Dental EHR-infused Persona Ontologies to Enrich Dental Dialogue Interaction of Agents

Abstract

The quality of patient-provider communication can predict the healthcare outcomes in patients, and therefore, training dental providers to handle the communication effort with patients is crucial. In our previous work, we developed an ontology model that can standardize and represent patient-provider communication, which can later be integrated in conversational agents as tools for dental communication training. In this study, we embark on enriching our previous model with an ontology of patient personas to portray and express types of dental patient archetypes. The Ontology of Patient Personas that we developed was rooted in terminologies from an OBO Foundry ontology and dental electronic health record data elements. We discuss how this ontology aims to enhance the aforementioned dialogue ontology and future direction in executing our model in software agents to train dental students.

I. INTRODUCTION

Patient-provider communication has proven to play a major role in improving patient’s health and building a trustworthy relationship between the patients and the providers. Communication in healthcare is the first tool providers use during the process of treating patients. Providers are not “born with excellent communication skills and hence they must learn and practice this skill” [1]. Providers have conversations with patients to “understand their medical history, illness experiences, and to formulate treatment recommendations” [2]. A review of patient satisfaction done by Yeh and Nagel reported that patient satisfaction continues to be associated with knowing the specific approach used when talking with patients, and addressing patients’ expectations and needs during the initial interaction between the patient and the provider [3]. Patient-Provider communication plays an essential role in clinical practice. When effective, patient-provider communication contributes to positive outcomes for providers in reducing their job-related stress, and for patients, in improving their experience and satisfaction during consultation [1] [4]. For a positive relationship to be built between healthcare providers and patients, “there is a need for the existence of effective patient-provider communication” [5], which will ensure that all parties involved listen and fully understand each other. Good patient-provider communication can be facilitated by establishing a positive rapport and effectively using both verbal and non-verbal communication skills during dialogue. In addition to the provider’s expertise, effective patient-provider communication is a necessity in building a therapeutic provider-patient relationship [6] [7]. In health care settings, proper interactions between patients and providers are essential for an understanding of the patient’s visit as well as their expectations of the provider [8].

A. Dental specific patient-provider communication

Effective communication forges the dentist-patient relationship. The ability of a dentist to communicate is a necessity in the management of the clinic and the patients. Previous studies indicated that patients consider communication as one of the crucial traits in the ideal dentist [9]. Poor communication between dentists and patients, resulting in frustration on both parties, has been reported [9] [10]. It is thus recommended that dentists be trained to enhance their communicative skills [9] [11]. All patients, irrespective of the practice they came from, longed for a caring and respectful dentist who would listen to their concerns, without “blaming” them for their oral health status [12]. According to Abrams et al, “Simply practicing dentistry with a high degree of technical expertise will not necessarily convince the patient that he has received high quality dental care. Other less technical aspects of dental treatment are recognized as being barometers of quality of dental treatment. Practitioners should not lose sight of the human and psychological aspects of care, and keep in mind that they are integral components of quality in dental treatment” [13]. In a study done by Sbaraini et al, they observed that even when patients were hesitant about the value of a recommended treatment, having a dentist that cared about their problems encouraged their compliance [14]. The findings from the report by yuan et al suggests the importance of “strengthening the trust of patients towards their dentist”, which can be achieved through “effective dentist-patient interaction” [15]. Communication being linked to reduced dental anxiety and increased trust “tentatively supports the proposition that effective communication acts as a driver for regular dental visits” [15].

B. Benefits of patient-provider communication

1). Shared decision making

Research findings indicates that “dental patients’ desire to be involved in decision making” [16] and patients in general want more involvement in the care they receive [17] [18]. Actively involving patients in their medical care, affects their adherence to treatment recommendations and improve their comprehension and understanding of it [19]. Patients who feel that “she/he has a decision/role in the selected course of treatment may be more apt to follow their treatment regiment, which may lead to better outcomes and increased patient satisfaction in turn” [3]. When Doctors display good communication skills, they allow patients to perceive themselves as fully participating in discussions relating to their health [20].

