The ontology was developed with the collaboration of 40 highly knowledgeable individuals from 13 companies located in 3 European countries, using a 5-phase methodology: (1) Elicitation of knowledge through systematic literature review and expert interviews (4 experts involved), (2) Conceptualization of a preliminary conceptual map, (3) Co-design of a refined map through 5 focus groups (23 experts involved), (4) Development of the ontology, and (5) Validation of the ontology (13 experts involved). Figure 1 shows the phases and their activities and outputs. The phases are described in the following paragraphs.

The ontology was developed by applying a knowledge-based approach since there is a lack of datasets about the knowledge domain. Indeed, no available dataset was retrieved, except for a few covering peculiar aspects of the US MD market, such as the one on “Manufacturer and User Facility Device Experience” [18]. These datasets do not cover the areas of interest in our research and do not address the European market. The fundamental knowledge-based ontology engineering activities [19] have been deeply rooted in a collaborative approach to foster the acquisition of relevant information and result in more reliable ontological models [20-22]. This adaptation was necessary due to the complexity and multidisciplinary nature of the knowledge to be conceptualized [23,24]. Specifically, in this study, the authors validated the effectiveness of using focus groups as a tool for knowledge acquisition and conceptualization phases in ontology engineering [25].

In the first phase, the knowledge relevant to the domain was elicited through a literature review and expert interviews, and the authors also analyzed any available DSS used by those responsible for deciding whether and how to introduce MDs into their organizations.

The literature review was guided by the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement approach [26] and aimed at investigating the existing conceptualizations of MDs. In particular, the literature review aims to answer the following questions (Qs):

Q1. What types of conceptualizations are adopted?

Q2. What are the purposes of the conceptualizations?

Q3. What are the intended users of the conceptualizations?

Q4. What is the domain of knowledge conceptualized?

Q4a. What are the main conceptual items addressed?

Q4b. Which knowledge sources (eg, standards, regulations, ontologies, and so on) guide the conceptualizations?

Q5. Have the identified conceptualizations been adopted in any applications?

Q6. Which EUnetHTA assessment elements are covered by the conceptualizations?

Q6 referred to the EUnetHTA [27], as the European Union has recognized it as the official model for HTA reports. The EUnetHTA suggests a list of Assessment Elements (ie, factors, concerns, objectives, or preferences relevant to the decision problem) that should be considered when assessing MDs. Q6 was included to check whether existing conceptualizations make accessible the knowledge that stakeholders should look for when assessing MDs’ value.

We searched Clarivate ISI Web of Science, Scopus, PubMed, Cochrane Database, and ProQuest for English-language journal articles, reviews, book chapters, and conference proceedings published after 2001 in “Computer science,” “Engineering,” “Decision sciences,” “Medicine,” “Business, Management and Accounting,” and “Multidisciplinary” subject areas. The resulting query searched for (ontolog* OR taxonom* OR thesaur* OR conceptualization* OR “conceptual map” OR “conceptual model” OR “decision support system” OR “decision-support system” OR dss OR “knowledge-based system” OR “expert system”) AND (medical AND devic*) in the “Title,” “Keywords,” and “Abstract” fields. The search was conducted in April 2024 and led to the analysis of 20 papers. In April 2026, the search query was rerun to verify whether relevant papers had been published in the meantime. Three additional relevant papers were found [28-30]. These works were not available when the initial conceptual map was developed during Phase 2; however, for completeness, they are analyzed and discussed in the “Results” section together with the other 20 works retrieved in April 2024. It is necessary to note that, although pertinent to the topic, these additional contributions do not address the gaps identified in April 2024, thus maintaining the relevance of this study. Please see Checklist 1 for further details regarding the PRISMA protocol we followed in this study.

Four participants (2 health care managers and 2 HTA analysts) from 2 leading hospitals in Italy were recruited for interviews using a purposive sampling strategy based on their expertise and direct involvement in MD adoption decisions. Additional details on the participants are reported in the “Results” section. Participants were identified through the authors’ professional networks.

Semistructured interviews were conducted, each lasting approximately 45 minutes. The interviews aimed at identifying the tools and sources experts usually use in MD’s adoption decisions and the problems they may face. Example prompts included (1) who is responsible for deciding which MDs to introduce, (2) which criteria are used when selecting and comparing MDs, (3) how the evaluation process is typically conducted, (4) what challenges are encountered in the selection and comparison of MDs, and (5) what type of information is required to support decision-makers and how a DSS could support this process.

Please see Checklist 2 for further details regarding the protocol for qualitative research we followed in this study.

