agitation”, and “Causes” as shown in Figure 2. We used a top-down hierarchical approach, which starts with the most general concepts, e.g., “Person” and “Agitation”, and subsequent sub-concepts, e.g., “Family-Carer” and “Nature”. The EDEM-CONNECTONTO includes types of agitation behaviors, the causes that trigger such behaviors, the consequences that a PwD might experience, and the possible interventions. Here, we also investigated types of family caregivers (e.g., friends, family carers, neighbors) of the PwD and their social activities.\u003c/p\u003e\n\u003ch2\u003e2.4 Ontology evaluation\u003c/h2\u003e\n\u003cp\u003eThe EDEM-CONNECTONTO was assessed in 5 steps. It started with the automatic logical reasoner, which evaluates the consistency of the model. The Protégé software and its reasoner were used to validate the ontology model during the implementing phase. In our case, Pellet [38] and HermiT [39] reasoners were used to identify redundant links in the ontology. Besides, the EDEM-CONNECTONTO was assessed to check whether it could answer the competency question (CQ) using the SPARQL query language. The EDEM-CONNECTONTO was evaluated by the standards of the biomedical ontology in terms of accuracy, clarity, completeness, conciseness, and consistency [40]. At last, two dementia care experts, an experienced professor of the Nursing Department at the University of Witten and an experienced manager of the eDEM-Connect project, were asked to manually assess the EDEM-CONNECTONTO in terms of accuracy, clarity, and completeness. Two annotators were also asked to annotate dementia-related data using an annotation codebook developed from the EDEM-CONNECTONTO. The annotated corpus was employed to assess whether the concepts defined in the EDEM-CONNECTONTO are recognizable by the annotator in dementia-related informal texts. \u003c/p\u003e"},{"header":"3 Results","content":"\u003cp\u003e3.1 \u0026nbsp; \u0026nbsp; The EDEM-CONNECTONTO\u003c/p\u003e\n\u003cp\u003eThe EDEM-CONNECTONTO has 241 concepts, 8 relations (see Table 1), and 240 individuals based on the agitation of PwD, social and physical environment, family members, causes, consequences, health issues of the PwD, and intervention strategies. The analysis of the literature has produced 142 concepts and 4 relationships between concepts. From the interviews with the family caregivers, 27 new concepts and 4 new relationships were added. 7 newly added concepts were also developed from the 18 qualitative interviews with family caregivers of PwD from the UK MARQUE study. The oval shapes in Figure 2 illustrate the concepts; the relationship (object property) is marked as a solid line that connects two concepts, and the subsumption relationship (data property) is denoted as a dotted line. The EDEM-CONNECTONTO is now publicly available at the NCBO BioPortal [41]. The 5 most abstract concepts in the EDEM-CONNECTONTO are \u0026ldquo;Person\u0026rdquo;, \u0026ldquo;Agitation\u0026rdquo;, \u0026ldquo;Causes\u0026rdquo;, \u0026ldquo;Consequences\u0026rdquo; and \u0026ldquo;Interventions\u0026rdquo;. As it is sensible to generate deeper knowledge and further specify the problem with additional diagnostic questions as child concepts, we added concepts in a top-down hierarchical way. For instance, the \u0026quot;Agitation\u0026quot; concept has the sub-concept \u0026quot;nature\u0026quot;, which is then divided into many sub-sub concepts e.g. \u0026quot;verbal_aggressive\u0026quot; and \u0026quot;verbal_nonaggressive\u0026quot;. In Figure 2, we created \u0026quot;exhibits\u0026quot; relations between the concept \u0026quot;Person with dementia\u0026quot; and \u0026quot;Agitation\u0026quot;, marked with a solid line\u0026rsquo;s arrow. The last step was creating individual instances of concepts in the hierarchy.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;As no single \u0026ldquo;correct\u0026rdquo; way or methodology for developing ontologies exists, we debugged and evaluated it with the domain experts. This iterative design process continued through the entire ontology development cycle. \u0026nbsp;The most abstract concept \u0026ldquo;Person\u0026rdquo; is categorized by considering the living conditions, activities, emotions, health conditions, and abilities of the PwD and the non-professional family carer who are responsible for providing long-term support to the PwD, shown in Figure 2 and 3. \u0026ldquo;Family-Carer\u0026rdquo; refers to the people who stay with the PwD or unprofessionally take care of the PwD. Generally, family caregivers and informal caregivers are unpaid family members, colleagues, and relatives who support the person undergoing acute or chronic conditions and want assistance to conduct various tasks, like bathing or dressing. An overview of the sub-subconcepts of the concept \u0026ldquo;Person with dementia\u0026rdquo; is depicted in Figure 3. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 1: Relationships between the concepts and their definitions\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 16.124%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRelationship\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83.876%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDefinition\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 16.124%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eexhibits\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83.876%;\"\u003e\n \u003cp\u003eIndicates relation between \u0026lsquo;PwD\u0026rsquo; and \u0026lsquo;Agitation\u0026rsquo;, \u0026lsquo;Verbal-aggressive\u0026rsquo;, \u0026lsquo;Verbal-nonaggressive\u0026rsquo;, \u0026lsquo;Physical-aggressive\u0026rsquo;, \u0026lsquo;Physical-nonaggressive\u0026rsquo; concepts.