Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Using Ethnographic Methods to Classify the Human Experience in
Medicine: A Case Study of the Presence Ontology
Amrapali Maitra, M.D., Ph.D.1,2, Maulik R. Kamdar, Ph.D.3, Donna M. Zulman, M.D., M.S.4,5,
Marie C. Haverfield, Ph.D.6, Cati Brown-Johnson, Ph.D.4, Rachel Schwartz, Ph.D.7, Sonoo Thadaney
Israni, M.B.A2, Abraham Verghese, M.D.2, Mark A. Musen, M.D., Ph.D.3
1 Department of Medicine, Brigham and Women’s Hospital, Boston, MA, USA
2 Stanford Presence Center, School of Medicine, Stanford University, Stanford, CA, USA
3 Center for Biomedical Informatics Research, Stanford University, Stanford, CA, USA
4 Division of Primary Care and Population Health, Stanford University, Stanford, CA, USA
5 Center for Innovation to Implementation, VA Palo Alto Health Care System, Menlo Park, CA, USA
6 Department of Communication Studies, San Jose State University, San Jose, CA, USA
7 WellMD Center, Stanford University School of Medicine, Stanford, CA, USA
Abstract
Objective
Although social and environmental factors are central to provider-patient interactions, the data that reflect
these factors can be incomplete, vague, and subjective. We sought to create a conceptual framework to
describe and classify data about presence, the domain of interpersonal connection in medicine.
Methods
Our top-down approach for ontology development based on the concept of “relationality” included the
following: 1) broad survey of social sciences literature and systematic literature review of >20,000 articles
around interpersonal connection in medicine, 3) relational ethnography of clinical encounters (n=5 pilot, 27
full) and 4) interviews about relational work with 40 medical and nonmedical professionals. We formalized
the model using the Web Ontology Language in the Protégé ontology editor. We iteratively evaluated and
refined the Presence Ontology through manual expert review and automated annotation of literature.
Results and Discussion
The Presence Ontology facilitates the naming and classification of concepts that would otherwise be vague.
Our model categorizes contributors to healthcare encounters and factors such as Communication, Emotions,
Tools, and Environment. Ontology evaluation indicated that Cognitive Models (both patients’ explanatory
models and providers’ caregiving approaches) influenced encounters and were subsequently incorporated.
We show how ethnographic methods based in relationality can aid the representation of experiential concepts
(e.g., empathy, trust). Our ontology could support informatics applications to improve healthcare such
annotation of videotaped encounters, clinical instruments to measure presence, or EHR-based reminders for
providers.
Conclusion
The Presence Ontology provides a model for using ethnographic approaches to classify interpersonal data.
!
!
2!
1 INTRODUCTION
Modern medicine has advanced the treatment of disease but at times infringes on the simple ritual of doctors
using compassion, listening, and skilled touch in the bedside exam to connect with patients1,2. The emotional
labor of creating connection is an important part of a healthcare provider’s role, yet such care is “far more
complex, uncertain, and unbounded than professional medical and nursing models suggest”3. Investing in
interpersonal connection may prevent burnout of healthcare providers and increases patient satisfaction4,5,
yet it is challenging in our technologized era of medicine2,6,7.
Presence is an emerging medical discourse that refers to the “purposeful practice of awareness, focus, and
attention with the intent to understand and connect with patients”8. There is currently no unifying framework
to describe, capture, and classify human and environmental data surrounding interpersonal connections in
clinical encounters, and the interpersonal interactions that comprise presence cannot be gleaned from the
EHR9. We define “interpersonal” as “a selective, systemic process that allows people to reflect and build
personal knowledge of one another and create shared meanings”10. The presence domain is of increasing
relevance to informatics research and applications that seek to improve the individual experience of
healthcare and delivery systems11 through electronic health record (EHR) innovations, scribe programs for
documentation, or integration of smartphones into clinical care.
Data related to human experiences and social interactions are often incomplete and sometimes subjective;
they are documented qualitatively in multiple, idiosyncratic, and partial ways. Such data arise through
interactions among different individuals, with diverse objects, across multiple physical and virtual spaces.
An example is the Social History within the EHR where demographic data like marital/partner status,
occupation, substance use, and sexual history are listed in series of drop-down boxes rather than elaborated
as an opportunity to situate medical complaints within patients’ complex life circumstances12.
In this study, we combine biomedical ontology engineering with ethnographic methods to define the factors
contributing to interpersonal connection in medicine. Specifically, we have conceptualized and developed
the Presence Ontology, a systematized vocabulary of terms that models the interactions taking place every
day among healthcare providers, patients, and their families and friends. Developing a conceptual vocabulary
for presence could generate informatics innovations to better evaluate the patient experience including
satisfaction13; mitigate clinician burnout and support joy of practice14,15; and equitably deliver personalized
care in the artificial intelligence (AI) transformation of medicine16.
The ontology was developed through interdisciplinary collaboration of experts in medicine, bioinformatics,
anthropology, linguistics, communication, psychiatry, and public health. Our approach may provide clarity
and consensus to the important but ill-defined domain of human experience in bioinformatics and has
relevance to informatics subfields where interpersonal data are central to knowledge classification domains.
2 CLASSIFYING PRESENCE
We sought to identify the elements of interpersonal connection in the patient-physician relationship and
engineer them into an explicit formal specification called an ontology. By utilizing a shared language with
defined relationships, we strove to make subjective data and metadata in healthcare more expressive and
precise—such data are often taken to be a black box by computational researchers because they are subjective
and may be vaguely defined. The development and use of ontologies for clinical care is a critical requirement
in the creation of automated decision support tools and clinical research databases for data harmonization
and semantic interoperability17. An example of ontology engineering of broad clinical concepts is the widely
used and exhaustive vocabulary known as the Systematized Nomenclature of Medicine – Clinical Terms
(SNOMED CT)18.