The second phase aimed to draft the first version of a conceptual map [31], leveraging the knowledge obtained in Phase 1. The literature review and expert interviews were jointly used to identify relevant gaps that the ontology should address, to gain insights into how the decision-making problem is currently solved, and to identify and consolidate the main conceptual elements of the domain, including entities, attributes, and relationships relevant to MD adoption. The synthesis followed an iterative and incremental process. First, key concepts and relationships emerging from the literature were extracted and organized into preliminary conceptual elements. These were then refined and complemented using insights from expert interviews, which provided contextual validation and highlighted practical requirements and decision-making needs.

The resulting knowledge was formalized into a conceptual map, a structured glossary, and a set of competency questions—that is, natural language questions defining and constraining the scope of the knowledge represented in the ontology [32]. The conceptual map resulting from Phase 2 was implemented using Draw.io (Gaudenz Alder) and is presented in the “Results” section.

The conceptual map resulting from the previous phase was adopted as the basis for the third phase, during which focus groups with domain experts were organized. Focus groups have proven effective in “co-designing” activities to elicit expert knowledge [33,34].

Domain experts were recruited through the authors’ professional networks using a purposive sampling strategy based on their roles and their direct involvement in MD adoption processes. Participation was voluntary, and experts were included to ensure a balanced representation of relevant stakeholders involved in the decision-making process. The experts who participated in the focus groups were health care managers, clinicians, HTA analysts, or MD manufacturers (ie, people involved in MD design, development, and sale). These latter were included because they have a comprehensive knowledge of the MDs and the sources (eg, databases) used for developing and selling MDs. Additional details on the participants are reported in the “Results” section.

Focus groups were conducted using a semistructured approach guided by the competency questions developed in Phase 2. During the sessions, participants were asked to comment on a populated version of the conceptual map based on a real MD and to identify missing elements or relationships. Each session lasted approximately 1 hour. Focus groups were conducted iteratively until no substantively new conceptual insights, missing conceptual elements, or relevant refinements to relationships emerged from focus group sessions.

All focus group sessions were recorded and transcribed verbatim. The transcripts were analyzed using thematic synthesis. Two authors independently coded the data, and discrepancies were resolved through discussion to reach consensus. Codes were iteratively grouped into higher-level themes, which were used to refine and extend the conceptual map, thereby improving the granularity (ie, the map allows the identification of the main relevant conceptual items and the relationships), instantiatability (ie, the map could be instantiated), and conciseness and completeness (ie, the map avoids any superfluous conceptual items and encompasses all essential ones) of the map [31,35]. The map resulting from this Phase is presented in the “Results” section.

The map resulting from the co-design phase was implemented into an ontology using the Protégé ontology editor [36]. Resource Description Framework [37] and Web Ontology Language [38] were adopted as ontological languages, with rules written in Semantic Web Rule Language [39]. The use of the Protégé ontology editor, together with W3C-endorsed languages for the semantic web, has become a common practice in biomedical and health care research, enabling semantic modeling, data integration, and knowledge-based decision support [40-42]. The prototypical ontology is described in the “Results” section and is available in Multimedia Appendix 1.

The prototypical ontology developed in the previous phase was populated with data from 6 real MDs. Thirteen experts were then interviewed to assess whether the ontology met the information needs they typically have when comparing, evaluating, or introducing MDs. During the interviews, one of the authors presented examples, drawn from the 6 MDs used to populate the ontology, illustrating the knowledge represented in it. Interviewees could ask questions or comment throughout the interview, and afterward were invited to provide feedback and complete a questionnaire. Each interview lasted about 30 minutes. The questionnaire consisted of 13 Likert-scale items adapted from the subscales composing the Technology Acceptance Model 3 [43], aimed at assessing perceived usefulness, relevance, and intention to use the DSS that may be developed based on the ontology. The items were adapted to reflect the specific context of MD adoption and ontology-based decision support. In addition, 2 open-ended questions were included to capture perceived strengths and limitations of the ontology. Quantitative questionnaire responses were analyzed using descriptive statistics, while qualitative ones were analyzed through thematic coding to identify recurring strengths and limitations of the ontology. The evaluation metrics used to validate the ontology and the results of this phase are reported in the “Results” section.