2). Patient satisfaction and relationship building

To build a good relationship, one requirement is for healthcare providers and patients to communicate effectively. patient satisfaction is “one of the most recognized and widely used methods for assessing the effectiveness of physician-patient communication” [21]. Patient satisfaction is especially critical because it is an indicator of how well the provider is meeting patients’ health care needs, expectations, and preferences [22]. Studies on doctor-patient communication, have shown discontent on the part of patients, even when doctors felt their communication was excellent during a patient encounter [6]. Results from the study done by Dahlem et al suggest proper patient-provider communication was a key component in building a long-lasting provider-patient relationship [23]. In consultations where patients were able to freely express their feelings, concerns, opinions, and questions, better outcomes and increase patient satisfaction was observed [24]. There is increasing evidence that continuity of care, where patients consistently see the same provider, is associated with higher levels of patient satisfaction [25].

3). Compliance and decreased malpractice

Most complaints about doctors are related to issues of communication and not clinical competency [26] [27]. Improvement in provider communication skills is associated with greater patient satisfaction during interactions [28], and decreased malpractice claims [29]. Some studies have found associations between patient-centered communication and a decreased rate of malpractice claims [29]. Barrier et al reported that, providers having no malpractice claims filed against them, generally allowed patients to express their concerns, make sure they understand the patient’s concerns, and are generally more warm and friendly than physicians who have malpractice claims filed against them [30]. Therefore, improvement in communication overall proves to have a positive impact in the decrease in malpractice lawsuits.

4). Improved health outcomes

Communication has been repeatedly identified as a clinical skill having potential in improving health outcomes. However, this skill is not innate and must be learned properly and repeatedly practiced [29]. A doctor’s communication skill must incorporate the ability to gather information, give recommended instructions and build a relationship with a patient in a manner which is professional, caring, and sympathetic. Poor communication between healthcare providers and patients can result in negative health outcomes [5], rendering the patient dissatisfied. A trusting dentist-patient relationship plays a crucial role in improving health outcomes like a reduction in dental anxiety, through effective communication [31].

C. Overview on Personas

Personas can be described as fictional archetypes that characterize users based on several factors like behaviors, goals, needs and fears [32] [33] [34]. They provide a means to model and communicate research about the people being studied. The concept of Personas was created by Alan Cooper and popularized in his book The Inmates are Running an Asylum: Why High-Tech Products drive us Crazy and How to Restore the Sanity. Since the publication of his book, the use of personas has become widely accepted and adopted in different domains of study. As stated by Alan Cooper, “Personas are not real people, but they represent them through the design process; they are hypothetical archetypes of actual users; Although they are imaginary they are defined with significant rigor and precision” [35]. Over the years, Personas have been described as meant to “capture the user’s mental model comprising of their expectations, prior experience, and anticipated behavior” [36]. Personas represent different categories of people in the real world and go further to represent a larger group of individuals. They are portrayed as particular individuals put together from the observations of several people, giving a specific representation of target users [37].

Floyd et al. stated a number of personas types such as: quantitative personas, data-driven personas, user archetypes (which is similar to personas, but more generic); and marketing personas (for marketing purposes) [38]. Researchers have shown that good personas have four main features: They are fictional, engaging, goal-directed, and role-based [32] [39]. Personas typically consist of various elements such as the name and birth date of the individual being described, a photo or sometimes a sketch of the individual, description of interests, demographic information, life circumstances, personal goals and so on [40]. Through the use of these real-life characteristics and photos, the personas receive a livelier, and personal character [41] [38].

1). Personas in health care

Persona development as mentioned earlier has been adopted and used in several fields including the healthcare field. They have been used in studies focused on patients with chronic conditions like older adults with diabetes [42], heart disease [43], multiple sclerosis [44], and so on. They have helped characterize users in many health domains such as heart failure self-management [45], home healthcare technologies [46], child cancer survivors [47], or older adults [48].

Five sample Studies Involving the use of personas in healthcare include:

1. The study done by Bhattacharyya et al, made use of persona development to create 4 patient personas and 2 caregiver personas to promote the creation of a digital health advisor for patients with complex needs (i.e., elderly patients or patients with chronic health conditions) [49].

