Sage Journals: Discover world-class research

Abstract

Objective

Exploring complex of digital health interventions (DHI), use cases, and applications in healthcare. Applications include integrated care architectures, mobile health, clinical decision support systems, and risk stratification. Thematic analysis aims to provide context for assurance methodologies aligned to DHIs. Use cases, patient outcomes, stakeholders, digital health technologies utilized, and the relationships will be established as a taxonomy.

Methods

Design - Scoping Review using PRISMA methodology, Data Sources from Engineering Village (Compendex and Inspec), PubMed, and OVID Embase (2019-2024). We introduce categories of DHIs and use cases. Records only relevant to the research question were included. We were interested in ontologies recorded in literature; screening was performed by independent reviewer(s). A further review was completed after this initial screening.

Results

Studies provide insight into the classification of DHIs, extending from workflow design, relationship analysis between common themes, and behavioural change intervention in mobile applications. The 41 included studies revealed four primary use case categories: diagnosis (n=12), detection/measurement (n=18), classification (n=7), and therapeutics (n=6), with some DHIs performing more than one use case. Four stakeholder groups were identified: patients, healthcare professionals, manufacturers, and regulators.

Conclusions

This method of hazard assessment, analysis and use of ontologies (DHI and HSC) improves the justification of safety claims and evidence articulation. DHI's incorporation of AI and other innovative technologies enhances the effectiveness of interventions. The operational aspects of implementing DHIs require consideration of the intended purpose, as well as a lack of experience or understanding of potential complex clinical decision support systems. We provide a classification of DHIs, extended from workflow design and behavioral change intervention, including use cases, resulting in a taxonomy of digital health technologies.

Keywords

Digital health intervention use cases taxonomy clinical decision support system

Introduction

The delivery of healthcare to patients utilizing technology can be defined in many ways. The industry in which healthcare technologies contribute to providing such care is one of the largest economies globally. Digital health is commonly used globally and encompasses many technologies, industries and academic and scientific disciplines. The World Health Organization (WHO) defines digital health as using digital technologies to improve health outcomes, efficiency and equity.¹ An extension of this term provides the concept of digital health intervention (DHI), a discrete function of digital health technology that offers a health objective.² These concepts are recognized globally and enable a better understanding of industry trends, advances in clinical practice and supporting research on enhancing clinical decision-support (CDS) systems.

To better understand the objectives of this study, we provide a clear and concise definition of the terms used and the associated problem statements under investigation. We also offer a summary of the medical device regulatory industry and regulation to aid in understanding the impact and challenges faced.

Purpose and scope of the document

This deliverable aims to review and categorize DHI use cases and applications. The literature review will provide a classification scheme for further insight into this research project.

Structure of this document

The remainder of this document is organized as follows.

Section “Digital health interventions as medical devices” introduces DHIs as medical devices, including terminologies essential to understanding the analysis undertaken, data analyzed and results. Section “Method” provides the method of analysis for the literature review.

Section “Results” details the analysis results and Section “Discussion” provides discussion and commentary on the findings. The classification of DHIs and use cases presented in graphical form and analysis is comprehensive.

Digital health interventions as medical devices

DHIs that provide a function of general wellness and lifestyle improvements are not typically subject to stringent regulatory requirements to place these products onto the market for general use. Those DHIs that function with specific medical purposes, such as diagnosing, treating or managing disease, are medical devices.³ Before being placed on the market and used, these products must meet stringent regulatory requirements to ensure their safety and security.

The International Medical Device Regulators Forum (IMDRF) promotes regulatory convergence in the global medical device industry.³ Many of the strategic aims of the IMDRF are to guide its members in efficient and effective regulatory processes. These strategies provide a consistent approach to regulatory requirements regardless of the country where the products are placed. The European Union (EU) regulates medical devices through the Medical Device Regulation (MDR). There are two regulations, EU MDR 2017/745 and EU MDR 2017/746. These regulations cover medical devices (MD) and in vitro diagnostic devices (IVD). The main difference between these two device types is that the medical devices are used to diagnose, prevent, monitor, treat, or alleviate disease or injury. IVDs are used to perform tests on samples taken from the human body to provide information about a person's health status.

DHIs that diagnose, monitor, treat, or alleviate disease or injury are medical devices. Similarly, testing on samples taken from the human body for health status is an IVD. The definition of a medical device leads to the description of use cases of diagnosis, monitoring and treatment, which are fundamental to this study.