The research was conducted in accordance with the Declaration of Helsinki and followed Good Clinical Practice guidelines. The study was performed within the Next Generation EU project ECS00000017 “Ecosistema dell’Innovazione” Tuscany Health Ecosystem (THE, PNRR, Spoke 9: Robotics and Automation for Health) and the project funded by NextGenerationEU-“Age-It-Aging well in an aging society” project (PE0000015), National Recovery and Resilience Plan (NRRP) -PE8-Mission4, C2, Intervention 1.3. This study did not involve clinical trials, the processing of sensitive patient data, medical record consultation, or sample archiving. According to national and regional regulations (Decreto Legislativo n. 211/2003, Legge Regione Toscana n.40/2005, Legge Regione Toscana n. 84/2015), ethics committee approval is not required for this type of research. Oral informed consent for meeting recording and the use of anonymized, aggregated data were obtained from all participants at the beginning of each session. All meetings were anonymized, and no identifying information was collected or stored. No identifiable features or personal data of participants are included in this publication. Data collection, processing, and management complied fully with national and European data protection regulations (EU Regulation Number 679/2016 and Decreto Legislativo Number 101/2018). Participants received no compensation or incentives for their participation.

A collaborative approach was adopted to engineer the ontology [19]. Table 1 provides details about the experts supporting the development of the ontology, their role, and the Phase (1, 3, or 4) they were involved in. It is worth noticing that the experts involved in the ontology creation were not involved in the ontology validation to ensure the independence and objectivity of the evaluation process.

The knowledge elicitation phase includes 3 main activities: a systematic literature review, the analysis of existing DSSs on MDs, and expert interviews. This section presents the results of these activities, as well as the knowledge emerging from them.

Figure 2 shows the PRISMA flow that the authors followed in the systematic literature review. Excluding duplicate works retrieved from more than one source, 2347 articles were retrieved from databases. The remaining works were scanned in their “Title” and “Abstract” fields to ensure consistency with the review aims; 57 works were retained. Works that could not be retrieved in their complete form were removed. Then, in the “Inclusion” phase, 2 authors read the screened works individually, and disagreements were resolved by a third author. Studies were excluded from the sample if they met one or more of the following exclusion criteria:

The total number of scientific works selected in this way is 19.

Finally, in accordance with all researchers, one article from Dogmus et al [44] was included by looking at the articles referenced in the included papers. In the end, 20 papers were eligible.

The literature review results are presented in tabular form in Tables S1-S3 in Multimedia Appendix 2 and are briefly commented on in the text. Taxonomies and ontologies are the most frequently used types of conceptualizations (Q1), and only one study presents a conceptual map. Tables S1 and S2 provide an analysis of the literature concerning questions Q2, Q3, Q4, Q4a, Q4b, and Q5: Table S1 focuses on taxonomies, while Table S2 examines ontologies and conceptual maps. The reason behind this split relies on the fact that taxonomies and ontologies (and conceptual maps) are characterized by different granularities [31]. Therefore, they can be analyzed using different conceptual items: first-order categories for taxonomies are investigated in Table S1, while first-order classes with properties are investigated for ontologies and conceptual maps in Table S2. Each row of Tables S1 and S2 represents a study, and each column corresponds to one of the questions the literature review aims to answer (refer to the “Methods” section).

Looking at Tables S1 and S2, it can be noticed that conceptualizations in literature are characterized by different purposes (Q2) based on their intended users (Q3). They can support MD manufacturers in designing better MDs, clinicians in identifying MDs better fitting their needs, health care managers in improving the services they are responsible for, policymakers and HTA analysts in performing HTA assessments of different MDs, and patients and caregivers in choosing the best MD. Surprisingly, the intended user is sometimes not mentioned in the conceptualizations reviewed. The domain addressed (Q4) typically consists of a few types of MDs sharing some key commonalities (eg, wearable or portable devices). Only 4 conceptualizations consider the MDs’ domain as a whole. The conceptual items (Q4a) considered by the included works are various. Several properties are typically used to characterize the MD, such as the product code, lot number, nomenclature codes, manufacturer name, and MDs’ hardware and software components. These conceptual items are sometimes identified by resorting to existing public sources (Q4b; eg, European Medical Device Nomenclature [EMDN]). Other conceptual items include the resources needed to use an MD in a service, the potential users of the MD (mainly expressed in terms of targeted medical conditions, user impairments, or activity limitations), the stakeholders (other than the end user) who an MD could potentially impact, the health service within which the MD can be used, the location of use, the roles and tasks of those operating the MD, and the expected outcomes associated with the MD’s use. Finally, it has been proposed to provide links to scientific publications to help stakeholders access expert knowledge on MDs. Only 4 studies describe an application (Q5) leveraging the conceptualization presented. Half explicitly reference the conceptualizations’ role as a backbone for an application layer, and the other half only partially hypothesize or describe the application. The limited applicability of the works is also witnessed by the fact that, in only one case, there is a second work [44] that improves the application proposed by another [45].