2. In the study done by Holden et al, they used quantitative methods to develop their personas. They develop biopsychosocial personas of older patients with heart failure using quantitative analysis of survey data collected on a sample of 32 older adults with chronic heart failure [43].

3. In the Study done by Agapie et al, they identified different persona types based off “hospitalized patients and their caregivers”. These personas included three patient personas (accommodating information seekers, involved safety guardians, and self-managing participators) and three caregiver personas (cooperative information seekers, vocal participators, hands-off safety guardians) [50].

4. In the study done by Huh et al they designed personas based on users of online health communities to “illustrate user’s needs and requirements in online health communities”. After interviews and survey responses, they emerged with four persona types which are the Caretakers, Opportunists, Scientists, and Adventurers [51].

5. In the study done by Reedner et al, they focused on the Oldest old demographic (85 years of age or older) and came up with two female personas ‘Hazel’ and ‘Rose’ from participants of their study in the wellness community setting [48].

D. Ontology

An ontology is a software artifact that models complex domain knowledge and information using machine readable syntax (i.e., OWL2 [52] and RDF[53], [54]). It is comprised of terms that are mapped to concepts and have labeled links between the terms that evoke meaning between concepts. For example, terms like “dentist” and “dental patient” with a labeled link of “treats” (dentist > treats > dental patient), expresses that all dentists treat dental patients.

Ontologies have a prominent role in the life sciences and have been embraced by biomedical community to assist modeling complex health data and linking very disparate health resources [55]. Some notable examples include the Gene Ontology [56], SNOMED-CT [57], Foundational Model of Anatomy [58], and many more. A quick glance of the open linked data cloud shows a large swath of ontologies belonging to the life sciences domain[59]. A goal is to be able to have a comprehensive knowledge base of health information by linking all of the health knowledge bases, which is theoretically achievable through ontologies [60].

In this study, we will utilize ontologies to model personas that contain information to express archetypes of patients that dental professionals may encounter. To the best of our knowledge this is the first to use ontologies to represent this specific type of domain information. In our later sections we will describe how this work will integrate into tools and enrich our previous ontology endeavor related to dialogue interaction between patient and provider [61].

E. Objective

From our earlier study [61], we introduced the use of speech-based conversational agents that can be utilized to train dental and medical students to experience some aspects of patient-provider communication. To enable the interaction, we created a standard dialogue ontology model that synthesize the discourse between patient and provider in a clinical setting. The objective in this work is to create diverse dental patient personas, that could be linked to that dialogue ontology model from our previous work, to enhance the machine’s impersonation of dental patients for the students to train with.

II. METHODS

A. Big Mouth Dental Data Repository

In practice, electronic health records are sometimes used as potential talking points when capturing information from patients. For this study, data regarding dental patients were obtained from the BigMouth dental data repository. BigMouth can be described as “a centralized oral health data repository derived from electronic health records (EHRs) at multiple dental schools participating in the Consortium of Oral Health Research and Informatics (COHRI)” [62]. This data repository contains data collected from over “3 million patients” and includes data such as patient demographic, medical history, dental history etc., which are made available for research purposes.

In the creation of the persona profiles, we obtained schema of the data from the data repository. This data schema was incorporated as questions directed to the patient and the patient’s response/manner of responding reflected the intended patient persona. A few examples of such data included: a) Patient social history which was an inquiry about the patient’s alcohol consumption and use of recreational drugs, tobacco, or nicotine products as well as the frequency at which it is done. b) Chief complaint, which included the patient’s reason for the dental visit and a description of their complaint. c) Patient Accommodations, which inquired about potential accommodations the patient may need during the dental visit such as the use of a sign language interpreter, accompaniment of service animal, transfer from the wheelchair, frequent bathroom breaks and so on.

B. Analyzing Persona Profiles

After reviewing works involving the use and description of personas [63] [64], common persona elements were observed and noted. Some of those elements such as a Fictional name, Job titles, Demographics, Status, Goals, and Photograph of the fictional, together with data from the electronic health records [62] such as patient Height, Weight, Temperature, Blood pressure, Heart rate, Respiratory rate, Body Mass Index, Pulse rate, were all incorporated in the creation of our persona.