Meeting the regulatory requirements and complying with legislation requires a consistent approach to development by the manufacturer of such medical devices. One characteristic of the regulatory requirements is the concept of harmonized standards. These international standards consider relevant parts of medical device requirements pertinent to the subject under consideration. For example, safety, quality and risk management, usability/human factors, software design and development lifecycle, information security and cybersecurity. These terms may be considered domains to understand their role in certification and meeting legislative requirements.

Any medical device manufacturer's objective is to place products on the market that are safe to use. The safety of the medical device is underpinned by the security of design (vulnerabilities are known and risks mitigated). The domain of usability and alignment with standards in this area ensures the product is developed with human behavior, abilities and limitations incorporated into the design. We define the terms to provide a shared understanding of our research question approach.

Safety = Freedom from unacceptable risk⁴

Security = Resistance to intentional or unauthorized act(s) designed to cause harm or damage to a system⁵

Usability = Human factors engineering application of knowledge about human behavior, abilities, limitations and other characteristics to the design of medical devices (including software), systems and tasks to achieve adequate usability.⁶ Achieving adequate USABILITY can result in acceptable RISK related to use

Risk = A combination of the probability of occurrence of harm and the severity of that harm⁴

Harm = The injury or damage to the health of people or damage to property or the environment⁴

Complying with the medical device regulations requires meeting the legislative requirements with supporting evidence. The evidence is reviewed independently by organizations (notified bodies) with devolved powers provided by the medical device regulators. Assurance activities throughout the development lifecycle of the DHI, from early-stage prototype to final marketable product, utilize this evidence-based approach to compliance.

The risk posed to patients and users by the functionality of these products is classified from low to high. In some regulatory markets, the classification is Class I to III, with I being the lowest risk. In Europe, the classification is Class I, IIa, IIb and III, with Class I being the lowest risk.

The higher the risk classification, the greater the regulatory burden on the manufacturer to provide assurance evidence when placing a safe product on the market.

Software as a medical device utilizes additional classifications of Class A, B and C if the manufacturer adopts one of the harmonized international standards aligned to the software development lifecycle (IEC 62304 Medical device software—Software life cycle processes).⁷

With all classes from market placement to software development, the effort to evidence regulatory compliance depends on risk. Higher risk naturally requires additional evidence and greater scrutiny from an assurance perspective.

Medical device registration

Once compliance and certification are achieved, the medical device can be placed on the market. This involves registration using Global Medical Device Nomenclature (GMDN) or European Medical Device Nomenclature (EMDN) codes. These codes are intended to reflect the medical purpose of the device and provide uniform classification and designation of medical devices.

GMDN and EMDN nomenclatures are similar; they are used to register products in the market. They are focused on product identification or categorization after completing the compliance activities. The classification schemes indicate that the intended purpose is incorporated into the coding structure. However, this is not consistently applied to software products such as medical devices and has only post-market certification objectives for safety incident reporting.⁸ One challenge we face is the disconnect between medical device class, product classification, and further software development classes that may or may not be applied to the device's development.⁹ Classification consistency can directly impact regulatory efforts at many other points in the device's lifecycle, including premarket assurance activities. Knowing the category of product that applies to the manufacturer, and an indication of its intended use, provides an opportunity to standardize assurance approaches for similar products. This benefits regulatory certification and compliance processes positively by creating efficiency, standardization and improvement in the classification schema. These are opportunities to meet the challenges regulators currently face.¹⁰

Artificial intelligence and clinical effectiveness

Artificial intelligence (AI) in digital health involves using complex dynamic algorithms to analyze data. Through AI, DHIs can improve diagnosis and accuracy, predict patient outcomes, personalize treatment plans and improve healthcare delivery. This technology can process vast amounts of data quickly, identify patterns and provide insights that might need to be noticed by human analysis alone. For this study, we use the term machine learning (ML) as a type of AI methodology.¹¹

We know that medical device product development utilizes product and software risk classes, which help to define the strength of evidence required to demonstrate safety. Current regulatory methods are not standardized, leading to variations in the application of assurance and validation techniques.¹² This problem is linked to the efforts made regarding medical device product nomenclatures (e.g. GMDN and EMDN).