Table S3 provides the information needed to answer Q6. The EUnetHTA outlines a comprehensive list of Assessment Elements, each designated by an alphanumeric code, representing a specific combination of Domain (ie, a broad category that groups assessment elements), Topic (subjects that characterize assessment elements), and Issue (ie, questions that need to be answered to evaluate the MD with regard to the assessment element). The EUnetHTA categorizes assessment elements into core (ie, the ones that should be included in all evaluations) and noncore ones and identifies 4 models (respectively for medical and surgical interventions, diagnostic technologies, screening technologies, and pharmaceuticals) that could differ in terms of core assessment elements. For each assessment element (corresponding to different rows in Table S3), the table provides its domain, topic, ID, issue, and a list of studies presenting conceptualization that have addressed this assessment element (even if they do not explicitly mention the EUnetHTA). Most of the assessment elements in the EUnetHTA (85%) have not been addressed in the reviewed conceptualizations. The selected conceptualizations mainly focus on identifying MD alternatives (B0001, 9 papers), describing the health conditions and populations (A0001, 8 papers), and describing the consequences and burden of the diseases targeted by the MD targets (A0009, 4 papers).

Our review highlighted that, although numerous DSS exist to assist various types of clinical decision-making [46,47], none have been specifically developed to support the evaluation of MDs and their introduction into health care organizations.

Currently, such evaluations can rely on a few databases, such as the European Database on Medical Devices (EUDAMED) [48], the “Public Access to Device Registration” (a database managed by the Medicines and Healthcare products Regulatory Agency on the MDs registered in the United Kingdom) [49], and the Global Unique Device Identification Database (a database managed by the FDA on MDs marketed in the United States) [50].

Interviews highlighted the need for a DSS to support MD introduction decisions and identified key stakeholder challenges. Interviews underscored that accessing expert knowledge could be essential to inform stakeholders in solving the decision problem concerning all the MDs for which proper HTA evaluations are not performed. Indeed, interviewees confirmed that, since HTA evaluations are time-consuming, they are typically reserved for MDs with significant expected clinical and economic impacts (ie, fewer than 5% of total purchase requests), while most MDs undergo less rigorous evaluations based on accessible available knowledge.

One of the main concerns raised by interviewees was the inadequacy of available knowledge-based apps and databases for solving the decision problem. For instance, interviewee INT3-HTA noted that EUDAMED provided only basic information and often contained incomplete records, limiting its usefulness. In addition, interviewee KE1-Manager1 highlighted the lack of detailed information on available databases on the organizational impact of MDs, including redistribution of staff tasks and modifications in health care service delivery processes. Moreover, interviewee KE1-Manager1 emphasized the need for a DSS that can assist them in comparing MDs that address the same medical condition and enable the provision of the same health service: “while clinicians often have specific MDs in mind when submitting purchase requests, for me it is still of paramount importance to evaluate alternative MDs.”

Furthermore, the interviews highlighted the relevance of the EUnetHTA while also emphasizing the challenges associated with using existing HTA reports developed under this framework. Indeed, in all the health care organizations where interviewees work, the EUnetHTA serves as a reference for conducting HTA assessments. However, as noted by interviewee KE2-HTA, accessing and using existing reports can be time-consuming and difficult, since they are produced by various national and international agencies, each following different reporting procedures.

Finally, it is worth noticing that none of the interviewees mentioned databases other than EUDAMED.

Neither the studies available in the literature nor the existing DSSs and databases adequately address the informational needs of stakeholders involved in the evaluation and selection of MDs.

Only a limited number of conceptualizations are available in the literature, and, to the best of the authors’ knowledge, only the EUDAMED database [48] considers the MD domain as a whole. Moreover, several conceptual elements that experts identified as essential to support decision-making during the interviews are either missing or only partially addressed. For example, the identification of medical conditions in which an MD can be used represents the starting point of most decision-making problems, as highlighted by both the literature [10] and the experts interviewed. Yet, to the best of the authors’ knowledge, this information is not included in existing DSSs and appears in only a few studies [51-53], mainly without reference to internationally recognized classifications.

Furthermore, experts emphasized the importance of assessing the overall impact of MDs, and the literature indicates that the effects of introducing an MD may vary considerably across different health services [54,55]. Nevertheless, only one study specifies the health services an MD can enable [56], and a few more limit themselves to identifying the locations of use rather than detailing the enabled services [51,54,57].