Before starting on the creation of the dental patient persona profiles, the following key aspects regarding Dental patients were obtained from studies previously done by scientists.

1). Dental patient demographics

Dental care providers come across various types of dental patients and these patients vary by sex, age, personalities, race, ethnicity and so on. An earlier study focused on, linking medical and dental health record data. In that study, out of 158,786 persons, the data obtained from oral health care providers showed that patients who received any kind of dental service were classified as follows; “four broad age groups (0–18, 19–39, 40–64 and 65+ years), race (white, black, Asian and other race), ethnicity (Hispanic and Non-Hispanic) and sex ( Male, Female)” [65]. From this data classification, it was derived that the greatest percentage of individuals with dental records where between the 0–18 age group, with the highest race being White, the most common ethnicity selection being non-Hispanic and the most common sex being Female.

2). Patient personalities

In a research done by Booth-Kewley et al, they demonstrated how personality traits predict health and disease [66]. A common way in which personalities are described is known as the “Big Five” personality traits, which includes Agreeableness, Conscientiousness, Neuroticism, Openness to Experience, Introversion and Extraversion” [67] [68] [69]. Hajek et al further elaborated on each of these traits in their study by stating that: “(i) Agreeableness refers to the tendency to get along well with others and is associated with altruism or modesty. (ii) Conscientiousness refers to the extent to which an individual is careful, reliable, and persevering. (iii) Extraversion refers to the tendency to experience positive emotions and to have a positive outlook on life. In general, extraverts are talkative, sociable, outgoing, enthusiastic, and energetic. Introverts are usually categorized as Type A (confident, independent, and good in interactions) or Type B (shy, lack communication skills, timid and withdrawn) [71]. (iv) Neuroticism refers to the tendency to experience negative emotional states including anxiety, depression, or anger. (v) Openness to experience refers to the tendency to be open-minded, imaginative, and curious” [70].

3). Patient Odontophobia

A prevalent condition among dental patients called Odontophobia wherein patients experience a phobia for dental procedures and dentist [72] is increasingly common. According to World Health Organization, Odontophobia is recognized as a real disease and about 15–20% of the population is affected by it [73]. Odontophobia can be classified into three classes namely: “Mild odontophobia, also called dental anxiety, is the most frequent among the population; Moderate odontophobia, also called dental fear; Severe dental phobia, the real dental phobia, decidedly rarer and more difficult to manage by the dentist” [72].

4). Dental patient types

Based on an article from Spear Education: Continuing Education for Dentist titled “The 5 Patients Every Dentist will treat”, Manji, described the five patients as; “The Event-Driven patient, (who only comes in for dentistry when it is absolutely necessary and arrives with a “must fix it” mindset), the Reactive patient (These are patients who understand the need for regular hygiene and exams, but who for one reason or another only participate in a limited way), The Proactive patient (who keep their appointments, take their oral health seriously and complete treatment that they accept), the Discretionary patient (one who, for whatever reason, is ready to look beyond tooth-based care and is ready to invest in esthetic treatments) and the Regenerative patient (who want the best care option, and they are willing to invest in making their oral health as good as possible)” [74].

After obtaining the necessary information/ data on dental patients, we aimed to create dental patient persona profiles which were inclusive of most of the data discussed above.

III. RESULTS

The final output of this work was coded in Web Ontology Language (OWL2) using Protege. In total, the Ontology for Patient Persona’s classes amounted 655, and 95 object properties and 6 data properties. The total axiom count resulted in 4744. This work is available through Github repository: https://github.com/UTHealth-Ontology/Ontology-of-Patient-Personas.

Figure 1 describes the basic “backbone” of our model. In that figure we have Persona concept that represent one Persona document represented as instance data. Also, for each Persona concept there are several Persona Elements. These Persona Elements describe the various features described in Section B (demographic information, personality types, EHR data elements, etc.). Essentially, these elements are representative of the textual content of the individual persona profiles.

Figure 1.