In order to model human experience in clinical encounters, we combine the conceptual abstraction of social
sciences theory with the granularity of ethnography. Literature from sociology, anthropology, and linguistics
provides fertile conceptual terrain to describe human experiences related to healthcare. Ethnographic
!
!
3!
methods are of increasing interest in bioinformatics19,20; they are fine-grained and describe the variability of
experience. Our methods build on ethnographic approaches for developing electronic knowledge bases21–23.
Our ontology shares some categories with existing frameworks24–26 including characteristics (identity
features that define both patients and providers), encounters (instances where multiple people come together)
and emotions (intrapersonal experiences through which interpersonal experience is mediated), but brings
additional rigor by naming terms with logic and consistency for usability across many providers and
interactions. We extend Ventres and Frankel’s “shared presence” framework focused on providers’ behaviors
and actions (e.g. to listen, examine, educate) by incorporating the behaviors and qualities of patients and
influence of environments in shaping presence within clinical encounters24. We also build on Larson and
Yao’s model of empathy which describes how antecedents (e.g. physician or patient characteristics or
situational characteristics) affect empathic processes, which in turn result in intrapersonal and interpersonal
outcomes that extend to physician and patient outcomes (e.g. burnout, patient satisfaction, and healthcare
outcomes)25. While the framework models clinical encounters as linear and one-dimensional, we elaborate
further to account for the multiple, intersecting ways in which people’s characteristics, environments, and
behaviors coalesce to shape presence.
The ontology leverages the Presence 5 framework, which describes evidence-based practices that promote
clinician presence: 1) prepare with intention, 2) listen intently and completely, 3) agree on what matters most,
4) connect with the patient’s story, and 5) explore emotional cues.9 These recommendations embed concrete,
measurable actions and behaviors within the clinical encounter, such as use of time, body position,
management of a computer screen, and communication style. Formative research for the Presence 5
framework included a systematic literature review, observations of clinical encounters, and interviews with
physicians and non-medical professionals, data, which also informed ontology development (Section 3).
3 METHODS
We applied ethnographic principles to ontology engineering centered on the concept of relationality. We
identified and modeled key concepts and relations surrounding the domain of presence and developed the
Presence Ontology into a formal, usable clinical artifact. We followed a top-down approach for ontology
development starting with the highest-level (most abstract) concepts in our domain and then defining sub-
concepts. Domain analysis was based upon literature survey, ethnographic observations of clinical
encounters, qualitative insights from professionals engaged in relational care, and meetings of clinical and
research experts over nine months.
3.1 Broad survey and systematic review of domain literature
We identified preliminary concepts pertaining to the domain of presence through a broad survey of literature
in both medicine (using PubMed27) and the social sciences (journals and books of anthropology,
communication, sociology, and psychology) around topics of interpersonal connection. Keywords for the
broad literature survey included Patient-Physician Relationship, Communication, Empathy, Power
Dynamics, Patient-Centered Care, Technology, Burnout, Trust, and Mindfulness. The broad survey allowed
for attention to concepts that may not readily appear in the conventional medical literature, such as the cultural
and political milieu of healthcare, the spoken and unspoken aspects of clinical care, as well as the hierarchies,
language, and feelings that structure encounters. Key concepts of relevance to presence included the clinical
encounter as ritual1,28, power dynamics in medicine29–31, the emotional labor and ethical stance of care
work26,32, the illness experience and explanatory models31,33, and the impact of technology and the built
environment on the connection between patients and providers34,35
The broad survey was expanded through a systematic literature review conducted for the Presence 5 study9,36.
Three databases were searched across biomedical and social sciences (PubMed27, EMBASE37, and
PsycINFO38) capturing research from January 1997 to August 2017 for randomized controlled trials and
controlled observational studies of evidence-based interpersonal interventions geared toward improving
!
!
4!
presence that included at least one outcome measure of the “quadruple aim” (i.e., patient health outcomes,
patient experience, clinician experience, or cost)36,39. A broad array of MeSH terms and keywords
encompassing domains such as trust, empathy, humanism, and communication were used. The review
yielded 21,835 articles; 77 of which were retained after screening of titles, abstracts, and full texts (and 73 of
which are the focus of a published systematic review36). For ontology development, the 77 papers were
reviewed in-depth for content and conceptual language related to presence. The abstracts were also used in
a later stage of ontology evaluation and revision (See Section 4.5).
The review began with 21,835 articles. After screening of titles, abstracts, and full texts (including systematic
reviews for component studies), 77 unique studies were preliminarily identified for quality assessment and
data extraction stage (in the published systematic review, 73 of the 77 articles were retained). These papers
were reviewed for content and conceptual language related to presence. The abstracts were also used in a
later stage of ontology evaluation and revision (See Section 4.5).
3.2 Relational ethnography to develop upper-level categories and
concepts
The literature survey resulted in an initial conceptual sketch of the domain, modeled around three key
components of clinical encounters: 1) providers, 2) patients, and 3) environments in which encounters take
place. However, subsequent discussion exposed many gaps; the model did not allow us to speak about factors
generated through interaction (such as trust) as well as identities shared across types of persons (both
providers and patients, for example). Seeking new ideas for upper-level abstraction, we used ethnographic
methods to elaborate the configuration of presence encounters.