Similarly, identifying the roles and activities of those operating the MDs within health care services would support understanding stakeholders’ responsibilities and needs, as well as assessing whether workflows should be modified or MDs adapted to fit existing processes. However, these aspects are rarely conceptualized [56], and existing DSS or databases do not provide information about the service in which the MD is used or the human resources involved.

Interviewees also emphasized the importance of accessing information related to manufacturers, as they envision using the system to search for MDs produced by specific manufacturers. Nevertheless, the manufacturer is explicitly specified in only 3 existing works [53,58,59], and none adopt unique manufacturer identifiers such as those required by EUDAMED [48].

Moreover, no study suggests using HTA reports as information sources, despite their rich, structured, and publicly available content. Most of the conceptual elements proposed in EUnetHTA, a framework officially recognized by the European Union and reportedly used by the interviewed experts, are not addressed in the reviewed conceptualizations.

Finally, specifying the MD application is also essential for MD evaluation [27], as it allows the identification of appropriate reference evaluation models. However, only one study addresses this aspect [60], and it does so without referring to existing nomenclatures.

The first draft of the conceptual map resulted from the Knowledge elicitation and Conceptualization phases.

Several properties are used to identify the MD: the Trade name with which the MD is sold on the market, the UDI-DI (ie, the unique numeric or alphanumeric code associated with all MDs placed on the European market), the EMDN code (ie, code of the MD according to the EMDN [61]), the Model of the MD, the Risk class of the MD, Scientific publications related to the MD, Technical sheets related to the MD, Patents related to the MD and its components (if any). Moreover, each MD is characterized by its Application (ie, whether it is used for diagnostic or screening reasons or medical and surgical interventions). Finally, each MD’s Type is specified (ie, whether it is a single component, a system, or a procedure pack [2]). The Manufacturer (ie, the company that produces the MD) is described by specifying its Name, the ID the manufacturer has received from the European Union, and the Country where the manufacturer is headquartered. The Medical Condition (ie, health conditions the MD is expected to treat that make it eligible to be used) is expressed as the intersection of the target user’s impairments, disease, and activity limitations—which can be described with the World Health Organization standards International Classification of Functioning, Disability, and Health (ICF) [62] and ICD-10 (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision) [63]. The Health Service (ie, individual services provided to maintain or improve a person’s health, such as medical visits, diagnostic tests, treatments, surgeries, and rehabilitation services) is identified by the Name and 2 Nomenclature codes, one corresponding to the Major Diagnostic Categories and the other to the Diagnosis-Related Groups. The Major Diagnostic Categories represent 25 codes organizing diagnoses for all organs and systems [64], while the Diagnosis-Related Groups were developed to identify the health products a patient can receive from a health service provider [65]. Together, the 2 classifications are used in Italy at a national level for administrative purposes and to organize the different health services that can be provided to patients. For each health service, the Human resources (ie, clinical personnel, caregivers, or patients) performing it and the Setting (ie, the location where the human resources performing the health service are, which can be the clinical setting or the patient’s house) are specified. The representation of the HTA reports on the MD is included in the map. The impacts generated by the introduction of MDs on stakeholders depend on how the health services are designed to deploy them [66]. Therefore, each HTA report is associated with one or more health services involving one or more MDs. One health service involving a given MD could be evaluated in more than one HTA report. Some properties are included in the conceptual map to identify the HTA report: the Link to the HTA report, the ID of the report, the Date on which the report was published, and the report’s Authors. Considering the importance of the EUnetHTA [67,68], only HTA reports produced following it are included. The “Assessment Elements” conceptual item refers to the specific aspects evaluated in the report. Each assessment element is characterized by an ID and consists of the combination of a Domain, a Topic, and an Issue. The “Results” item corresponds to the results for a given assessment element that are provided in the HTA report. The “Method” item describes the method by which the assessment element’s results have been obtained. Starting from the list of assessment elements in the EUnetHTA, the authors included only 23 elements in the map. Table S4 in Multimedia Appendix 2 describes the exclusion criteria that guided the selection of the assessment elements, while Table S5 in Multimedia Appendix 2 lists the assessment elements included in the map. Both tables were discussed with the domain experts who participated in the focus groups.

The conceptual map resulting from the previous phase was adopted as the basis for this phase, during which 5 focus groups with 23 domain experts were organized (refer to Table 1). An improved version of the conceptual map resulted from the Co-design phase. This phase consisted of 5 focus groups where 23 experts were asked to comment on a purposely populated map, referring to a real MD. All participants found that the conceptual items in the map were necessary for representing the domain, but they suggested that some properties should have been added to the map to better describe the MD:

Figure 3 shows the map; ovals correspond to classes, arrows correspond to relationships between classes, and properties are placed on dashed arrows related to the class they refer to. The conceptual items in red were added after Phase 3.