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 16.124%;\"\u003e\n \u003cp\u003e\u003cstrong\u003etriggers\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83.876%;\"\u003e\n \u003cp\u003eIndicates the relation between the \u0026lsquo;Causes\u0026rsquo; and \u0026lsquo;Agitation\u0026rsquo; concepts\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 16.124%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eleads\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83.876%;\"\u003e\n \u003cp\u003eIndicates the relation between the \u0026lsquo;Agitation\u0026rsquo; and \u0026lsquo;Consequences\u0026rsquo; concepts\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 16.124%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eaccepts\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83.876%;\"\u003e\n \u003cp\u003eIndicates the relation between the \u0026lsquo;Acceptance\u0026rsquo; and \u0026lsquo;Agitation\u0026rsquo; concepts\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 16.124%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eprevents\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83.876%;\"\u003e\n \u003cp\u003eIndicates the relation between the \u0026lsquo;Prevention\u0026rsquo; and \u0026lsquo;Agitation\u0026rsquo; concepts\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 16.124%;\"\u003e\n \u003cp\u003e\u003cstrong\u003ereduces\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83.876%;\"\u003e\n \u003cp\u003eIndicates the relation between the \u0026lsquo;Reduction of negative consequences\u0026rsquo; and \u0026lsquo;Consequences\u0026rsquo; concepts\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 16.124%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eresolves\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83.876%;\"\u003e\n \u003cp\u003eIndicates the relation between the \u0026lsquo;Problem-solving\u0026rsquo; and \u0026lsquo;Causes\u0026rsquo; concepts\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 16.124%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eis-a\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 83.876%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDenotes that the concept is a sub-concept of another concept\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Agitation can also be defined as any inappropriate physical or verbal, aggressive or non-aggressive act that is not an obvious expression of need or disorientation. It can be defined as emotional distress such as anger, mood swings, negativism, or outbursts. During the challenging behavior analysis of the PwD, four types of agitation-based challenging behaviors were identified. These are \u0026ldquo;Physical_aggressive\u0026rdquo;, \u0026ldquo;Physical_nonaggressive\u0026rdquo;, \u0026ldquo;Verbal_aggressive\u0026rdquo; and \u0026ldquo;Verbal_nonaggressive\u0026rdquo;. Apart from them, the location, atmosphere, and topography of the PwD were added to the ontology (see Figure 4).\u003c/p\u003e\n\u003cp\u003eOur domain experts analyzed the general causes that lead to the above-mentioned challenging behavior for the PwD. Based on the analysis of the general causes of agitated behavior in PwD, we added three top-level sub-concepts under the \u0026ldquo;Causes\u0026rdquo; category: \u0026ldquo;Factors related to the person with dementia\u0026rdquo;, \u0026ldquo;Environmental factors\u0026rdquo;, and \u0026ldquo;Interpersonal factors\u0026rdquo;. In the ontology, one of the most higher-up concepts of \u0026ldquo;Causes\u0026rdquo; is \u0026ldquo;environmental factors\u0026rdquo; which is categorized as the monotony of the nursing home environment, location of the PwD, the color of the room, lack of natural material, dark places, noise (e.g., fan noise, television noise, water noise), temperature, loneliness, activities, weather. For example, if a person lives in an isolated place, it can be responsible for the agitated behavior of the patient. \u0026ldquo;Factors related to the person with dementia\u0026rdquo; and \u0026ldquo;Interpersonal factors\u0026rdquo; are also added as the sub-concepts of the concept \u0026ldquo;Causes\u0026rdquo; that could be the reasons for the challenging behavior of PwD. All the concepts of the EDEM-CONNECTONTO with the definitions are published \u0026nbsp;in BioPortal [41] at https://bioportal.bioontology.org/ontologies/EDEM-CONNECTONTO.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Agitated behavior of the PwD might lead to physical and psychological stress on the caregivers taking care of the PwD, such as heavy workload, disruptions to daily care routines, depression, and poor quality of life ( see [42], [43],[44], [45]). There are a few other adverse outcomes that may include the PwD\u0026rsquo;s premature institutionalization [46]. Our domain experts identified these outcomes that the PwD might experience and added these outcomes as concepts in the EDEM-CONNECTONTO ( see the sub-concepts of \u003cstrong\u003e\u0026quot;Consequences\u0026quot;\u003c/strong\u003e concepts in BioPortal at https://bioportal.bioontology.org/ontologies/EDEM-CONNECTONTO). In general, family-caregivers support the PwD to manage agitation in order to minimize the likelihood of adverse outcomes. Probable interventions are studied that could help the PwD to improve their conditions. Thus, we included interventions based concepts such as \u0026ldquo;Prevention\u0026rdquo;, \u0026ldquo;Problem solving\u0026rdquo;, \u0026ldquo;Acceptance\u0026rdquo;, \u0026ldquo;Reduction of negative consequences\u0026rdquo; (see Figure 5).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.2 \u0026nbsp; \u0026nbsp; Results of evaluating the ontology model by the competency questions using SPARQL queries\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe capability of the EDEM-CONNECTONTO to answer the competency questions (CQ) was evaluated using SPARQL query [47] which is defined as a semantic query language for retrieving and manipulating data stored in the ontology that are described in resource description framework (RDF) format. By using SPARQL query language, we can retrieve data from the EDEM-CONNECTONTO based on the 12 CQ. The retrieved data were checked to see if the EDEM-CONNECTONTO could deliver the correct response for each CQ. To assess the capability of the developed model to answer the CQ, each CQ was represented by SPARQL queries to retrieve data from the ontology. For instance, CQ 1: \u003cem\u003e\u0026ldquo;What are the types of the agitated behaviors, described for the PwD?\u0026rdquo;\u003c/em\u003e in SPARQL language was\u003c/p\u003e\n\u003cp skip=\"true\"\u003e\u003cstrong\u003e\u003cem\u003eprefix rdf:\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u0026lt;http://www.w3.org/1999/02/22-rdf-syntax-ns#\u0026gt;\u003c/p\u003e\n\u003cp skip=\"true\"\u003e\u003cstrong\u003e\u003cem\u003eprefix rdfs:\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u0026lt;http://www.w3.org/2000/01/rdf-schema#\u0026gt;\u003cstrong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp skip=\"true\"\u003e\u003cstrong\u003e\u003cem\u003eprefix owl:\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u0026lt;http://www.w3.org/2002/07/owl#\u0026gt;\u003c/p\u003e\n\u003cp skip=\"true\"\u003e\u003cstrong\u003e\u003cem\u003eSELECT ?subclass ?subclassLabel ?subclassDefinition\u0026nbsp;\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp skip=\"true\"\u003e\u003cstrong\u003e\u003cem\u003eWHERE {\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp skip=\"true\"\u003e\u003cstrong\u003e\u003cem\u003e?subclass rdfs:subClassOf myont:Nature .