In addition to the technical verification, validating such devices’ clinical effectiveness and appropriateness will require testing them against datasets representing the targeted population. However, this task is challenging, as datasets cannot be shared with manufacturers due to privacy concerns.¹³ Even when sharing is possible, the process may take several months. The problem is exacerbated by (a) the fact that datasets need to be produced specifically for each application, depending on its scope, and (b) the increasing volume of DHIs requiring validation. If a manufacturer has access to their datasets, it is easier to perform validation by a third party (e.g. a regulator) with a common dataset. Understanding aspects of the application training is crucial to provide safety assurance of acceptable use of such applications. This may include considering such problems as clinical relevance and specificity of data, the pedigree of data and their evidence base, alignment with the guidelines, data quality (e.g. accuracy and timeliness) and potential biases. This also has implications in the certification process, as each application is often non-standardized, and the stakeholders need more guidance on how to approach assurance, frequently resorting to their internal policy, which needs more EU, professional and international guidance. As a result, this wave of innovation cannot be effectively harnessed, with applications being deployed to patients. It may deprive patients of healthcare improvements and extension of optimum outcomes.

Method

Search strategy details and study criteria

The full search terms used are provided below:

(mHealth OR “mobile health” OR health OR “health software” OR “medical device”)

AND

(Classification OR terminology OR ontology)

AND

(“Use case” OR “intended purpose”)

Database search conducted on 10/05/2024.

Date restrictions: 2019–2024 (5-year limit as stated, with provision to extend to 10 years if needed)

Language: English only.

Records must be published in English to be included in the review. The Engineering Village databases, including Compendex and Inspec, PubMed biomedical and life sciences literature, and OVID Embase scientific, medical and healthcare databases, will be used to include articles from the past five years to the present. This will be extended to 10 years if unsatisfactory results are found (low volume of records returned compared to other reviews in this field).

Screening process

Initial screening: Two independent reviewers (SI and HW) screened titles and abstracts using predefined inclusion/exclusion criteria.

Full-text review: Potentially relevant articles underwent full-text review by the same reviewers.

Disagreement resolution: Conflicts were resolved through discussion between reviewers, with a third reviewer (GE) consulted when consensus could not be reached.

Inclusion criteria: Articles discussing DHIs and use cases, published in English, published 2019–2024, focus on ontologies and classification systems.

Exclusion criteria: Articles not focused on DHI use cases, non-English publications, outside date range, conference abstracts without full text.

Criteria for study inclusion/exclusion

Records must be relevant to the research question for inclusion and selection—articles discussing DHIs and their use cases are critical. We were particularly interested in ontologies recorded in the research literature. Study titles and abstracts have been reviewed, and screening has been performed by the independent reviewer(s). Following this initial screening, a further review of the full articles has been completed. A pictorial representation of the review and selection process has been provided in the review using the PRISMA flow diagram, shown in Figure 1.

Figure 1.

PRISMA diagram.

Data extraction, collection and analysis

Data was extracted using a standardized form developed by the research team and piloted on five studies. The following data were systematically extracted: study characteristics (author, year, country, study type), DHI characteristics (type, technology used, terminology/standards), use cases and intended purposes, stakeholder involvement, patient/clinical outcomes, safety, security and usability concerns, regulatory considerations.

The taxonomy was developed through an iterative process: initial categorization based on regulatory definitions, thematic analysis of extracted use cases, stakeholder mapping based on identified roles, relationship mapping between categories and expert review and refinement.

A narrative synthesis approach was employed, supplemented by: frequency analysis of use case categories, stakeholder role mapping, thematic analysis of safety/security concerns, development of visual taxonomy representations, cross-referencing with existing regulatory frameworks (GMDN, EMDN, medical device regulations).

Records from all three databases were extracted in raw format, imported into a recognized reference manager tool, checked for duplicates and then exported into a Microsoft Excel spreadsheet pro forma. Each record was downloaded and reviewed based on the initial inclusion criteria of the DHI subject theme. This provided a set of excluded and included records from which a more detailed review was undertaken. The analysis was completed, and the pro forma contents are highlighted in Table 1.

Table 1.

Data collection definitions.

Data collected	Definition
Digital Health Intervention	A discrete function of digital health technology, providing a health objective
Technology	A particular method by which science is used for practical purposes
Terminology of use case/standards	Applicable standards and terminologies used to define the use case
Use Case or Common Functions	Usage scenario for the DHI
Level of concern (safety, security and usability)	Summary of the further work, concerns and issues in the context of safety, security and usability
Stakeholders	Individuals, groups, or organizations that have an interest or role in the services provided and their outcomes
Patient/clinical outcome	Performance outcome measures that assess how a DHI functions.
Intended Use or Purpose	The use for which the device is intended according to the data supplied by the manufacturer on the labeling, instructions and promotional materials
Type of article reported in the literature	The different types and formats of articles presented in the research
Country	Origin of article

Results

Paper identification

Using the search method discussed, 1776 papers were identified. Of these, 1735 papers were excluded and 41 were included for review. Duplicates were removed when the three database sources were combined.