Relational ethnography of clinical encounters (n=5 pilot, 27 full) allowed us to examine tension,
incompleteness, or unexpectedness. Relationality provided a conceptual foundation to situate patient
encounters as hierarchical, interpersonal, spontaneous, and unbounded, rather than modeling patients as fixed
entities with rigid roles. We use “relationality” as defined by sociologist Desmond, as a focus on
understanding processes, fields, and conflicts40. Relational ethnography conceptualizes interactions beyond
the boundaries of place or group; instead, it seeks to “broaden and expand” the field of objects and
relationships therein. This method is useful for health informatics because it is grounded in practice theory,
attending to the ways in which “technology and social practice are mutually elaborated” in clinical
interactions41.
An anthropologist observed five pilot clinical encounters with IRB approval (Stanford Institutional Review
Board Protocol No. 30711). Written informed consent was obtained from patients prior to observations.
Observations were centered on the elements of presence in patient-physician interactions with one provider
at a medicine subspecialty clinic in California. Fieldnotes focused on structure and content of conversation,
body position, movement and touch during the physical examination, and patterns of speech and silence.
Patients were interviewed post-visit using a semi-structured interview script to elicit perceptions about the
level of connection to provider, feelings of vulnerability, perceptions of care and empathy, and satisfaction.
To validate findings from the pilot observations, we analyzed data from 27 patient-physician encounters that
were conducted for the Presence 5 study described previously9. These encounters spanned three primary care
settings in California (an academic medical center, a Veterans Affair facility, and a federally qualified health
center) and used a “rapid ethnography” approach42 centered around a conceptual model for Presence
observations (Figure 1). Methods included observation and video- or audio-recording of clinical encounters,
fieldnotes, team debriefing, post-visit interviews with clinicians and patients about strategies to foster
presence, and consensus coding9. Procedures were approved by Stanford University IRB (Protocol No.
42397). Physicians and patients provided written informed consent for observations, recordings, and
interviews.
!
!
5!
Figure 1. Conceptual model for clinical observations to develop Presence relational ethnography. The model
depicts the conceptual hierarchy for themes related to the clinical encounter. The model was developed through
literature survey and expert review and finalized prior to conducting the relational ethnography. Presence research team
members were trained using this model in order to structure fieldnotes for observed patient-physician encounters
(n=27) using a rapid ethnography approach. At the core is the clinical ritual, upon which is layered interpersonal
interaction (with attention to verbal and non-verbal communication, timing, and silence), then individual identity
features of both the clinician and patient, structural and systems-level features such as clinic resources or wait time, and
finally the environmental milieu within which encounters occur. Additional elements that mediate the encounter
include power dynamics, care team members, the patient’s family and friends, technology, tools, and touch.
3.3 Qualitative insights from relational care professions to refine
categories
The ethnographic study was supplemented by trans-disciplinary qualitative insights, leveraging data from
formative research for the Presence 5 study, in which purposive sampling was used to identify 30
professionals from outside the field of medicine whose work involves relational care, such as police officers,
personal care services (e.g. yoga, massage therapy), management fields (CEO, school principal), education,
the arts, and social services9,43. Participants were interviewed about their approaches to interpersonally
intense encounters. An inductive analysis was used to identify themes. 10 physicians were also interviewed
to compare and correlate themes8. Themes included conscientious approaches to self-care that permit greater
presence, protected time for peer interaction, and an emphasis on fostering a bidirectional exchange to
enhance professional fulfillment. These insights refined our upper-level categories of presence.
3.4 Collective domain expertise of experienced clinicians and researchers
Collective expertise was elicited from a team of researchers whose backgrounds encompass medicine,
bioinformatics, anthropology, communication, linguistics, psychiatry, and public health. The research team
met biweekly over six months to iterate on conceptual categories and relationships within the ontology. At
each stage, the team of experts reviewed evidence to date, refined subsequent methods, and discussed
applicability of concepts (through reflection on moments of presence breakdown in individual clinical
experience, for example). This allowed for synthesis of mixed methods to generate ontology concepts.
!
!
6!
3.5 Formalization of ontology using OWL in Protégé
The ontology was implemented using OWL (Web Ontology Language) in Protégé, one of the most widely
used open source ontology editors44. Using the preliminary steps, we enumerated different class entities and
properties. We modeled these entities using appropriate OWL axioms such as classes, object properties, data
properties, and annotations. Then, following best practices45, we externally cross-referenced relevant terms
from two existing ontologies, SNOMED CT and the Emotion Ontology18,46. We did not import either
ontology because the contexts in which they were developed are different from our own—that is, SNOMED
CT is supposed to be an exhaustive clinical reference terminology, whereas the Emotion Ontology is focused
heavily on emotions rather than clinical encounters or their participants. While we could have extracted
modules from SNOMED CT for import within the Presence Ontology, such modules could have led to
semantic inconsistency. Finally, we uploaded the Presence Ontology in the BioPortal repository47, the
world’s largest open source repository of biomedical ontologies, for annotations and dissemination.
3.6 Evaluation and revision of ontology from the presence systematic
review
A corpus of abstracts from the systematic review of presence (Section 3.1) was used to evaluate the Presence
Ontology36. We conducted a manual formative assessment of the abstracts along with a review from clinician
experts to identify key conceptual gaps in our initial ontology. We also completed a data-driven summative
evaluation using the Presence ontology (uploaded in the BioPortal repository) and the BioPortal Annotator48,
which is widely used for the annotation of biomedical texts and electronic health records with UMLS
concepts. The BioPortal Annotator takes as input a dictionary of term labels as well as a set of their synonyms
(e.g., “Clinician,” “Physician,” and “Doctor”) and generates annotated abstracts as output. We annotated 77
abstracts with 306 terms (classes and object properties) from the Presence Ontology. We also identified terms
that frequently co-occur (i.e., mentioned together in the same abstract) with a given term.