\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp skip=\"true\"\u003e\u003cstrong\u003e\u003cem\u003e?subclass rdfs:label ?subclassLabel .\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp skip=\"true\"\u003e\u003cstrong\u003e\u003cem\u003eOPTIONAL { ?subclass rdfs:comment ?subclassDefinition }\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp skip=\"true\"\u003e\u003cstrong\u003e\u003cem\u003e?individual rdf:type myont:Nature .\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp skip=\"true\"\u003e\u003cstrong\u003e\u003cem\u003e}\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp skip=\"true\"\u003eWe can retrieve information on the types of agitated behavior along with the definitions from CQ 1. The retrieved data are the 5 types of agitated behavior. The retrieved outputs are the name of the concepts (\u0026ldquo;Strong_emotion\u0026rdquo;, \u0026ldquo;Verbal_aggressive\u0026rdquo;, \u0026ldquo;Verbal_nonaggressive\u0026rdquo;, \u0026ldquo;Physical_aggressive\u0026rdquo;, \u0026ldquo;Physical_nonaggressive\u0026rdquo;) along with the \u0026ldquo;rdfs:label,\u0026rdquo; \u0026ldquo;rdfs:comment,\u0026rdquo; and \u0026ldquo;definitions\u0026rdquo;. The 12 CQ, along with the retrieved output and definition of one of the concepts, are presented in Table 2. We calculated accuracy as the primary evaluation metric to measure the overall correctness of the model. This involved calculating the ratio of the number of accurate predictions to the absolute number of predictions. We specifically evaluated the model\u0026rsquo;s accuracy with the 12 CQ (see Section 3.3). The consistency between the retrieved data and the actual developed concepts was a key indicator that EDEM-CONNECTONTO can deliver the correct response for each query. As a result, we achieved 100% accuracy for each query.\u003c/p\u003e\n\u003cp\u003eTable 2: We retrieved the output from the ontology model based on the competency question (CQ). The consistency between the retrieved outcomes and the developed concepts indicates that the EDEM-CONNECTONTO provides the correct response for each CQ.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCQ\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRetrieved output\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDefinition of the concepts\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWhat are the types of agitated behaviors described for the person with dementa?\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u0026lsquo;Strong_emotion\u0026rsquo;, \u0026lsquo;Verbal_aggressive\u0026rsquo;,\u003c/p\u003e\n \u003cp\u003e\u0026lsquo;Verbal_nonaggressive\u0026rsquo;,\u003c/p\u003e\n \u003cp\u003e\u0026lsquo;Physical_aggressive\u0026rsquo;, and \u0026lsquo;Physical_nonaggressive\u0026rsquo;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eDefinition of \u0026lsquo;Physical_aggressive\u0026rsquo;:\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026lsquo;It refers to the physically-aggressiveness behaviors of the people with dementia. For example: biting, kicking, hurting and others\u0026rsquo; (see [35]).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWhat are the causal factors of agitation in person with dementia?\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u0026lsquo;Environmental factors\u0026rsquo;, \u0026lsquo;Interpersonal factors\u0026rsquo;, and \u0026lsquo;Factors related to the person with dementia\u0026rsquo;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eDefinition of \u0026lsquo;Environmental factors\u0026rsquo;: \u0026lsquo;It refers to the environmental factors that can cause agitation\u0026rsquo; (see [35]).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWhat are the causal environmental factors of agitation in person with dementia?\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u0026lsquo;Lack of activity\u0026rsquo;, \u0026lsquo;Lack of established routines\u0026rsquo;, \u0026lsquo;Safety issues\u0026rsquo;, \u0026lsquo;Change in routines\u0026rsquo;, \u0026lsquo;Light level\u0026rsquo;, and \u0026lsquo;Noise level\u0026rsquo;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eDefinition of \u0026lsquo;Lack of activity\u0026rsquo;: \u0026lsquo;It refers to lack of appropriate engaging activities for a people with dementia that can cause agitation\u0026rsquo; (see [35]).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWhat are verbal_aggressive behaviours?\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u0026lsquo;Screaming\u0026rsquo;, \u0026lsquo;Cursing\u0026rsquo;, \u0026lsquo;Safety issues\u0026rsquo;, and \u0026lsquo;Making verbal sexual advances\u0026rsquo;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eDefinition of \u0026lsquo;Screaming\u0026rsquo;: \u0026lsquo;It refers to speaking loudly / shouting, high-pitched vocalisations\u0026rsquo; (see [35]).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWhat are physical_nonaggressive behaviours ?\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u0026lsquo;Aimless wandering\u0026rsquo;, \u0026lsquo;Pacing\u0026rsquo;, \u0026lsquo;Hyper activity\u0026rsquo;, \u0026lsquo;Hiding things\u0026rsquo;, \u0026lsquo;Intentional falling\u0026rsquo;, \u0026lsquo;Handling things inappropriately\u0026rsquo;, \u0026lsquo;Eating inappropriate substances\u0026rsquo;and \u0026lsquo;Performing repetitive mannerism\u0026rsquo;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eDefinition of \u0026lsquo;Aimless wandering\u0026rsquo;: \u0026lsquo;It refers to a certain type of behaviour that includes walking aimlessly from place to place\u0026rsquo; (see [35]).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWhat is the background information of person with dementia?\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u0026lsquo;Mood and emotion\u0026rsquo;, \u0026lsquo;Personality\u0026rsquo;, \u0026lsquo;Lifestyle before dementia\u0026rsquo;, \u0026lsquo;Age of the person with dementia\u0026rsquo;, \u0026lsquo;Preferences\u0026rsquo;, and \u0026lsquo;Resources\u0026rsquo;.