This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Lifestyle modification is the most important factor in the management of obesity. It is therefore essential to enhance client participation in voluntary and continuous weight control.
The aim of this study was to develop an obesity management ontology for application in the mobile-device domain. We considered the concepts of client participation in behavioral modification for obesity management and focused on minimizing the amount of information exchange between the application and the database when providing tailored interventions.
An obesity management ontology was developed in seven phases: (1) defining the scope of obesity management, (2) selecting a foundational ontology, (3) extracting the concepts, (4) assigning relationships between these concepts, (5) evaluating representative layers of ontology content, (6) representing the ontology formally with Protégé, and (7) developing a prototype application for obesity management.
Behavioral interventions, dietary advice, and physical activity were proposed as obesity management strategies. The nursing process was selected as a foundation of ontology, representing the obesity management process. We extracted 127 concepts, which included assessment data (eg, sex, body mass index, and waist circumference), inferred data to represent nursing diagnoses and evaluations (eg, degree of and reason for obesity, and success or failure of lifestyle modifications), and implementation (eg, education and advice). The relationship linking concepts were “part of”, “instance of”, “derives of”, “derives into”, “has plan”, “followed by”, and “has intention”. The concepts and relationships were formally represented using Protégé. The evaluation score of the obesity management ontology was 4.5 out of 5. An Android-based obesity management application comprising both agent and client parts was developed.
We have developed an ontology for representing obesity management with the nursing process as a foundation of ontology.
Advancements in socioeconomic status and lifestyle changes in Korea have led to an increasing number of obese people and a consequent increase in the importance of obesity management to prevent illness and promote health among the population [
This study investigated two strategies for empowering clients to effectively manage their obesity: (1) use of mobile devices, which are now an integral part of everyday life in many countries [
Some clients lack knowledge regarding how to manage obesity, and the information provided through mobile devices should be based on the available evidence. The best sources for evidence-based knowledge are published clinical practice guidelines. The content for obesity management should vary with the health behavior and diet patterns of the individual client. The effectiveness of obesity management can be improved if time-stamped behavioral data are collected and tailored coaching is provided remotely, based on published clinical practice guidelines, using mobile devices [
Since these guidelines are written in natural language, it is necessary to translate them into a computer-interpretable format. Ontology is a computer-interpretable knowledge model for formalizing and representing shared concepts in a specific domain of interest. It is considered to be a highly effective way of improving the integration, interoperability, and sharing of data. Moreover, the enhanced reusability of clinical data and evidence-based knowledge support clinical decision making by explicitly defining and delivering semantic concepts in a specific domain [
This project is a part of the Health Avatar Project in Korea, which promotes health and manages health problems using personal data and knowledge in virtual space. The health avatar consists of an agent avatar, which is a representation of the expert knowledge or published knowledge (such as clinical practice guidelines), and a client avatar, which is a representation of personal data from molecular to community levels (Figure 1). The health avatar collects data from the client avatar and provides it with tailored care solutions based on the personal data and knowledge or evidence provided by the agent avatar. The client avatar collects and sends personal data to the agent avatar. Upon receiving that personal data, the agent avatar makes judgments and provides the client avatar with tailored recommendations.
The obesity management application is a type of health avatar. The client avatar can be represented by data such as weight, height, abdominal circumference, physical activity, and diet, while the agent avatar can be represented by the clinical practice guidelines published by the National Institute for Health and Clinical Excellence (NICE). Personal data such as weight and height, and tailored recommendations, alarms, or reminders such as “educate about low-calorie diet” are transferred between the client and agent avatars. The front end serves as a user interface to collect these personal data and display tailored recommendations suggested by the agent avatar.
This study considered two objectives during the development of the obesity management ontology. The first objective was to identify concepts related to client participation in behavioral modification for obesity management, since it is important to identify the factors that motivate and encourage participation in the process. The possibility of using the nursing process as a foundation of ontology was explored, since that process is a patient-focused clinical reasoning method for nursing problems, and each phase of the nursing process emphasizes the active participation of the client [
With this background, a clinical practice guideline-based obesity management ontology was developed by identifying the concepts and relationship between concepts according to the nursing process, using ontology development methods that are used in the biomedical arena. In addition, concepts describing client participation in the application program, which is the most important aspect of obesity management, were identified and reflected in the ontology, and finally, the ontology was evaluated.