4 RESULTS
4.1 Presence Ontology
Our ontology models presence within clinical medicine. Patients and providers are involved in encounters,
which contain several subparts. Encounters are influenced by various personal, environmental, and
relationship factors, as well as emotions, qualities, and characteristics of the participants. Encounters often
result in health outcomes. They also produce qualities that set the stage for future encounters. Interactions
are also modeled as time-bound (with a start time and an end time).
A diagrammatic representation of the model behind the Presence Ontology is illustrated in Figure 2. The
colored regions in the representation represent the different upper-level class entities with the subclasses
listed in those boxes. Object properties are italicized alongside the arrows between class entities.
Our model includes the following upper-level class entities shown in the colored boxes:
• Person - All individuals and specific Healthcare Roles like Patients, Providers, or “Framily” (i.e.,
Friends and Family) involved in an encounter
• Characteristic - Defining features of individuals such as age, gender, race, etc.
• Encounter - A subclass of Events, which involves at least one Provider and at least one Patient.
Encounters can consist of multiple sub-encounters and smaller interactions or “Encounter
Components”. Encounters result in outcomes for patients, providers, and healthcare systems.
• Action - Each Encounter Component may involve several actions (e.g., patient workup) that are
performed by the individuals involved in the Encounter
• Object - An Action often involves the use of certain tools, and on the opposite spectrum, an Action
may be interrupted by certain Objects (e.g., interruption by pager alert)
!
!
7!
• Factor - Encounters between a patient and a provider may be influenced by several external factors,
which often include Communication (Verbal, Nonverbal, Paralinguistic), the nature of the Patient
Provider Relationship, elements of the Patient History, and features of the Environment.
• Quality - Non-relational qualities (e.g., experienced internally by a single individual) and relational
qualities (e.g., interpersonal connection experienced by two or more individuals) can be generated
through Encounters or can influence them.
• Emotion - Encounters are also influenced by emotions of the different participants in that encounter.
!
Figure 2. Class diagram of the Presence Ontology elaborating upper-level class hierarchies and object properties.
Each colored box indicates an upper-level class (e.g., “Factor”), with subsequent inner boxes depicting the hierarchy
under the upper-level class (e.g., “Patient History” is a subclass of “Factor”, and in turn “Family History” is a subclass
of “Patient History”). Different classes are connected using object properties in the Presence Ontology. For example, the
object property “performs” associates the class “Person” with the class “Action” (an individual under the class “Person”,
whether a “Patient” or a “Provider”, will perform some “Action”), whereas, the object property “hasCharacteristic”
associates the class “Person” with the class “Characteristic” (an individual under the class “Person” has at least one
“Characteristic”, such as “Age”, “Occupation”, “Race”, etc.). For simplicity, we have only shown the most relevant
classes in this class diagram and refer the reader to explore the Presence Ontology in the BioPortal repository for more
information around the class hierarchies and the object properties.
!
!
8!
The Presence Ontology allows for naming and classification of concepts that would otherwise be vague. The
ontology also allows mapping of the relationships among these disparate components in an encounter. Our
ontology uses the object property “produceQuality” to model how a relational quality like compassion is
produced within a specific interaction. This is depicted in Figure 2 where the “produceQuality” object
property is associated with the classes “Encounter Component” and “Quality”, where “Quality” is further
subcategorized into relational (e.g., compassion) and non-relational qualities (e.g., self-confidence). Recent
ethnographic studies of medical encounters support the idea that qualities like compassion are both
“dispositional” (an individual quality or characteristic) and “situational” (related to the encounter context)49.
The Presence Ontology also suggests new conceptualizations of technology and time in clinical encounters.
The ontology uses object properties such as “InvolvesUseOf” and “InterruptedBy” to describe negative and
positive ways a technological object (computer, smartphone, etc.) could act within an encounter. Thus, it
offers a way to integrate data around technologies or devices in healthcare as both a tool and a barrier to
developing interpersonal connection. Second, conceptualizing medical encounters as time-bound entities
allows clinical applications like tracking time-related data. In medicine, encounters length is an important
variable, as time (e.g., length of encounter, wait time) may affects patients’ perceptions of the interaction. In
one study of 5,000 patients concerning prior healthcare encounters, time with physicians was a stronger
predictor of patient satisfaction than wait time50.
4.2 Ontology Evaluation and Revision
4.2.1 Formative Evaluation
Through the manual formative assessment of the initial ontology and the abstracts extracted from the
systematic review, we identified a key conceptual gap in our ontology: cognitive models. We revised the
ontology to include Cognitive Model as a subclass of Factors that influence Encounters. Cognitive Model
comprises both Caregiving Approaches (approaches held by providers such as patient-centered care, shared
decision-making, or motivational interviewing) and Illness Models (also termed illness representations or
illness scripts) conceptualized by patients. While intrapersonal experiences like emotions had been modeled,
cognitive processes were an important addition as they indicated how attitudes regarding care or illness may
shape encounters.
4.2.2 Summative Evaluation
Figure 3 depicts the top 20 terms from the Presence Ontology that frequently appeared in our corpus of 77
abstracts. Each red bar indicates the total number of abstracts in which the represented term is mentioned,
whereas the corresponding blue bars indicate the total number of other terms from the Presence Ontology
with which the represented term co-occurs in the corpus (i.e. in a given abstract, the represented term is
identified with a few other terms from the Presence Ontology). For example, from the histogram, it is
apparent that the concept of “Patient” is identified in more than 70 abstracts and co-occurs with
approximately 90 other terms from the Presence Ontology.