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eDefinition of \u0026lsquo;Resources\u0026rsquo;: \u0026nbsp;\u0026lsquo;It refers to the resources for people with dementia to assist individuals in managing their condition and enhancing their quality of life (e.g., support systems, services, and tools available to assist the people with dementia)\u0026rsquo; (see [35]).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWho are the informal caregivers of person with dementia?\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u0026lsquo;Family_Carer\u0026rsquo;, \u0026lsquo;Neighbours\u0026rsquo;, \u0026lsquo;Friends\u0026rsquo;, and \u0026lsquo;Further_family_members\u0026rsquo;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eDefinition of \u0026lsquo;Family_Carer\u0026rsquo; : \u0026lsquo;It refers to the informal caregiver who takes care of the person suspected/diagnosed with dementia\u0026rsquo; (see [35]).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWhat are the consequences for person with dementia experiencing agitation due to dementia?\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u0026lsquo;Disability\u0026rsquo;, \u0026lsquo;Institutionalisation early placement in a nursing home\u0026rsquo;, \u0026lsquo;Decreased quality of life\u0026rsquo;, \u0026lsquo;Anorexia weight loss\u0026rsquo;, \u0026lsquo;Hospital admission\u0026rsquo; and \u0026lsquo;Poor caregiver outcomes\u0026rsquo;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eDefinition of \u0026lsquo;Decreased quality of life\u0026rsquo;: \u0026lsquo;It refers to the concept that the agitation has an effect on the individual\u0026apos;s wellbeing and the ability to enjoy life\u0026rsquo; (see [35]).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWhat non-pharmacological interventions are employed to mitigate the effect of agitation in dementia?\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u0026lsquo;Maintain routines\u0026rsquo;, \u0026lsquo;provide for social contacts\u0026rsquo;, \u0026lsquo;Promote well-being\u0026rsquo;, \u0026lsquo;Improve the quality of relationship between the PWD and the relative\u0026rsquo;, and \u0026lsquo;Identify unmet needs and meet them when possibles\u0026rsquo;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eDefinition of \u0026lsquo;Maintain routines\u0026rsquo;: \u0026lsquo;It refers to the action of the family carer that maintains the care routine of a people with dementia according to his/her needs\u0026rsquo; (see [35]).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWhat intervention can be used for verbal_aggressive behaviour of the person with dementia?\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u0026lsquo;Distract\u0026rsquo;, \u0026lsquo;Screening for physical/medical causes\u0026rsquo;, \u0026lsquo;Environmental design\u0026rsquo;, \u0026lsquo;Recognize and avoid triggers\u0026rsquo; and \u0026nbsp;\u0026lsquo;Screening for physical/medical causes\u0026rsquo;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eDefinition of \u0026lsquo;Distract\u0026rsquo;: \u0026lsquo;It refers to the intervention taken by the family carer which is to distract the person with dementia from problematic situations to another task, such as, to offer a snack or put on some familiar music to interrupt behaviours that are becoming difficult\u0026rsquo; (see [35]).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWhat are the interpersonal factors that are responsible for verbally non-agressive behavior?\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u0026lsquo;Lack of caring skills\u0026rsquo;, \u0026lsquo;Language of the caregiver\u0026rsquo;, \u0026lsquo;verbal interaction\u0026rsquo; and \u0026lsquo;Low mental capabilities\u0026rsquo;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eDefinition of \u0026lsquo;Lack of caring skills\u0026rsquo;: \u0026lsquo;It refers to the lack of caring skills of the caregivers to deal with person with dementia\u0026rsquo; (see [35]).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWhat intervention can be used to cope with the screaming of the PwD?\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u0026lsquo;Communication: controlling the emotional tone of your own language\u0026rsquo; and \u0026lsquo;Distract\u0026rsquo;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eDefinition of \u0026lsquo;Distract\u0026rsquo;: \u0026lsquo;It refers to the intervention taken by the family carer which is to distract the person with dementia from problematic situations to another task, such as, to offer a snack or put on some familiar music to interrupt behaviors that are becoming difficult (see [35]).\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;3.3 \u0026nbsp; \u0026nbsp; Results of evaluating EDEM-CONNECTONTO against biomedical ontology criteria\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEDEM-CONNECTONTO was assessed with respect to consistency, clarity, completeness, and accuracy following the biomedical ontology standards [40].\u003c/p\u003e\n\u003cp\u003e\u0026bull; \u0026nbsp;Accuracy: We checked the definitions, descriptions, and properties of the concepts. They meet the biomedical ontology standard (see [40]) as the concepts and their definitions were extracted from the scientific papers and assessed by the domain experts.\u003c/p\u003e\n\u003cp\u003e\u0026bull; \u0026nbsp;Completeness: It measure the ontology\u0026apos;s domain knowledge coverage. Two domain experts and an experienced manager of the eDEM-CONNECT project evaluated it manually. The 12 CQs were also used to assess the completeness of the EDEM-CONNECTONTO.\u003c/p\u003e\n\u003cp\u003e\u0026bull; \u0026nbsp;Consistency: It means that the ontology does not keep or permit any contradictions. \u0026nbsp;We used Pellet and Hermit Reasoners, and the results indicate that the ontology model is consistent.\u003c/p\u003e\n\u003cp\u003e\u0026bull; \u0026nbsp;Clarity: It ensures that the EDEM-CONNECTONTO is unambiguous, with well-defined\u003c/p\u003e\n\u003cp\u003emeanings for all concepts and relationships. Names of concepts and their definitions should b\u003cstrong\u003ee\u0026nbsp;\u003c/strong\u003ecomprehensible. Clarity of this ontology is accomplished by assigning a specific label with a relevant explanation to the concepts individually using \u0026ldquo;rdfs: label,\u0026rdquo; \u0026ldquo;rdfs:comment,\u0026rdquo; and \u0026ldquo;definitions\u0026rdquo; (see an illustration about the annotation of the \u0026ldquo;Person with dementia\u0026rdquo; in Figure 6). It also ensures that the ontology can communicate properly to the readers with a relevant explanation of the concepts and the relationships between these concepts.