An obesity management ontology was developed using the General Formal Ontology method [
To determine the scope of obesity management services for adult clients, a guideline developed by NICE [
We reviewed top-level ontologies such as Basic Formal Ontology and a theoretical framework to determine the most appropriate foundational ontology for obesity management, thus minimizing errors during the creation of a domain-specific ontology [
The concepts needed to assess the client, make a diagnosis, determine client-specific interventions, and evaluate the outcome of those interventions within the scope of obesity management were extracted. These concepts included those used to represent not only personal information but also client-specific interventions from the guidelines based on personal data. This phase considered the concepts needed to enhance the client participation, together with ways to operationalize those concepts.
The extracted concepts were specified as classes, individuals, and relationships between classes and individuals in order to define the obesity management ontology according to the foundational ontology determined in the second phase [
The content of the obesity management ontology was evaluated using scenarios via peer-review by 3 ontology and domain experts [
The extracted concepts and relationships were formalized using Protégé. The feasibility of the representation of the nursing-process ontology based on Protégé was assessed.
A prototype application was developed based on an obesity management ontology incorporating the two key features of client participation and minimizing information exchange between the application and the database.
The clinical practice guidelines on obesity reported by NICE (2006) [
The nursing process met the requirements of promoting client participation and minimizing data transactions between databases and the obesity management application in this study. The nursing process comprises five cyclical phases: assessment, nursing diagnosis, outcome identification, implementation, and evaluation. The nursing process as a framework for nursing practice is a systematic, patient-centered, and goal-oriented method with the following phases: (1) identifying client status and need, (2) making nursing diagnoses based on the available client data, (3) determining the expected outcome for a nursing problem for the client, (4) determining and implementing the best nursing interventions to reach the goal, and (5) evaluating whether the expected outcome has been reached. This framework was selected as a foundation of ontology because it describes well both the obesity management process and the client participation therein. In obesity management, the expected outcome is determined by the clients, the evaluation is performed repeatedly to determine whether the outcome has been reached, and the tailored best intervention is implemented to reach expected outcomes based on the client’s assessment data.
Figure 2 illustrates the nursing process used for obesity management as a foundation of ontology. The nursing process can be subdivided into five phases according to the time taken to access the application when specifying the obesity management ontology. This was represented as a dynamic ontology with a spiral reflecting two cyclical nursing processes that are connected to each other. The blue color in the figure (ie, boxes 1, 2, 3, and 4) indicates the initial process of the obesity management service from the beginning to the end, while the green color (ie, boxes 5 and 6) indicates the repetitive process of obesity management from the beginning of the revisit to the service until a client reaches his or her identified outcome. These two cyclical nursing processes contribute to minimizing the amount of data transactions. The nursing diagnosis inferred using the personal data and evidence-based knowledge in a single, long-interval process can be reused in short-interval periodic processes; therefore, data needed in short-interval, periodic processes would be minimized.
Figure 3 shows the obesity management information, which in part reflects the cyclical characteristics of the service provision and login information, with client information guiding that cyclic process. This initial nursing process ontology was elaborated based on these information characteristics. The model has three levels of category: temporal, nursing process, and analytical. The temporal level was an abstract category corresponding to the obesity management. This was divided into assessment, nursing diagnosis, goal, outcome evaluation, and implementation phases, according to the nursing process. The analytical category shows the data linkage to each phase of the nursing process. In the assessment phase, data are collected from a client, and the nursing diagnosis phase defines the degree of obesity using the data collected in the assessment phase. A target weight loss and the date by which to achieve that goal are set in the goal phase, while the intervention for obesity management is provided in the implementation phase. Finally, the state of the client after the intervention is defined in the outcome evaluation phase by comparing the goal with the reassessment data.
The concepts required to make nursing diagnoses, generate client-specific interventions to enable the individual to reach the goal that they have set, and evaluate outcomes in the obesity management process were extracted. In total, 127 concepts were identified (
The 127 concepts extracted in the previous phase were arranged as classes and individuals depending on the foundational ontology, using the relationship-linking classes listed in
The evaluation scores of the nursing-process-based obesity management ontology for various criteria are presented in
Relationship-linking classes in the obesity management ontology.