The records were found to vary between the following types of articles:

Interrater reliability study (1), journal articles (21), mixed methods study (1), narrative review (1), pre-post study (1), qualitative study (1), scoping review (4) and systematic review (11).

PRISMA paper extraction

The PRISMA flow diagram below, in Figure 1, shows the results of the systematic review:

Publication of articles over time

From the records identified, papers reviewed were published between 2019 and 2024. Articles published in 2019 = 3, 2020 = 7, 2021 = 7, 2022 = 12, 2023 = 10 and 2024 = 2.

Country of study

The records used for this review originate from 22 countries, as outlined in Figure 2. The United States provided the most articles, followed by the UK, Spain and Germany, which each provided three.

Figure 2.

Country of study.

Included studies and research focus (digital health intervention)

From the included studies, we present in Table 2 the DHI-led research focus, technology in use, terminology of the use case, or standards in use, and documented level of concern from the study. Where possible, the technology and terminology of the use case have been reported in full. Missing data items not presented in the study are explained as N/A (not available). An examination of this in full is presented in Appendix.

Table 2.

Results—digital health intervention, research focus and level of concern.

DHI category	Ref.	Key technologies	Common use cases	Primary concerns
Mobile Applications	¹⁴	Mobile health/mHealth	General health apps	Ambiguous classifications, safety claims, usability definitions
Clinical Decision Support	¹⁵	Electronic Health Record	Diagnosis	Clinical decision support accuracy, identification of diagnostic gaps
Mental Health Intervention	¹⁶	Mobile health/mHealth	Secondary intervention, therapy, fall risk assessment	Reliability of passive data collection, missing contextual data
Classification Systems	¹⁷	Web-based software	Quality of classification	Harmonization of anatomical concepts, classification safety
Medical Imaging AI	¹⁸	Radiology AI software, PACS	Image acquisition, interpretation and reporting	Detection sensitivity/specificity, workflow impact, user experience
Medical Device Coding	¹⁹	Dataset management	Device identification	Safety analysis, post-market surveillance of high-risk devices
Disease Diagnosis	²⁰	IoT, AI software	Communicable disease diagnosis, clinical decision support	Accuracy of diagnosis and reliability
Behavioral Change	²¹	Mobile health/mHealth	Clinical decision support, digital behavioral intervention	Utilization in workflow, intervention design clarity
Professional Guidance	²²	Mobile health/mHealth, eMental Health	Professional guidance for mental health	Patient safety issues, absence of professional contact
Assistive Technology	²³	Mobile health/mHealth, IoT	Clinical decision support, assistive technology	Safety issues with classifier selection in technology adoption
Medical Device Classification	²⁴	Medical devices	Device categorization and terminology	Inconsistent terminologies leading to confused safety incident reports
Activity Recognition	²⁵	IoT, AI software	Human activity recognition	System accuracy, data processing speed, device usability
Paediatric Therapy	²⁶	Mobile health/mHealth	Paediatric speech therapy, caregiving	Lack of practical application and regulatory compliance
Cancer Risk Assessment	²⁷	AI software, web-based	Hereditary cancer clinical practice guidelines	Risk classification improvement, computational efficiency
Social Care Monitoring	²⁸	Mobile health/mHealth	Monitoring and alerting for social care	Quality, detection and safety improvements through mobile applications
Teledermatology	²⁹	Health software, medical device	Clinical decision support, teledermatology	Safety of diagnosis, AI outperforming humans
Clinical Decision Support	³⁰	Health software	Clinical decision support	Increased safety and improved decision-making quality
mHealth Clinical Support	³¹	Mobile health/mHealth	mHealth clinical decision support	Increased safety and improved decision-making through experience-based CDS
Physical Activity Intervention	³²	IoT, mobile health/mHealth	Mobile health, telehealth, physical activity, telemedicine	Challenges of sustaining patient involvement and user engagement
Taxonomies	³³	N/A	Taxonomies	N/A

Technology distribution analysis shows mobile health technologies are reported in 25 studies, AI software in 8, web-based solutions in 10 and Internet of Things in 4 studies. Primary concerns are the safety of the DHI (14 studies), usability (5 studies) and accuracy (4 studies).