\u003c/p\u003e\n\u003cp\u003e3.4 \u0026nbsp; \u0026nbsp;Evaluation by domain experts\u003c/p\u003e\n\u003cp\u003eOur domain experts reviewed and assessed the ontology model manually for measuring accuracy and completeness. Their revision on EDEM-CONNECTONTO is viewed as a genuine assessment. Based on their suggestions, we added four new concepts as the ontology\u0026apos;s sub-subconcepts of the \u0026quot;Interventions\u0026quot; concept. They were: \u0026ldquo;Self-care: taking time for yourself\u0026rdquo;, \u0026ldquo;Looking for distractions\u0026rdquo;, \u0026ldquo;Safety for PWD: provide a safe environment\u0026rdquo; and \u0026ldquo;Safety for relatives: leave the room, leave PWD alone\u0026rdquo; as a subclass of \u0026ldquo;Reduction of negative consequences\u0026rdquo;. In addition, following the domain expert\u0026rsquo;s recommendation about redundancy, the term \u0026ldquo;Talking about feelings and experiences of the PwD\u0026rdquo; was removed from the subclass of \u0026ldquo;Reduction of negative consequences\u0026rdquo;. The concepts \u0026ldquo;distress\u0026rdquo;, \u0026ldquo;reduced income from employment\u0026rdquo; and \u0026ldquo;depression\u0026rdquo; were also removed from the subconcepts of the \u0026ldquo;Consequences\u0026rdquo; concept. The domain experts and the project manager then evaluated the modified ontology. They verified its accuracy and completeness without additional revision.\u003c/p\u003e\n\u003cp\u003e3.5 \u0026nbsp; \u0026nbsp;Data annotation\u003c/p\u003e\n\u003cp\u003eWe proposed an annotation scheme that is based on the 8 most general concepts and relations from the EDEM-CONNECTONTO for annotating textual data from online dementia forums. \u0026quot;Agitation\u0026quot;, \u0026quot;Verbal_aggressive\u0026quot;, \u0026quot;Verbal_nonaggressive\u0026quot;, \u0026quot;Physical_aggressive\u0026quot;, \u0026quot;Physical_nonaggressive\u0026quot;, \u0026quot;Cause\u0026quot;, \u0026quot;PwD\u0026quot;, \u0026quot;Family-carer\u0026quot; are the labels used for the annotation task (see details in [15]). Two annotators were employed to label instances independently of the entities (concepts) and the relationships in the text corpus. We used Cohen\u0026rsquo;s kappa score [48] as a measure of interrater reliability to evaluate the quality of the annotated data. We observed that the annotators mainly annotated entities containing noun phrases (e.g., father, he) based on \u0026ldquo;PwD\u0026rdquo; and \u0026ldquo;Family-carer\u0026rdquo; labels compared to the labels describing types of agitation of PwD (e.g., verbal_nonaggressive, physical_aggressive) (see [15]). We found that the domain of PwD is difficult to conceptualize. Our domain experts and the annotators struggled with identifying the concepts and relationships that define the domain. Nonetheless, the outcomes demonstrated that both annotators could observe the labels from the concepts in EDEM-CONNECTONTO with moderate agreement.\u003c/p\u003e"},{"header":"4 Discussion","content":"\u003cp\u003eThe eDEM-CONNECT project has succeeded in developing a comprehensive ontology for managing the agitation of PwD. This ontology consists of 241 concepts that can be helpful in various aspects of the care of PwD. It is incorporated as a knowledge base for the eDEM-Connect chatbot (see [13]) where SPARQL query is used to search associated questions in the ontology and is forwarded to the user as the response. The eDEM-CONNECT chatbot dedicated to support the family caregivers of \u0026nbsp;PwD on agitation topics, empowering them by providing structured knowledge which can ease their daily difficulties with their relatives with dementia. Both domain experts and family caregivers have been involved from the beginning of the ontology development phase. This has resulted in scientific findings and practical knowledge being incorporated into the ontology. A unique characteristic of this ontology is its focus on family caregivers. This is also reflected in the fact that aspects of the dyadic relationship have been elaborated in detail, both in terms of the possible causes of agitation and especially in terms of interventions. This distinguishes the ontology from comparable ontologies, such as the DRANPTO [21] ontology. The first challenge was identifying the relevant resources to extract knowledge in the dementia domain. We focused on the types of agitation, non-pharmacological interventions, family caregivers, and the socio-economic condition of the PwD, as no standardized document supplies vast coverage of evidence-based knowledge. We observed that it was challenging for the domain experts to define the concepts after extracting relevant information. The second challenge was looking for an efficient ontology assessment approach for the EDEM-CONNECTONTO because of the need for standardized methods to assess ontologies. A few ontology assessment approaches have been developed to determine the quality of ontology models, such as the gold standard assessment, domain expert-based assessment, tool-based assessment approach, CQ-based assessment, and quality criteria-based assessment [40]. Generally, ontology developers use any one of the assessment approaches or a mixture of these assessment methods to measure the quality of ontology models (see [40], [49]). Therefore, we evaluated the EDEM-CONNECTONTO using five approaches that include 1) evaluation by automatic reasoner 2) answering SPARQL queries 3) applying ontology quality criteria e.g., accuracy, clarity 4) evaluating the ontology by domain experts, and 5) using concepts from the ontology for an annotation task. This combined assessment approach has generated excellent results. The EDEM-CONNECTONTO satisfies the quality standards defined in [40]. Another challenging aspect was engaging domain experts to assess the EDEM-CONNECTONTO manually. In the domain expert–based assessment method, domain experts need to manually check all entities (concepts and relationships) of the suggested ontology. As this method requires lots of effort and time from domain experts, it is considered a costly assessment approach. Two domain experts and the eDEM-CONNECT project manager manually assessed the EDEM-CONNECTONTO concerning accuracy, clarity, and completeness. We also conducted 2 workshops where the concepts and the relationships were discussed to ensure the clarity of the EDEM-CONNECTONTO. The EDEM-CONNECTONTO has many potentials. One of the prospects is that we can convert dementia-specific textual data into machine-processable data by producing semantic annotations that directly map a data element to an ontology concept. Another potential of the EDEM-CONNECTONTO is to be employed as an annotation codebook for data annotation tasks where the annotated corpus could be used for concept and relationship extraction tasks in the domain of PwD and the annotated corpus will generate more insights about agitation in dementia. This, in turn, will enable the training of machine learning models that could support PwD and their caregivers in challenging situations and potentially improve their overall wellbeing.