The systematic review identified papers with a high patient-centric focus, which previous models in the
domain of presence often lack. Through this histogram, it is evident that the Presence Ontology is able to
identify concepts such as “patient-centered care”, “patient satisfaction”, “confidence”, and “stress,” which
focus more on the qualities and emotions that are the outcome of patient-provider encounters. The inclusion
of these patient-centric terms in the Presence Ontology broaden the focus from providers as the only possible
driver of interactions, a focus that is prevalent in existing models, toward a relational view of encounters.
Finally, concepts which refer to “Cognitive Models” (11th most common term in Figure 3) and were
identified in the formative assessment are often mentioned in literature surrounding presence.
!
!
9!
Figure 3. Concepts from the Presence Ontology that were most commonly identified in the 77 Abstracts related
to Presence literature. The X-axis in this histogram showcases the 20 most commonly identified concepts from the
Presence Ontology in the Presence literature. Each red bar in the histogram indicates the total number of abstracts in
which the represented concept is mentioned, whereas each blue bar indicates the total number of Presence concepts that
co-occur with the represented concept in these abstracts. The importance of patient-centric concepts in our approach
broadens the focus from providers as drivers of human experience in medicine toward a relational framework for
presence, the direct outcome of our ethnographic methods for ontology development.
Using this evaluation approach and the Presence Ontology, we can construct presence co-occurrence
networks to understand which factors and qualities are simultaneously experienced in clinical encounters.
For example, as shown in Figure 4, the emotion “Stress” is mentioned in 13 abstracts, and in those abstracts
it co-occurs with approximately 45 other terms from the ontology. “Stress” frequently co-occurs with both
“Patient” and “Physician,” as well as with the concepts of “Outcome,” “Confidence,” “Anger,” “Empathy,”
and “Trust”. The co-occurrence analysis can provide insight into features that may diminish presence in
encounters.
!
!
10!
Figure 4: Co-occurrence network for the concept of “stress” generated from the identification of Presence
concepts in literature. The concept of “stress” from the Presence Ontology generally co-occurs with common concepts
of “patient” or “physician”, but also with concepts such as “empathy”, “anger”, “trust”, etc. The size and the color of the
nodes is indicative of the number of abstracts in which the presence concept is identified, and the thickness of the
connecting edges between two nodes is indicative of the number of abstracts in which the connected concepts co-occur
together.
!
5 DISCUSSION
The specificity of controlled vocabularies can push forward the fields of medicine and clinical informatics.
We have sought to introduce specificity to the domain of presence with challenging data qualities that are
subjective (based on individual perception and experience), partial (not all information about the domain can
be known), unpredictable (both patients and providers often improvise in interactions), and ever-changing
(any changes in social world produces a change in the domain of human connection).
The ethnographic approach of relationality, grounded in practice theory, provides a useful model for
development of knowledge systems that strive for ontological realism while remaining rooted in the core of
healthcare: human interactions occurring over time. The role of realism in ontologies has been a debate in
clinical and biomedical informatics; Smith and Ceusters argue that ontologies should comprise universals
taken from an objective reality51. The Presence Ontology uses social theory and relational ethnography to
model the multiple, idiosyncratic, unbounded interactions of clinical care into a controlled vocabulary that
attempts to define the reified elements of presence and resultant relationships. While this may appear
paradoxical, a relational approach is useful in order to elucidate what would otherwise be a black box for
!
!
11!
knowledge classification, even if such an approach suggests the limits of ever perfectly transcribing
experiential data that is situated, relativistic, and inherently partial. We sought to develop a usable
interdisciplinary biomedical ontology that provides a shared language for the subjective components in
medicine, without compromising formal rigor and cross-domain interoperability qualities. This approach is
resonant with the move toward “social interactionism” in informatics, based on Kaplan’s model of
communication, control, care, and context52, and in the pluralistic methods of socio-informatics41.
Using relational approaches, we modeled abstract entities (e.g. individuals, encounters, characteristics,
emotions, qualities, tools) within which domain-specific categories (such as doctors, patients, stethoscopes,
waiting rooms, medications, or diseases) can reside. Encounters are interpersonal experiences where concepts
like empathy and trust cannot be attributed to one entity but are the result of complex interactions. To this
end, we developed the class entity “Relational Quality” that defines qualities emerging from encounters such
as empathy, compassion, or trust.
Good patient care is found not on a computer screen but in being truly present with patients53. The social
sciences theory and ethnographic methods used to develop our ontology achieve a broader reach and greater
usability towards developing “good patient care” than existing frameworks discussed in Section 2. Our
ontology does not solely focus on providers; it suggests that many interactions in healthcare meaningfully
modify patient-provider interactions. These warrant further exploration. By modeling both providers and
patients as individuals with specific characteristics, the model can help elucidate if providers’ individual
characteristics (gender, race, age, personality, etc.) and congruence with patients’ identities can influence
relational experiences like empathy or outcomes like patient satisfaction or clinician burnout54.
Our study uses relational ethnography to refine the ontology engineering process in order to bring greater
specificity and accuracy to our model of subjective healthcare experiences. Recent research in bioinformatics
has favored automated observational methods to study the clinical environment. For example, one study
suggests incorporating automated techniques like simulation models or tracking sensors to tag healthcare
providers and to visualize and analyze clinical workflow in an emergency room55. However, automated
methods alone miss key features of interactions that render meaning, especially those that are not visible or
tangible (such as feelings, qualities, or cognitive models). Space is not the only variable for interactional
content. Serious clinical encounters can involve social talk, humor, or self-disclosure—thus appearing less
“professional”—while casual hallway conversations can highlight qualities of professionalism that may belie
the label of “casual.” Thus, using only narrow ethnographic or spatial data cannot provide the breadth that a
relational approach allows.