Table 2 shows the glossary of conceptual items included in the conceptual map, while Table 3 reports the list of competency questions. In both tables, the column “First (1) or second (2) draft” specifies whether the conceptual item and competency question were present in the first draft of the map or were added in the second draft, considering the comments received during focus groups.

As elicited from focus groups, the health services in which the MD could be used are gathered from HTA reports concerning the MD. The properties referring to classes in the map could be obtained from HTA reports, scientific publications, or technical sheets concerning the MD and the data on EUDAMED [48] (if available).

During one of the focus groups, a participant (FG3-Clinician2) pointed out that the applications of MD (screening, diagnosis, and medical and surgical interventions) should not be disjointed since the same MD could be used for more than one application. The suggestion demonstrates that focus groups could stimulate a high level of collaboration among experts and ontologists [21,35].

The ontology development phase adopted a modular approach—that is, each of the domains composing the prototype is modeled on a separate ontology [70]. The modules are imported into the general box and are developed using the Resource Description Framework and the Web Ontology Language representations of the conceptual map resulting from Phase 3. Thus, the ontology is organized into 5 modules (HealthService, MedicalCondition, HumanResources, and HTA modules, imported into a general box named MedDev, in which rules are specified). The ontology includes a total of 1432 classes, 32 object properties, 28 data-type properties, 118 individuals, and 64 rules. The ontology foresees the possibility of indicating whether an MD has a previous or subsequent version by reusing the sequence Ontology Design Pattern [71].

The prototypical ontology is available in Multimedia Appendix 1.

Figure 4 shows the ontograph generated using Protégé, which provides a graphical representation of the ontology structure. In the ontograph, nodes represent ontology classes, while arrows denote semantic relationships between them. Solid blue arrows indicate explicitly asserted relationships defined in the ontology, whereas dashed arrows represent inferred relations generated by the reasoning process. Different arrow colors correspond to different object properties connecting the classes. Multimedia Appendix 3 comprises a figure (named “Modules composing the MedDev ontology”) that shows the graphical representation of the modules composing the MedDev ontology. In the figure, arrows indicate that a certain module is imported into another, and black rectangles specify reused ontological resources.

The ontology is instantiated with 6 MDs, whose information can be retrieved from 4 HTA reports. For example, 2 individuals, OTCA17 (Multimedia Appendix 4) and OTCA27 (Multimedia Appendix 5), belonging to the class HTA_Report assess GreenLight-XPS and compare it with Holmium_Laser_Therapeutic_Apparatus. The GreenLight-XPS is a system devoted to Medical_and_surgical_intervention which can be used for the ICD-10 disease N40.1 (benign prostatic hyperplasia with lower urinary tract symptoms) and for the ICF b610 (urinary excretory functions). This MD is adopted in Health_Services that involves MDC 12 “Diseases and disorders of the male reproductive system” and involves DRG 39 “benign prostatic hyperplasia” a health service performed by the human resource Surgeon.

Leveraging the Pellet reasoner [72], the ontology can represent the conceptual map to its full extent, enabling the representation of any MD and its characteristics, as well as its HTA reports (and results) characterizing it.

Furthermore, the semantic model of the ontology can answer the main questions stakeholders might raise when gathering information about MDs (refer to Table 3 reporting the competency questions). For example, leveraging the SPARQL query language, they could ask the ontology to retrieve all the MDs that treat a specific medical condition or ask for the single-use MDs that could be used in a certain health service. Similarly, it is possible to run a SPARQL query to search for specific strings within the HTA reports’ textual results (which are also reported using a specific data-type property; Textbox 1):

To assess whether a DSS based on the proposed ontology can meet the information needs that stakeholders typically have when comparing, evaluating, or introducing MDs, 13 stakeholders from 3 European countries were interviewed and surveyed (refer to Table 1). During the interviews, one of the authors presented examples, taken from the six MDs used to populate the ontology, to illustrate the types of knowledge represented within the proposed ontology. During and after the presentation of the examples, interviewees were invited to share their impressions and comments. Afterward, they were asked to complete a questionnaire composed of 13 Likert-scale questions and 2 open-ended questions, which allowed participants to elaborate on the strengths and weaknesses of the ontology.

Table S6 in Multimedia Appendix 2 reports the quantitative results of the questionnaire. The results show consistently high mean values with low variability, indicating overall positive expert feedback. Participants generally perceived that a DSS based on the proposed ontology may be useful, relevant, and well-aligned with their work tasks. The data also suggests a strong intention to use the system in the future.