\u003c/p\u003e\n\u003cp\u003eFrom the beginning, we have involved both domain experts and the family caregivers of the PwD in the ontology’s development phase. However, the participation of family caregivers in fundamental decisions on the selection of knowledge for ontology could not be learned due to restrictions during the COVID-19 pandemic. The EDEM-CONNECTONTO is limited to the non-clinical intervention strategies for managing agitation in dementia as we have a considerable volume of clinical data that are publicly available for addressing agitation in dementia.\u003c/p\u003e\n\u003cp\u003eThe EDEM-CONNECTONTO has a particular value for nursing science. This is highly relevant because one of the existing ontologies concerning nursing, the International Classification for Nursing Practice (ICNP) Ontology, does not contain any concepts that are important for dealing with the agitation of PwD. The structured exploration of the ontology concepts clearly showed weak evidence on the individual concepts, especially on the weighting and causal relationships between the concepts. Most of the literature originates from the field of medicine. Structured nursing knowledge of the concepts analyzed needs to be improved. In particular, there is a lack of evidence on interventions that are effective for family caregivers in specific situations. In the future, we plan to conduct workshops with the family caregivers of the PwD to define more concepts and relationship entities on the intervention strategies. We plan to annotate more textual unstructured data for the concepts and relationships extraction task using a machine learning approach, which will be integrated into the EDEM-CONNECTONTO. Furthermore, EDEM-CONNECTONTO will also be extended by incorporating information from additional sources (interviews and group discussions with relatives as well as publicly available sources such as online forums). As mentioned before, the vision we have is a chatbot that relies on the domain knowledge from the ontology and through interaction with the user it could provide him or her with helpful information regarding a given agitation-related situation with a PwD. During the project we have implemented a minimal probabilistic chatbot [13] that relies on parts of the ontology. In the future we plan to use modern LLM architectures and combine them with our ontology either in the form of RAG, through fine-tuning or even by incorporating an additional logical layer on top of the LLM to ensure the validity of the provided information.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eAcknowledgements\u003c/p\u003e\n\u003cp\u003eWe recognize the efforts of our annotators in annotating the dementia forum texts.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFundings\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the German Federal Ministry of Education and Research (BMBF, reference number: 16SV8335)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of Interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors have no conflict of interest to report.\u003c/p\u003e\n\u003cp\u003eData availability\u003c/p\u003e\n\u003cp\u003eThe EDEM-CONNECTONTO ontology is publicly available at the NCBO BioPortal at https://bioportal.bioontolog\u003cu\u003ey.\u0026nbsp;\u003c/u\u003eorg/ontologies/EDEM-CONNECTONTO\u003ca href=\"https://bioportal.bioontology.org/ontologies/EDEM-CONNECTONTO\"\u003e.\u003c/a\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAdditional information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003eAUTHOR CONTRIBUTIONS STATEMENT\u003c/p\u003e\n\u003cp\u003eS.S., C.P., I.H., and K.Y. conceived the idea. All authors discussed the results and commented on the paper. C.P., I.H., M.H., and B.H. conceptualized all the concepts of the ontology and S.S. designed, developed and evaluated the EDEM-CONNECTONTO ontology. \u0026nbsp;All authors reviewed the definition of the ontology and the S.S., C.P., I.H., and K.Y. wrote the paper.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eDementia (2023) World Health Organization. Available at: https://www.who.int/news-room/fact-sheets/detail/dementia (Accessed: 11 June 2024).\u003c/li\u003e\n\u003cli\u003eThyrian JR, Eichler T, Hertel J, Wucherer D, Dreier A, Michalowsky B, Killimann I, Teipel S, Hoffmann W. Burden of behavioral and psychiatric symptoms in people screened positive for dementia in primary care: results of the DelpHi-study. Journal of Alzheimer\u0026apos;s Disease. 2015 Jan 1;46(2):451-9.\u003c/li\u003e\n\u003cli\u003eCummings J, Mintzer J, Brodaty H, Sano M, Banerjee S, Devanand DP, Gauthier S, Howard R, Lanct\u0026ocirc;t K, Lyketsos CG, Peskind E. Agitation in cognitive disorders: International Psychogeriatric Association provisional consensus clinical and research definition. International psychogeriatrics. 2015 Jan;27(1):7-17.\u003c/li\u003e\n\u003cli\u003eBrijoux T, Zank S. Auswirkungen kognitiver Einschr\u0026auml;nkungen (Demenz) auf Lebensqualit\u0026auml;t und Versorgung. InHohes Alter in Deutschland 2023 Jul 5 (pp. 173-195). Berlin, Heidelberg: Springer Berlin Heidelberg.\u003c/li\u003e\n\u003cli\u003eYaffe K, Fox P, Newcomer R, Sands L, Lindquist K, Dane K, Covinsky KE. Patient and caregiver characteristics and nursing home placement in patients with dementia. Jama. 2002 Apr 24;287(16):2090-7.\u003c/li\u003e\n\u003cli\u003eGeorge C, Ferreira N, Evans R, Honeyman V. A systematic review of the association between individual behavioural and psychological symptoms in dementia and carer burden. Working with Older People. 2020 Sep 18;24(3):181-203.