Conceptual description of the Ontology of Patient Personas in alignment with OBO Foundry’s Information Artifact Ontology concepts.

With the basic underlying structure, we aligned the model to the Information Artifact Ontology (IAO) [75], a realism-based reference ontology that describe high-level information entities, such as documents, digital assets, data sources, etc. The IAO is one of OBO Foundry ontologies that are community supported by life science researchers to curate ontologies that strictly adhere to design principles, like openness and interoperability, and ontological realism.

In user experience research, a persona is sometimes manifested as document but the persona itself is a fictious representation of some archetype of a target user [76]. We classified a persona as an information content entity (ICE) which is defined as a “a generically dependent continuant that is about something” ¹. Similarly, for the data elements that compose the persona were classified the same. In the same figure (Figure 1), we show how the persona model we invented is aligned with IAO. The main concepts Persona is subclassed as an information content entity, and the Persona Elements that were derived from demographic and EHR schema data are subclassed as an information content entity. The various Persona Elements were linked by the “has part” object property to denote that the Persona and its subclasses have various Persona Elements.

We specified the Persona concept by creating a Health Patient Persona. This would allow for creating types of Personas that are not necessarily health related. We also further specified the Health Patient Persona by creating a concept specific to dental, Dental Patient Persona. Using research we surveyed, we identified five types of possible persona types. We incorporated the Event Driven Patient Persona, Proactive Patient Persona, Reactive patient persona, and Regenerative Patient Persona from the work done by Manji. These were previously discussed in the section Dental patient types. In addition to that, we included the Fearful Patient Persona to depict patients who may experience the different levels of odontophobia.

Noted before, Persona Elements are based on demographic and data schema from EHR records. Briefly, we present Table 1, that describes the various subclasses of Persona Elements to furnish our ontology model.

Table 1.

Sample concepts from the ontology

Data item (and subtypes)	Description
Patient_Persona_Baseline_Information (Blood_pressure_Information, Body_Mass_Index_Information, Heart_Rate_Information, Height_Information, Respiratory_Rate_Information, Temperature_Information, Weight_Information)	Information gathered from patients during the first initial consultation
Patient_Persona_Chief_Complaint (Cavities/lost_or_broken_fillings_compliant, Check-up/cleaning, Crooked_teeth/need_braces, Crowns/bridgework, Dentures/patient_dentures, Gum_Problems, Missing_teeth/spaces, Oral_pathology, Pain_Complaint_Root_canal, Toothache_Compliant, Unhappy_with_smile_or_other_aspect_of_the_way_teeth_look_or_feel, Wisdom_Teeth)	The patient’s reason for consultation
Patient_Persona_Family_History (Family_History_of_Arthritis, Family_History_of_Cancer, Family_History_of_Diabetes, Family_History_of_Heart_Attack, Family_History_of_Heart_Disease, Family_History_of_High_Blood_Pressure, Family_History_of_Stroke)	Diseases that run through/affecting the patient’s family
Patient_Persona_Odontophobia_Level (MILD_level_of_Odontophobia _'dental_anxiety', MODERATE_level_of_Odontophodia _‘dental_fear’, SEVERE_level_of_Odontophobia _'dental_phobia' )	A measure of patient’s odontophobia (phobia of dental procedures)
Patient_Persona_Required_Accomodations (Anti-Anxiety_Protocol_(e.g._nitrous_oxide/O2), Operatory_Close_to_Waiting_Area_ (mobility_concerns), Accompaniment_of_Service_Animal, Caregiver_to_Stay_in_Operatory_ During_Appointment, Frequent_bathroom_breaks, Modification_from_Standard_ Chair_Position, Transfer_from_Wheelchair, Use_of_Additional_Neck_or_ Back_Supports, Use_of_Primary_Language_Interpreter, Use_of_Sign_language_Interpreter, Use_of_Supplemental_O2)	Accomdations required by the patient during the dental visits.

Lastly, one of the primary motivation to develop this ontology was to utilize it for application purposes. For each concept utilized in this model, we have a data property called hasUtteranceExpression. This data property permits the addition of various natural language text to express how the persona would express this in their own “voice”. In many personas, there is often an utterance in the persona document that represents the voice of the character of the persona. Not only we use this data property to represent that feature, but we also intend to use this as an application feature when the ontology is used as component for the conversational agent.