Moreover, qualitative feedback was collected both during the presentation of the examples and through the open-ended questions in the questionnaire. Positive feedback primarily highlighted the system’s comprehensiveness, modular structure, interconnectivity, and usefulness in aggregating and linking information from different sources. Negative comments focused on the absence of cost-related information, the need for integration with national or regional databases, and the initial learning effort required to use the system effectively.

Overall, these findings confirm that the ontology can effectively support users in analyzing, selecting, and introducing new MDs within their organizations by satisfying the experts’ informational requirements.

This paragraph provides a nonexhaustive list of examples of how the ontology-based DSS could assist stakeholders in deciding whether and how to introduce MDs in real settings. The relevance of these applications emerged during the interviews and focus groups carried out during this study (Phases 1, 3, and 5).

As highlighted by VAL2-Manager, clinicians might use the DSS to retrieve information about a specific MD they have already selected. By consulting scientific publications, existing HTA evaluations, and available clinical trials on the MD, clinicians can assess whether it meets their needs before proposing it for purchase. Scientific publications may provide insights into how other organizations use the MD, while HTA evaluations and the final or preliminary results of clinical trials can further inform them of its potential impact.

Similarly, health care managers can use the DSS to explore information on selected MDs. They can access data on the medical conditions in which the MD can be used, identify the target population, and possibly realize that introducing the MD could modify care pathways for other conditions beyond those initially considered. Managers can also review technical sheets and patents to understand an MD’s functionality, application, and usage conditions, as well as check its availability, maintenance duration, and spare parts. As specified by VAL1-HTA and VAL3-Manager, such information helps assess organizational readiness, required adaptations, and regulatory compliance. For instance, they might discover that patient training is required or that the MD could be used in multiple health services, optimizing resource allocation.

The DSS also enables comparisons between MDs for a given medical condition. As highlighted by VAL6-Manager, this helps stakeholders identify the most suitable options to improve clinical pathways.

VAL5-Manager noted that managers can also search for MDs from specific manufacturers based on past positive purchase experiences.

Finally, as specified by VAL10-HTA, for MDs with minimal organizational and economic impact, typically not subjected to a comprehensive HTA analysis, the DSS alone might provide sufficient information to support decisions on whether to introduce a given MD.

This study aimed to develop a comprehensive, evidence-based ontology to support the informed adoption of MDs in health care organizations. The authors successfully designed a modular ontology serving as the backbone of a DSS, capturing key elements to inform adoption decisions. It integrates conceptual components such as medical conditions, health services enabled by MDs, professional roles and activities, manufacturers, HTA report data, and MD applications, often absent in existing systems and prior literature (see the “PRISMA Literature Review” section). At the same time, it deliberately excludes components irrelevant to the decision problem (eg, hardware and software infrastructure), achieving a balance between granularity, conciseness, and comprehensiveness. The ontology is grounded in established international sources [61-63] and national nomenclature codes to identify the health service [64,65].

The ontology was developed through a custom knowledge-based methodology that integrates focus groups to elicit knowledge. Our approach builds on existing ontology engineering practices [19] and draws on AgiSCOnt principles [20,21], using conceptual maps to formalize complex, interdisciplinary domains. The ontology’s instantiability was validated by populating it with data from 4 HTA reports, confirming both its robustness and practical applicability. Furthermore, the validation phase demonstrated that the ontology effectively addresses the core information needs of stakeholders.

The ontology presented in this study represents, to the best of our knowledge, the first comprehensive, generalizable, and evidence-based conceptualization developed specifically to support informed adoption decisions for MDs. It fills key gaps in existing literature and DSSs by capturing the full spectrum of stakeholders’ knowledge needs. Only a limited number of existing conceptualizations consider the MD domain holistically, and key elements highlighted by expert interviews, such as the identification of medical conditions for MD use, the roles and activities of health care professionals, service-level impact, manufacturer-specific information, HTA reports, and MD applications, are either missing or incompletely addressed in prior works.

Beyond its conceptual coverage, the robustness of the proposed ontology is rooted in the knowledge-based and collaborative ontology engineering methodology adopted. Given the complexity and multidisciplinary nature of the MD adoption domain, traditional ontology engineering approaches are often insufficient to capture implicit knowledge. By integrating focus groups into the knowledge acquisition and conceptualization phases, this study empirically validates their effectiveness in supporting the cocreation of reliable ontological models, enabling the systematic elicitation and formalization of heterogeneous expert perspectives [25].