\u003c/li\u003e\n\u003cli\u003eDe Vugt ME, Stevens F, Aalten P, Lousberg R, Jaspers N, Winkens I, Jolles J, Verhey FR. Behavioural disturbances in dementia patients and quality of the marital relationship. International journal of geriatric psychiatry. 2003 Feb;18(2):149-54.\u003c/li\u003e\n\u003cli\u003ePozzebon M, Douglas J, Ames D. Spouses\u0026rsquo; experience of living with a partner diagnosed with a dementia: A synthesis of the qualitative research. International psychogeriatrics. 2016 Apr;28(4):537-56.\u003c/li\u003e\n\u003cli\u003eLuiu AL, Favez N, Betrancourt M, Szilas N, Ehrler F. Family relationships and alzheimer\u0026rsquo;s disease: A systematic review. Journal of Alzheimer\u0026apos;s Disease. 2020 Jan 1;76(4):1595-608.\u003c/li\u003e\n\u003cli\u003eStephan A, Bieber A, Hopper L, Joyce R, Irving K, Zanetti O, Portolani E, Kerpershoek L, Verhey F, de Vugt M, Wolfs C. Barriers and facilitators to the access to and use of formal dementia care: findings of a focus group study with people with dementia, informal carers and health and social care professionals in eight European countries. BMC geriatrics. 2018 Dec;18:1-6.\u003c/li\u003e\n\u003cli\u003eVecchio N, Fitzgerald JA, Radford K, Fisher R. The association between cognitive impairment and community service use patterns in older people living in A ustralia. Health \u0026amp; social care in the community. 2016 May;24(3):321-33.\u003c/li\u003e\n\u003cli\u003eMichalowsky B, Kaczynski A, Hoffmann W. \u0026Ouml;konomische und gesellschaftliche Herausforderungen der Demenz in Deutschland\u0026ndash;Eine Metaanalyse. Bundesgesundheitsblatt-Gesundheitsforschung-Gesundheitsschutz. 2019 Jul 11;62(8):981-92. \u003c/li\u003e\n\u003cli\u003eBoiting M, Tschorn N, Suravee S, Yordanova K, Halek M, Jagoda FA, L\u0026uuml;dtke S, Burmann A. eDEM-CONNECT: An Ontology-Based Chatbot for Family Caregivers of People with Dementia. InInternational Conference on Wireless Mobile Communication and Healthcare 2023 Nov 29 (pp. 73-91). Cham: Springer Nature Switzerland.\u003c/li\u003e\n\u003cli\u003eDe Angelis L, Baglivo F, Arzilli G, Privitera GP, Ferragina P, Tozzi AE, Rizzo C. ChatGPT and the rise of large language models: the new AI-driven infodemic threat in public health. Frontiers in Public Health. 2023 Apr 25;11:1166120.\u003c/li\u003e\n\u003cli\u003eSuravee S, Stoev T, Schindler D, Hochgraeber I, Pinkert C, Holle B, Halek M, Kr\u0026uuml;ger F, Yordanova K. Annotation scheme for named entity recognition and relation extraction tasks in the domain of people with dementia. In2022 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops) 2022 Mar 21 (pp. 236-241). IEEE.\u003c/li\u003e\n\u003cli\u003eLewis P, Perez E, Piktus A, Petroni F, Karpukhin V, Goyal N, K\u0026uuml;ttler H, Lewis M, Yih WT, Rockt\u0026auml;schel T, Riedel S. Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in neural information processing systems. 2020;33:9459-74.\u003c/li\u003e\n\u003cli\u003eGruber TR. A translation approach to portable ontology specifications. Knowledge acquisition. 1993 Jun 1;5(2):199-220.\u003c/li\u003e\n\u003cli\u003eRoantree M, O\u0026rsquo;Donoghue J, O\u0026rsquo;Kelly N, Pierce M, Irving K, Van Boxtel M, K\u0026ouml;hler S. Mapping longitudinal studies to risk factors in an ontology for dementia. Health Informatics Journal. 2016 Jun;22(2):414-26.\u003c/li\u003e\n\u003cli\u003eMeditskos G. Multimodal sensing and fusion for comprehensive monitoring and feedback: the integrated Dem@ Care system.2015.\u003c/li\u003e\n\u003cli\u003eFook VF, Tay SC, Jayachandran M, Biswas J, Zhang D. An ontology-based context model in monitoring and handling agitation behavior for persons with dementia. InFourth Annual IEEE International Conference on Pervasive Computing and Communications Workshops (PERCOMW\u0026apos;06) 2006 Mar 13 (pp. 5-pp). IEEE.\u003c/li\u003e\n\u003cli\u003eZhang Z, Yu P, Chang HC, Lau SK, Tao C, Wang N, Yin M, Deng C. Developing an ontology for representing the domain knowledge specific to non‐pharmacological treatment for agitation in dementia. Alzheimer\u0026apos;s \u0026amp; Dementia: Translational Research \u0026amp; Clinical Interventions. 2020;6(1):e12061.\u003c/li\u003e\n\u003cli\u003eMalhotra A, Younesi E, G\u0026uuml;ndel M, M\u0026uuml;ller B, Heneka MT, Hofmann-Apitius M. ADO: A disease ontology representing the domain knowledge specific to Alzheimer\u0026apos;s disease. Alzheimer\u0026apos;s \u0026amp; dementia. 2014 Mar 1;10(2):238-46.\u003c/li\u003e\n\u003cli\u003eSkillen KL, Chen L, Nugent CD, Donnelly MP, Solheim I. A user profile ontology based approach for assisting people with dementia in mobile environments. In2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society 2012 Aug 28 (pp. 6390-6393). IEEE.\u003c/li\u003e\n\u003cli\u003eZhang Z, Yu P, Yin M, Chang HC, Thomas SJ, Wei W, Song T, Deng C. Developing an ontology of non-pharmacological treatment for emotional and mood disturbances in dementia. Scientific Reports. 2024 Jan 22;14(1):1937.\u003c/li\u003e\n\u003cli\u003eYordanova K, Koldrack P, Heine C, Henkel R, Martin M, Teipel S, Kirste T. Situation model for situation-aware assistance of dementia patients in outdoor mobility. Journal of Alzheimer\u0026apos;s Disease. 2017 Jan 1;60(4):1461-76.\u003c/li\u003e\n\u003cli\u003eSimperl EP, Tempich C. Ontology engineering: A reality check. InOn the Move to Meaningful Internet Systems 2006: CoopIS, DOA, GADA, and ODBASE: OTM Confederated International Conferences, CoopIS, DOA, GADA, and ODBASE 2006, Montpellier, France, October 29-November 3, 2006. Proceedings, Part I 2006 (pp. 836-854). Springer Berlin Heidelberg.\u003c/li\u003e\n\u003cli\u003eAntoniou G, Harmelen FV. Web ontology language: Owl. Handbook on ontologies. 2009:91-110.\u003c/li\u003e\n\u003cli\u003eMusen MA. The prot\u0026eacute;g\u0026eacute; project: a look back and a look forward. AI matters. 2015 Jun 16;1(4):4-12.\u003c/li\u003e\n\u003cli\u003eKales HC, Gitlin LN, Lyketsos CG. Assessment and management of behavioral and psychological symptoms of dementia. Bmj. 2015 Mar 2;350.\u003c/li\u003e\n\u003cli\u003eBartholomeyczik S, Halek M, editors. Assessmentinstrumente in der Pflege: M\u0026ouml;glichkeiten und Grenzen. Schl\u0026uuml;tersche; 2010 Dec 6.\u003c/li\u003e\n\u003cli\u003eKong EH. Agitation in dementia: concept clarification. Journal of advanced nursing. 2005 Dec;52(5):526-36.\u003c/li\u003e\n\u003cli\u003eK\u0026ouml;hler K, Dreyer J, Hochgraeber I, von Kutzleben M, Pinkert C, Roes M, Holle B. Towards a middle-range theory of \u0026lsquo;Stability of home-based care arrangements for people living with dementia\u0026rsquo;(SoCA-Dem): findings from a meta-study on mixed research. BMJ open. 2021 Apr 1;11(4):e042515.