We acknowledge the trade-offs of using a relational approach for ontology engineering: since we seek a
wider framework encompassing all possible modifiers of presence in clinical interactions, some of our
categories are not modeled with enough specificity to provide depth into a specific subcomponent. However,
this attempt is the first of its kind. Future work should combine our conceptual breadth with greater depth.
5.1 Study Limitations
Although observations of physician-patient interactions informed the ontology, we did not interview patients
or elicit their feedback about the ontology. Future work should engage this important community of
stakeholders in ontology validation and applications. Similarly, while we envision that our ontology will be
applicable to a range of healthcare providers (such as nurses, social workers, respiratory therapists, etc.), the
providers on our research team are currently limited to physicians, and physicians author the majority of the
domain literature. In future work, we hope to collaborate with a broader array of providers. Finally, despite
the diversity of backgrounds on our team, the ontology may not capture all aspects of presence related to
health equity. We plan a follow-up study to evaluate the Presence Ontology with a racial justice lens, given
growing evidence of structural racism and inequitable care in healthcare systems and informatics56.
5.2 Future Directions
We envision the Presence Ontology will be useful to a broad range of end-users, including providers, patients,
their families and friends, as well as those who manage, design, or work within healthcare systems or in non-
!
!
12!
medical fields where connection is analogously important to professional roles and identity. Interactions in
healthcare matter, not only for the subjective experiences of patients and providers but also for measurable
outcomes such as minimization of medical errors, increased efficiency, equity, and population health. Our
ontology can support research on connection in medicine that seeks to make claims about how presence
affects outcomes such as these.
Future work should involve gathering more information about presence in the real world in order to refine
the Presence Ontology and adapt it to develop research and/or clinical tools aimed at improving healthcare:
1. Ethnographic Data Mining: Patient-provider videotaped encounters or transcripts of audiotaped
encounters could generate useful and novel data and meta-data about presence in healthcare. The
Presence Ontology could be used to annotate transcripts or create a codebook for real-time
ethnographic analysis that could be analyzed using machine learning methods, for example.
2. Documentation of Presence: The ontology could offer precision and specificity to scene analysis
methods using ambient intelligence (combining artificial intelligence and contactless sensors) to
assess metaphorical “dark spaces” in medicine and explore the interplay between environment and
health behaviors57.
3. Clinical Instruments for Presence: While numerous clinical instruments exist for rating aspects of
presence such as empathy, burnout, or patient satisfaction, a unified clinical instrument could
improve the uptake and measure the success of high-yield, teachable behaviors to improve
connection (such as the Presence Center’s five recommendations to enhance presence9). Such an
instrument could be used to educate trainees in the art of connection and adapted into a checklist to
empower patients about aspects of presence that they should expect in encounters.
6 CONCLUSION
We have demonstrated a novel classification of the subjective domain of human experience using an
ethnographic approach to ontology engineering. Our Presence Ontology synthesizes multiple forms of data
and uses relational ethnography to model connection at a high level of abstraction and with clarity. The
Presence Ontology focuses on the interpersonal dynamics among providers, patients, and families and
friends; the factors that may influence these interactions; and the outcomes they generate. As a result, our
conceptual model may have broader reach and greater usability than existing frameworks. A model for
presence and future applications of our ontology can offer shared agendas and support novel informatics
applications to improve human connection in healthcare.
!
!
13!
AUTHOR CONTRIBUTIONS
A.M. conducted the pilot ethnographic study to observe clinical encounters, conducted the broad literature
survey, and decided on concepts and relations in the Presence Ontology. M.K. formalized the Presence
Ontology in OWL using the Protégé Ontology Editor and published it on BioPortal. Along with A.M., M.M.
developed the ontology using anthropological methods. M.M. and A.V. supervised A.M. in iterative ontology
development. M.H. led the systematic review of interpersonal interventions geared toward improving
presence. C.B. led the ethnographic study of 27 patient-physician encounters and contributed toward
development of upper-level concepts in the ontology. R.S. led the qualitative study of trans-disciplinary
professional insights about interpersonal connection. D.Z. led the Stanford Presence 5 research team. A.V.
and S.T provided insight to refine ontology concepts. A.M. and M.K. prepared the initial manuscript draft.
All authors were involved in critical revisions, editing, and manuscript approval.
ACKNOWLEDGMENTS
The authors acknowledge Michelle B. Bass, PhD, MSI, Population Research Librarian, Stanford School of
Medicine, for her coordination of the literature review. Stanford University’s Institutional Review Board
approved human subjects research led by Maitra (Protocol No. 30711) and led by Brown-Johnson (Protocol
No. 42397). We thank Dr. Sheila Lahijani for her helpful input.
DATA AVAILABILITY
The Presence Ontology is made available via the BioPortal repository of biomedical ontologies at
http://bioportal.bioontology.org/ontologies/PREO.
COMPETING INTERESTS
We have no disclosures to report.
FUNDING
This work was supported by the Arthur Vining Davis Foundation, a grant from the Gordon & Betty Moore
Foundation (#6382, Zulman & Verghese, PIs), and by grant R01 GM121724 from the U.S. National Institute
of General Medical Sciences.