Stakeholder mapping

We considered all stakeholders involved in using each DHI within the studies reviewed. The roles within each stakeholder grouping are presented in Table 3 and are organized to highlight the commonalities among users as presented in the literature. The stakeholders identified in Table 3 provide context when examining the relationships between the four primary stakeholder groups—Manufacturer, Regulator, Patient and Healthcare Professional (incorporating Health Organization, Administrator, Care Giver and Local Authorities).

Table 3.

Stakeholder relationship mapping.

Relationship	Manufacturer	Healthcare ProfessionalHealth OrganizationAdministratorCaregiverLocal Authorities	Regulator
Patient	Feedback—forums, app stores and customer support.Usability feedback.Post-market surveillance activities.Safety incident reporting.	Prescribing recommendation.Patient education and guidance.Clinical oversight.Professional endorsement.	Reporting of adverse events.Clinical trials.Public feedback (patient advocacy groups).
Regulator	Requirements for regulatory compliance.Demonstration of the DHI's safety.Regulator market approval.Harmonized standards and legislative requirements development.Post-market surveillance.	Action of clinical studies for regulatory evidence. Professional and implementation guidance.Learning from safety incidents.
Healthcare ProfessionalHealth OrganizationAdministratorCaregiverLocal Authorities	Clinical requirements.Workflow integration.Usability feedback.Training, support and guidance.Real word evidence.

The Manufacturer to Regulator relationship provides context regarding the requirements for regulatory compliance and the demonstration of the DHI's safety. Regulators provide approval of DHI compliance based on evidence to permit products to be placed on the market and monitored for their use. Both stakeholders work within harmonized standards and legislative requirements to demonstrate the compliance of the DHI. At such a time when the DHI is placed onto the market for use, this enables a feedback loop of post-market surveillance to be implemented to provide a mechanism for safety incidents to be reported as and when they occur and assessed periodically.

The relationship between the Manufacturer to Healthcare Professional and others encompasses several distinct actions. Clinical requirements, workflow integration considerations and usability feedback are all provided to the manufacturer during DHI development. A manufacturer may provide training, support and guidance to implement and use the DHIs within the setting in which they are used. Healthcare professionals and others generate real-world evidence-based information on usage, clinical outcomes and safety reports to inform future developments or safety incident analysis.

The relationship between Healthcare Professionals and other patients can be summarized in several ways. The prescribing, recommendation and integration of DHIs into the patient care pathways is a key action. Patients can receive education and guidance on the use and interpretation of outputs from the DHIs. Clinical oversight concerns outcome assessment, management of treatment plans and often professional endorsement.

The Patient to Manufacturer relationship includes direct feedback from a DHI user through forums, app stores and customer support. Patient groups can provide usability feedback during the development of a DHI in prototyping stages. Manufacturers utilize the data collected for regulatory post-market surveillance activities. Finally, safety reporting allows a patient to report adverse DHI events to manufacturers directly.

The Patient to Regulator relationship also includes the reporting of adverse events concerning the use of the DHI within the context of market analysis, incident detection and learning. Clinical trials often incorporate patient feedback through clinical studies for regulatory approval and certification of the DHI. Regulators usually seek public feedback (patient advocacy groups) and consultation on standards, subjects of interest, influencing policy and safety awareness as part of more proactive safety measures.

The Healthcare Professional and others-Regulator relationship concerns the provision of and action of clinical studies for regulatory evidence. Professional and implementation guidance, new use cases and learning from safety incidents are also key actions within this context.

Whilst individual relationships are common, there are multi-stakeholder interactions that impact the quality and safety ecosystem of DHIs, evidence generation, innovation and development lifecycle. This stakeholder mapping demonstrates that the safety of DHIs requires technical innovation and the effective interaction of these relationships throughout the product lifecycle.

Use case, common functions and distribution analysis

The use cases defined in each study were grouped into the standard functions of Diagnosis, Therapeutics, Detection/Measurement and Classification. The classification of DHI from the WHO framework was aligned with each study, together with the technology utilized and the standard functions of the DHI. The use case analysis can be seen in Figures 3 to 6. Figure 3 shows the WHO DHI classification distribution and highlights the eight categories with counts in the literature from the scoping review. Use Case, Common Functions and relationship analysis can be shown in Figure 4. The Use case distribution analysis in Figure 4 reported eight use case patterns, with multi-function DHIs the most common reported in literature at 31%. The Relationship analysis displayed in Figure 5, across the use cases and WHO classification categories, reveals clear patterns indicating that CDS and multi-functional DHIs are the most common. Healthcare Provider Decision Support + multi-function DHI accounts for the most significant percentage of studies examined, at 35.7%. The Technology distribution analysis shown in Figure 6, considers the reported DHI, with most involved mobile health (mHealth) in 31% of the studies analyzed. The functions represented in literature are common legislative terms within medical device regulations and standards.