\u003c/li\u003e\n\u003cli\u003eThomas J, Harden A. Methods for the thematic synthesis of qualitative research in systematic reviews. BMC medical research methodology. 2008 Dec;8:1-0.\u003c/li\u003e\n\u003cli\u003eWard-Griffin C, Oudshoorn A, Clark K, Bol N. Mother-adult daughter relationships within dementia care: A critical analysis. Journal of Family Nursing. 2007 Feb;13(1):13-32.\u003c/li\u003e\n\u003cli\u003eCohen-Mansfield J. Cohen-Mansfield Agitation Inventory. International Journal of Geriatric Psychiatry. 2012.\u003c/li\u003e\n\u003cli\u003eLaybourne, Anne (2019), \u0026ldquo;Managing agitation and raising quality of life: semi structured interviews with family carers of people living with dementia\u0026rdquo;, Mendeley Data, V1, doi: 10.17632/s8wtptnhyc.1\u003c/li\u003e\n\u003cli\u003eNoy NF, McGuinness DL. Ontology development 101: A guide to creating your first ontology.\u003c/li\u003e\n\u003cli\u003eSirin E, Parsia B, Grau BC, Kalyanpur A, Katz Y. Pellet: A practical owl-dl reasoner. Journal of Web Semantics. 2007 Jun 1;5(2):51-3.\u003c/li\u003e\n\u003cli\u003eMotik B, Shearer R, Horrocks I. Optimized reasoning in description logics using hypertableaux. InAutomated Deduction\u0026ndash;CADE-21: 21st International Conference on Automated Deduction Bremen, Germany, July 17-20, 2007 Proceedings 21 2007 (pp. 67-83). Springer Berlin Heidelberg.\u003c/li\u003e\n\u003cli\u003eAmith M, He Z, Bian J, Lossio-Ventura JA, Tao C. Assessing the practice of biomedical ontology evaluation: Gaps and opportunities. Journal of biomedical informatics. 2018 Apr 1;80:1-3.\u003c/li\u003e\n\u003cli\u003eGeller J, Keloth VK, Musen MA. How sustainable are biomedical ontologies?. InAMIA Annual Symposium proceedings 2018 (Vol. 2018, p. 470). American Medical Informatics Association.\u003c/li\u003e\n\u003cli\u003eCerejeira J, Lagarto L. Behavioral and psychological symptoms of dementia. Frontiers in neurology. 2012 May 7;3:23573.\u003c/li\u003e\n\u003cli\u003eCubit K, Farrell G, Robinson A, Myhill M. A survey of the frequency and impact of behaviours of concern in dementia on residential aged care staff. Australasian Journal on Ageing. 2007 Jun;26(2):64-70.\u003c/li\u003e\n\u003cli\u003eFeast A, Orrell M, Charlesworth G, Melunsky N, Poland F, Moniz-Cook E. Behavioural and psychological symptoms in dementia and the challenges for family carers: systematic review. The British Journal of Psychiatry. 2016 May;208(5):429-34.\u003c/li\u003e\n\u003cli\u003eP\u0026eacute;rez J, Arenas M, Gutierrez C. Semantics and complexity of SPARQL. ACM Transactions on Database Systems (TODS). 2009 Sep 3;34(3):1-45.\u003c/li\u003e\n\u003cli\u003eKales HC, Gitlin LN, Lyketsos CG. Assessment and management of behavioral and psychological symptoms of dementia. Bmj. 2015 Mar 2;350.\u003c/li\u003e\n\u003cli\u003eP\u0026eacute;rez J, Arenas M, Gutierrez C. Semantics and complexity of SPARQL. ACM Transactions on Database Systems (TODS). 2009 Sep 3;34(3):1-45.\u003c/li\u003e\n\u003cli\u003eKv\u0026aring;lseth TO. Note on Cohen\u0026apos;s kappa. Psychological reports. 1989 Aug;65(1):223-6.\u003c/li\u003e\n\u003cli\u003eRaad J, Cruz C. A survey on ontology evaluation methods. In Proceedings of the International Conference on Knowledge Engineering and Ontology Development, part of the 7th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management 2015 Nov 12.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Dementia, agitation, ontology, non-clinical interventions","lastPublishedDoi":"10.21203/rs.3.rs-6289849/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6289849/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003ePersons with dementia (PwD) face cognitive decline, placing added stress on family caregivers. Challenging behaviour like agitation is one of the prominent behaviors exhibited in PwD. We envision a chatbot that provides information to unprofessional caregivers to ease the stress factor while dealing with agitation in PwD. To develop the chatbot, one needs domain knowledge, including types of agitated behavior, living and socio-economic conditions of the PwD. We call this structured knowledge an ontology. This study focuses on developing the eDEM-Connect Ontology: Ontology of Dementia-related Agitation and Relationship between Informal Caregivers and Persons with Dementia (EDEM-CONNECTONTO) as the domain knowledge that chatbot needs to use for providing adequate information. We perform a systematic literature review, analyze existing ontologies, hold workshops with experts, and interview informal caregivers. We then develop and validate the EDEM-CONNECTONTO with the Prot\u0026eacute;g\u0026eacute; software. EDEM-CONNECTONTO consists of 241 Concepts, 8 relations, and 240 individuals. The results from the evaluation show that it meets the standard for biomedical ontologies. The EDEM-CONNECTONTO addresses agitation in PwD, prioritizing support for family caregivers and incorporating non-pharmacological interventions. It fills gaps in nursing science by formalizing knowledge relevant to dementia care and agitation, guiding future research in the field.\u003c/p\u003e","manuscriptTitle":"eDEM-CONNECT: Agitation ontology for the chatbot-based support of informal caregivers of people with dementia","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-22 11:54:21","doi":"10.21203/rs.3.rs-6289849/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"099e614c-a0a3-4d3c-b6c3-33b8685b5002","owner":[],"postedDate":"April 22nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":46913018,"name":"Health sciences/Health care"},{"id":46913019,"name":"Biological sciences/Computational biology and bioinformatics/Classification and taxonomy"},{"id":46913020,"name":"Biological sciences/Computational biology and bioinformatics/Data integration"},{"id":46913021,"name":"Biological sciences/Computational biology and bioinformatics/Databases"},{"id":46913022,"name":"Biological sciences/Computational biology and bioinformatics/Literature mining"}],"tags":[],"updatedAt":"2025-05-15T10:23:49+00:00","versionOfRecord":[],"versionCreatedAt":"2025-04-22 11:54:21","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6289849","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6289849","identity":"rs-6289849","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}