Grounded in established sources, the proposed ontology ensures interoperability across health care settings. Moreover, the ontology engineering methodology used by the authors followed and the ontology’s modular design ensure both generalizability and practical applicability across diverse contexts: the ontology’s core conceptual structure captures regulatory, organizational, and assessment processes that are widely shared across health care systems, and modules can be easily refined or extended to align with local practices.

Four key issues remain open for further development of the domain ontology of MDs.

First, most experts involved in this study were from public and private service providers in Italy, and the conceptual item Health Service was modeled using 2 Italian nomenclature codes, as no international standard currently exists. Involving mainly Italian experts in the ontology development does not compromise the conceptual soundness of the ontology at the structural level, and the core classes and relations were defined to capture regulatory, organizational, and assessment processes that are common across European health care systems. However, the use of Italian nomenclatures affects specific implementation-level components (ie, the health service conceptual item). Second, the HTA module is based on the EUnetHTA model (which is the one the European Union has recognized as the official model for HTA reports and is currently the most adopted framework for HTA in Europe), but other relevant models could be used to create HTA reports, such as models developed specifically for hospitals [73]. Third, the effort required to populate the ontology is substantial, and the low quality and limited availability of certain knowledge sources (eg, HTA reports, technical sheets, and so on) could pose a challenge for populating it. Fourth, while the ontology has been validated in terms of conceptual completeness, consistency, and instantiability, its full clinical and financial decision-making value can only be demonstrated through the implementation and pilot testing of a functioning ontology-based DSS.

This paper proposes a validated ontology that aims to support the informed adoption of MDs in health care organizations. While prior literature has mostly focused on partial conceptualizations and existing databases cover only limited aspects of MDs, the proposed ontology fills this gap by providing a comprehensive conceptualization that addresses all the key information needs of stakeholders involved in the decision problem. The ontology’s ability to capture this full spectrum of information stems from the custom, collaborative, knowledge-based ontology engineering methodology adopted in its development. This methodological choice enabled the elicitation and integration of diverse forms of expert knowledge, ensuring that the resulting conceptual model is both robust and closely aligned with stakeholders’ needs. Thus, this study demonstrates the value of cocreating conceptualizations with stakeholders when dealing with complex and multidisciplinary domains and provides useful guidance for future efforts aimed at conceptualizing complex domains characterized by multiple stakeholders with heterogeneous information needs. This collaborative methodology supports the abstraction of locally elicited expertise into higher-level conceptual models, thereby enhancing the transferability of the resulting ontology beyond the specific national context in which it was developed.

From a managerial perspective, the ontology provides a structured, knowledge-driven foundation to support decision-making in the adoption of MDs. By integrating heterogeneous sources of explicit and implicit knowledge, it enables stakeholders to systematically compare MDs and select those best suited to their specific contexts. The ontology can guide structured evaluation and integration processes, ensuring that decisions are evidence-informed, transparent, and reproducible. A promising and still largely unexplored field of application could be oral and dental health care [74], a domain characterized by rapid innovation that requires informed and well-structured adoption processes. This approach can help health care organizations streamline MDs’ introduction, enhance clinical workflows, and reduce the underuse of innovations. The modular design allows managers to adapt components, such as health service nomenclatures or HTA modules, to local practices without altering the overarching conceptual structure, facilitating implementation across diverse health care systems. The purpose of this study is not to provide a system to be officially adopted at the European level, nor does it have the institutional mandate or resources to do so. Rather, its managerial value lies in illustrating how to rationalize expert knowledge and identify the key informational components necessary to support informed MD adoption decisions, thus offering a conceptual foundation for future research and development initiatives at the European level.

In light of the identified limitations and challenges, this study represents a call for the development of international projects aimed at establishing shared nomenclatures and replicating the knowledge acquisition process with stakeholders from other European health care systems. The ontology’s modular design allows the HTA module, based on the EUnetHTA framework, to be replaced with alternative HTA models if needed, and the HumanResources module to be adapted to other health care systems’ nomenclatures without affecting the overall structure. This modularity ensures that updating and adapting the ontology remains straightforward. Future efforts may also focus on ontology population, which could be automated through ontology learning frameworks, provided that datasets such as EUDAMED or other sources become publicly available. In this regard, the EU Regulation 2021/2282 on HTA may facilitate the creation of higher-quality knowledge sources, supporting ontology population. Finally, integrating the ontology into a DSS and evaluating its clinical and economic impact via real-world pilot studies involving multiple health care organizations is essential for full validation. Engaging experts and stakeholders from different countries in these pilots will further verify the ontology’s applicability across diverse health care contexts.