REFERENCES
1. Verghese A, Brady E, Kapur CC, Horwitz RI. The Bedside Evaluation: Ritual and Reason. Ann
Intern Med. 2011;155(8):550-553. doi:10.7326/0003-4819-155-8-201110180-00013
2. Verghese A. Culture shock - Patient as icon, icon as patient. N Engl J Med. 2008;359(26):2748-
2751. doi:10.1056/NEJMp0807461
3. Kleinman A. Caregiving: the odyssey of becoming more human. Lancet. 2009;373(9660):292-293.
doi:10.1016/S0140-6736(09)60087-8
4. Kerasidou A, Horn R. Making space for empathy: Supporting doctors in the emotional labour of
clinical care Ethics in Clinical Practice. BMC Med Ethics. 2016;17(1):8. doi:10.1186/s12910-016-
0091-7
5. Smith CK, Polis E, Hadac RR. Characteristics of the initial medical interview associated with
patient satisfaction and understanding. J Fam Pract. 1981;12(2):283-288.
6. Verghese A, Shah NH, Harrington RA. What this computer needs is a physician humanism and
artificial intelligence. JAMA - J Am Med Assoc. 2018;319(1):19-20. doi:10.1001/jama.2017.19198
7. Martin SA, Sinsky CA. The map is not the territory: medical records and 21st century practice.
Lancet. 2016;388(10055):2053-2056. doi:10.1016/S0140-6736(16)00338-X
!
!
14!
8. Brown-Johnson C, Schwartz R, Maitra A, et al. What is clinician presence? A qualitative interview
study comparing physician and non-physician insights about practices of human connection. BMJ
Open. 2019;9(11):e030831. doi:10.1136/bmjopen-2019-030831
9. Zulman DM, Haverfield MC, Shaw JG, et al. Practices to Foster Physician Presence and
Connection with Patients in the Clinical Encounter. JAMA - J Am Med Assoc. 2020;323(1):70-81.
doi:10.1001/jama.2019.19003
10. Wood JT. Interpersonal Communication: Everyday Encounters, 8th Edition - 9781285445830 -
Cengage.
11. Hersh WR. Medical InformaticsImproving Health Care Through Information. JAMA.
2002;288(16):1955-1958. doi:10.1001/jama.288.16.1955
12. Waitzkin H, Britt T. A critical theory of medical discourse: How patients and health professionals
deal with social problems. Int J Heal Serv. 1989;19(4):577-597. doi:10.2190/L84U-N4MQ-9YAC-
D4PP
13. Presson AP, Zhang C, Abtahi AM, Kean J, Hung M, Tyser AR. Psychometric properties of the
Press Ganey® Outpatient Medical Practice Survey. Health Qual Life Outcomes. 2017;15(1):1-7.
doi:10.1186/s12955-017-0610-3
14. Gardner RL, Cooper E, Haskell J, et al. Physician stress and burnout: the impact of health
information technology. J Am Med Informatics Assoc. 2019;26(2):106-114.
doi:10.1093/jamia/ocy145
15. Bakken S. Can informatics innovation help mitigate clinician burnout? J Am Med Informatics
Assoc. 2019;26(2):93-94. doi:10.1093/jamia/ocy186
16. Israni ST, Verghese A. Humanizing Artificial Intelligence. JAMA - J Am Med Assoc.
2019;321(1):29-30. doi:10.1001/jama.2018.19398
17. Kohane IS. Bioinformatics and Clinical Informatics: The Imperative to Collaborate. J Am Med
Informatics Assoc. 2000;7(5):512-516. doi:10.1136/jamia.2000.0070512
18. Donnelly K. {SNOMED-CT}: The advanced terminology and coding system for eHealth. Stud
Health Technol Inform. 2006;121:279-290.
19. Unertl KM, Schaefbauer CL, Campbell TR, et al. Integrating community-based participatory
research and informatics approaches to improve the engagement and health of underserved
populations. J Am Med Informatics Assoc. 2015;23(1):60-73. doi:10.1093/jamia/ocv094
20. Unertl KM, Novak LL, Johnson KB, Lorenzi NM. Traversing the many paths of workflow
research: Developing a conceptual framework of workflow terminology through a systematic
literature review. J Am Med Informatics Assoc. 2010;17(3):265-273.
doi:10.1136/jamia.2010.004333
21. Forsythe DE, Buchanan BG. Knowledge Acquisition for Expert Systems: Some Pitfalls and
Suggestions. IEEE Trans Syst Man Cybern. 1989;19(3):435-442. doi:10.1109/21.31050
22. Forsythe DE. Engineering Knowledge: The Construction of Knowledge in Artificial Intelligence.
Soc Stud Sci. 1993;23(3):445-477. doi:10.1177/0306312793023003002
23. Belkin NJ, Brooks HM, Daniels PJ. Knowledge elicitation using discourse analysis. Int J Man
Mach Stud. 1987;27(2):127-144. doi:10.1016/S0020-7373(87)80047-0
24. Ventres WB, Frankel RM. Shared presence in physician-patient communication: A graphic
representation. Fam Syst Heal. 2015;33(3):270-279. doi:10.1037/fsh0000123
25. Larson EB, Yao X. Clinical empathy as emotional labor in the patient-physician relationship. J Am
Med Assoc. 2005;293(9):1100-1106. doi:10.1001/jama.293.9.1100
26. Hochschild AR. The Commercialization of Intimate Life: Notes from Home and Work. Berkeley:
Univ of California Press, 2003; 2003.
27. PubMed. https://pubmed.ncbi.nlm.nih.gov/.
28. Frank JD, Frank JB. Persuasion and Healing: A Comparative Study of Psychotherapy. JHU Press;
1993.
29. Ainsworth-Vaughn N. Claiming Power in Doctor-Patient Talk. Oxford University Press on
Demand; 1998.