Relationship name | Defined at concept |
Part of | Relationship between class and subclasses |
Instance of | Relationship between instances and class |
Derives from | Relationships of nursing diagnosis and evaluation to assessment data |
Derives into | Relationships of assessment data to nursing diagnosis and evaluation |
Has plan | Relationships of nursing diagnosis and evaluation to implementation |
Followed by | Relationships of implementation to nursing diagnosis and evaluation |
Intention | Relationship of nursing diagnosis to goal identification |
Grades of evaluation on the representation layer.
Criterion | Score | ||
|
Mean | Minimum | Maximum |
Match between formal and cognitive semantics | 4.7 | 4 | 5 |
Clarity | 4.0 | 4 | 5 |
Explicitness | 4.3 | 4 | 5 |
Interpretability | 5.0 | 4 | 5 |
Accuracy | 4.7 | 4 | 5 |
Consistency | 5.0 | 5 | 5 |
Comprehensiveness | 4.3 | 4 | 5 |
Granularity | 4.1 | 4 | 4 |
Relevance | 4.8 | 4 | 5 |
Total | 4.5 |
|
|
Figure 4 shows some of the relationships among the classes represented by obesity management using Protégé. The rectangles represent classes and the arrows represent the relationship between the 127 concepts. The color of the arrow reflects the relationship among the classes and individuals. For example, red arrows (n=1) represent the relationship between each individual and class, which is “instance of”; green arrows (n=2) mean the relationship of “part of”; purple arrows (n=3) represent “followed by”, which is the relationship between implementation and evaluation in this figure; and gray arrows (n=4) represent “derives into”, which is the relationship between the assessment data and evaluation.
The prototype application consisted of two parts: an agent avatar and a client avatar as described in the
The Health Avatar Project model.
Cyclical process for obesity management.
Initial obesity management ontology based on the nursing process.
Some of the relationships among classes represented using Protégé.
Screenshot showing an example of a client-specific recommendation.
Ontology is an effective way of representing specific domain knowledge in the field of biomedicine. It is also used for knowledge model development to support the clinical decision making of health care providers [
The nursing process was used as a foundation of ontology to describe the client participation in obesity management. Since the nursing process is a scientific and systematic problem-solving framework that allows the client to participate in their own care [
The client was considered to be the principal agent of goal setting and outcome evaluation in obesity management. The nursing-process–based obesity management ontology developed in this study allows the client to participate in setting a goal, such as a target weight loss and duration to reach to the target weight, and to monitor and evaluate their behavioral modification toward obesity management. To this end, a set of concepts was introduced to describe whether the goal and rate of weight loss were realistic and adequate, and another set was designed to describe whether the obesity management behavior was adequate based on the self-monitoring data.
One of the success factors in the existing Web-based intervention programs is login intervals by clients reflecting client participation [
Another key characteristic of the nursing process is that it comprises cyclical phases. In this study, two cyclical nursing processes were connected to each other based on the sequences of logins to the application in our obesity management ontology: the first and second cycles of the nursing process correspond to the first visit and revisits, respectively. At the first visit, the degree of obesity is inferred based on height, weight, abdominal circumference, sex, and past history, and a goal is set. From the second visit, obesity management is evaluated by comparing current weight with the defined outcome. Based on the login information, which is used to guide the cyclical process of obesity management, if the client is either a first-time user of the application or a previous user who did not meet the continuity criteria, a long-interval, single process will start; otherwise, a short-interval, periodic process starts. This feature minimizes the real-time data traffic.
The obesity management ontology developed in this study was compatible with the Initial Clinical Information Ontology of the Danish General Electronic Patient Journal Conceptual Model [
Concepts extracted based on the nursing process are specified as classes and individuals. The relationships between these concepts were extracted from the OBO [
The obesity management ontology developed herein was evaluated at a representation layer of the basic internal layers from various dimensions of the ontology [
This study was subject to the following limitations. First, the developed ontology cannot be used for other types of nursing problems because the concepts used in its development were extracted specifically from obesity management guidelines. We recommend further development of this ontology for other types of nursing problems to determine whether the nursing process can be used as a foundation of ontology to represent nursing care in general. Second, the effectiveness of the prototype Android application developed in this study was not tested. The effectiveness of the application will be established in another study in the near future.
Some of the concepts extracted from the guidelines.
National Institute for Health and Clinical Excellence
Open Biological and Biomedical Ontologies
Systematized Nomenclature of Medicine Clinical Terms
Extensible Markup Language
This work was supported by the National Research Foundation of Korea (NRF) grants funded by the Korean government (MSIP) (Nos. 2012-012257 and 2010-0028631).
None declared.