IV. DISCUSSION

To the best of our knowledge this work is the first attempt to utilize ontology-based method to model persona profiles, specifically for patients in healthcare. The Ontology of Patient Persona aims to be reusable computable artifact for persona profiles that be integrated for a software agent. This ontology relies on the IAO framework to help abstract the information entity of a persona and align it to the body of biomedical ontologies existing on the OBO Foundry.

The innovative aspect of this work is that it presents the application of EHR-related data schema to furnish a computable persona profile as an ontology artifact. The data schema derived from the EHR characterizes a description of the patient. Future work could involve expanding the use of this ontology for other applications aside from our software agent use case. Another contribution of this work is the usage of an OBO Foundry and their common framework (BFO) to construct this ontology. This effort could lead to further work in developing standard personas for healthcare (and non-healthcare) supported by taxonomies and semantic architectures that are grounded in community agreement and scientific rigor. Lastly, the other contribution is the enrichment of our previous work to link this persona ontology to a dialogue interaction ontology for patient-provider communication, giving conversational agents individualized utterance expression representative of archetypes.

From our previous study we cited the need for tools that can help improve patient-provider communication, and one approach is to mimic patients through conversational agents where a dental student could acclimate their verbal skills to the level of the patient. This could improve a variety of health outcomes such as improved diagnoses and patient satisfaction. One of the first steps was to develop the interaction model to assist the machine to perform discourse with a human agent (i.e. dental student). From sample scripts, we derived a formal model of interaction and encoded an ontology that leveraged the Patient Health Information Dialogue Ontology (PHIDO) [77] model framework as a foundation. One future goal mentioned in that paper was to manifest patient variety that a dental student can train on to handle diverse patients - which led to the genesis of this current work.

The diverse dental patient personas created depict various types of patients a dentist may encounter in their clinical practice. Understanding these patient personas may help providers have better insight on how dental patients differ from each other; by their personalities, life experiences and so on. This may enable the providers incorporate that knowledge in personalizing the care suitable for each type of patient.

In Figure 2, we show how both the persona ontology and the dialogue ontology would potentially link to each other. The dialogue ontology contains a series of linked utterances forming speech objectives. These utterance concepts would link to persona elements concepts to fetch the utterance expression data property. Essentially, these linked utterance expression would give each persona model (and by default the agent that uses the model) their “individual voice”.

Figure 2.

Usage of Ontology of Patient Personas linking to enrich the dialogue ontology.

A. Future Direction

Through semantic interoperability of ontologies, our next phase is to bond the two ontologies and integrate it to our software engine to demonstrate the dynamic multiple patient expression that machine can embody. The predicted output of this future effort will be an ontology-driven dialogue software system to be deployed for devices for usability evaluation and feasibility testing.

In addition, we will create sample persona profiles manifested as instance data of the five identified dental patient types. We currently have developed six personas that we intend to convert to instance data for the ontology. These personas include: (i) The Ideal patient persona representing a Proactive patient, (ii) The Committed patient persona representing a Regenerative patient, (iii) The Immigrant patient persona representing an Event-Driven patient, (iv) the Adolescent patient persona representing a Reactive patient, (v) The Anxious patient persona representing a Fearful patient and (vi) The Disabled patient persona created to add diversity to our patient archetypes.

V. CONCLUSION

We introduce an ontology, Ontology of Patient Personas, that represent patient-specific persona artifacts that describe potential archetypes of dental patients. The ontology integrates with Information Artifact Ontology that leverages OBO Foundry terminologies and its standard underlying structure for biomedical ontologies. Also we based on our ontology on elements in dental electronic health records to capture a data-driven “picture” of a dental patient. We intend to align this ontology with our previous work on a dialogue ontology for a conversational agent that can take on various patient archetypes that dental students can train to improve their communication skills. Our future direction will include this and deploying the integrated ontology mode in a software engine for usability assessment and testing.