30. Foucault M. The Birth of the Clinic. Routledge; 2012.
31. Jain SL. Malignant: How Cancer Becomes Us. Univ of California Press; 2013.
32. Mol A. The Logic of Care: Health and the Problem of Patient Choice. Routledge; 2008.
33. KLEINMAN A, EISENBERG L, GOOD B. Culture, Illness, and Care. Ann Intern Med.
1978;88(2):251-258. doi:10.7326/0003-4819-88-2-251
!
!
15!
34. Forsythe D. Studying Those Who Study Us: An Anthropologist in the World of Artificial
Intelligence. Stanford University Press; 2001.
35. Ozdalga E, Ozdalga A, Ahuja N. The Smartphone in Medicine: A Review of Current and Potential
Use Among Physicians and Students. J Med Internet Res. 2012;14(5):e128. doi:10.2196/jmir.1994
36. Haverfield MC, Tierney A, Schwartz R, et al. Can Patient–Provider Interpersonal Interventions
Achieve the Quadruple Aim of Healthcare? A Systematic Review. J Gen Intern Med.
2020;35(7):2107-2117. doi:10.1007/s11606-019-05525-2
37. Embase. https://www.embase.com.
38. PsycINFO. https://www.apa.org/pubs/databases/psycinfo.
39. Sikka R, Morath JM, Leape L. The quadruple aim: Care, health, cost and meaning in work. BMJ
Qual Saf. 2015;24(10):608-610. doi:10.1136/bmjqs-2015-004160
40. Desmond M. Relational ethnography. Theory Soc. 2014;43(5):547-579. doi:10.1007/s11186-014-
9232-5
41. Wulf V, Pipek V, Randall D, Rohde M, Schmidt K, Stevens G. Socio-Informatics. Oxford
University Press; 2018.
42. Handwerker PW. Quick Ethnography: A Guide to Rapid Multi-Method Research. Rowman
Altamira; 2001.
43. Schwartz R, Haverfield MC, Brown-Johnson C, et al. Transdisciplinary Strategies for Physician
Wellness: Qualitative Insights from Diverse Fields. J Gen Intern Med. 2019;34(7):1251-1257.
doi:10.1007/s11606-019-04913-y
44. Tudorache T, Noy NF, Tu S, Musen MA. Supporting Collaborative Ontology Development in
Protégé. In: Springer, Berlin, Heidelberg; 2008:17-32. doi:10.1007/978-3-540-88564-1_2
45. Kamdar MR, Tudorache T, Musen MA. A systematic analysis of term reuse and term overlap
across biomedical ontologies. Semant Web. 2017;8(6):853-871. doi:10.3233/SW-160238
46. Hastings J, Ceusters W, Smith B, Mulligan K. The emotion ontology: Enabling interdisciplinary
research in the affective sciences. In: Lecture Notes in Computer Science (Including Subseries
Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol 6967 LNAI.
Springer, Berlin, Heidelberg; 2011:119-123. doi:10.1007/978-3-642-24279-3_14
47. Whetzel PL, Noy NF, Shah NH, et al. BioPortal: Enhanced functionality via new Web services
from the National Center for Biomedical Ontology to access and use ontologies in software
applications. Nucleic Acids Res. 2011;39(SUPPL. 2):W541-W545. doi:10.1093/nar/gkr469
48. Jonquet C, Shah NH, Youn CH, Callendar C, Storey M-A, Musen MA. NCBO Annotator:
Semantic Annotation of Biomedical Data. In: Proceedings of the International Semantic Web
Conference (ISWC), Poster and Demo Session, Vol. 110. ; 2009.
49. Sinclair S, McClement S, Raffin-Bouchal S, et al. Compassion in Health Care: An Empirical
Model. J Pain Symptom Manage. 2016;51(2):193-203. doi:10.1016/j.jpainsymman.2015.10.009
50. Anderson RT, Camacho FT, Balkrishnan R. Willing to wait? The influence of patient wait time on
satisfaction with primary care. BMC Health Serv Res. 2007;7(1):1-5. doi:10.1186/1472-6963-7-31
51. Smith B, Ceusters W. Ontological realism: A methodology for coordinated evolution of scientific
ontologies. Appl Ontol. 2010;5(3-4):139-188. doi:10.3233/AO-2010-0079
52. Kaplan B. Evaluating informatics applications—some alternative approaches: theory, social
interactionism, and call for methodological pluralism. Int J Med Inform. 2001;64(1):39-56.
doi:https://doi.org/10.1016/S1386-5056(01)00184-8
53. Verghese A. The importance of being. Health Aff. 2016;35(10):1924-1927.
doi:10.1377/HLTHAFF.2016.0837
54. Gleichgerrcht E, Decety J. Empathy in Clinical Practice: How Individual Dispositions, Gender, and
Experience Moderate Empathic Concern, Burnout, and Emotional Distress in Physicians. Zalla T,
ed. PLoS One. 2013;8(4):e61526. doi:10.1371/journal.pone.0061526
55. Vankipuram A, Traub S, Patel VL. A method for the analysis and visualization of clinical
workflow in dynamic environments. J Biomed Inform. 2018;79:20-31.
doi:10.1016/j.jbi.2018.01.007
56. Cahan EM, Hernandez-Boussard T, Thadaney-Israni S, Rubin DL. Putting the data before the
algorithm in big data addressing personalized healthcare. npj Digit Med. 2019;2(1):78.
doi:10.1038/s41746-019-0157-2
57. Haque A, Milstein A, Fei-Fei L. Illuminating the dark spaces of healthcare with ambient
intelligence. Nature. 2020;585(7824):193-202. doi:10.1038/s41586-020-2669-y