Figure 3.

WHO DHI classification distribution.

Figure 4.

Use case distribution.

Figure 5.

Use case, common functions and relationship analysis.

Figure 6.

Technology distribution analysis.

Intended use and patient/clinical outcome

To reaffirm the use case and standard functions, we examine the intended use and outcome (patient or clinical) for each DHI under review, shown in Table 4. The focus of the study, along with its objectives, accounts for the missing data. We can see from the results that the use cases of Therapeutic, Detection/Measurement, Diagnosis, and Classification are common. This is as expected. Similar patient/clinical outcomes have been removed as they duplicate the confirmed outcome.

Table 4.

Intended use and patient/clinical outcome.

Patient/clinical outcome	Intended use or purpose	Use case
Improvement in diagnosisAccuracy of detection from patient data and algorithm.	Diagnosis of a periodontal dental condition	DiagnosisDetection/Measurement
Reliable passive and active data collection, possibly involving local (on-phone) analysis of the collected data to provide timely interventions and/or have high expectations to bring them into clinical practice.	SyMptOMS is a smartphone application for the delivery of psycho-educational location-based notifications for the treatment of mental disorders associated with specific places.Mental disorders—Panic Disorder, agoraphobia and gambling disorder.TUG Test is a fully instrumented version of the well-known Timed Up and Go (TUG) test to assess a person's mobility, implemented through a smartphone and a wearable application.SyMptOMS-ET, a rewrite, and extension of the original SyMptOMS application, will automate place-based emotional exposure therapy using timely assessments and a rule-based ecological momentary intervention.	Detection/Measurement
Improved diagnosisEfficiency of operation	N/A	Diagnosis
Improved diagnosisEfficiency of operation	Detection, classification and measurement of pulmonary nodules in computed tomography scans of the chest	Diagnosis
Improved safety alertsDevice identification	N/A	Detection/Measurement
Improved diagnosisEfficiency of operation	Clinical Decision Support	Diagnosis
Improved suitability of application to patient needs	Alleviation of fatigue in cancer patientsSleep improvement, stress/anxiety reduction,Mood improvement, physical activity increase	Therapeutic
Improved accuracy of diagnosis, earlier detection	Enable healthcare providers to recognize and diagnose AKI accurately	Diagnosis
Potential for automated decision-making using FCM methods	LEAVES is a self-help online intervention designed to soothe the mourning process of older adults who have lost their spouses.	Classification
N/A	Knowledge-based approach to assistive technology provision and adoption	Therapeutic
Standardized clinical outcome and quality improvementStandardized Risk classificationSupporting clinical assessments	Diagnosis and patient management	DiagnosisDetection/Measurement
Real-time elderly activity classificationElderly caregivers can track the location of older people on an online map with a route guidance system to the elderly.	An activity tracking system for older people using machine learning can perform tracking in both indoor and outdoor environments.	ClassificationTherapeuticDetection/Measurement
N/A	Framework & conceptual model aimed at guiding specialists in adopting existing software platforms or building new ones—health intervention applications (mHealth)	TherapeuticDetection/Measurement
Standardized clinical outcome and quality improvementRisk classificationSupporting clinical assessments	Heredity Cancer identification—risk classification and identification for further clinical diagnosis and consultation	Classification
Standardized clinical outcome and quality improvementRisk classificationSupporting clinical assessments	Diagnosis and patient management	DiagnosisClassification
Standardized clinical outcome and quality improvementRisk classificationSupporting clinical assessments	Diagnosis and patient management	Classification
InformativePreventiveDiagnostic	Informative, preventative and diagnostic cancer application	DiagnosisDetection/Measurement
Standardized clinical outcome and quality improvementRisk classificationSupporting clinical assessments	Management of patient careMonitoring and alerting	Detection/MeasurementDiagnosis
Standardized clinical outcome and quality improvementRisk classificationSupporting clinical assessments	Management of patient careMonitoring and alerting	ClassificationDetection/Measurement
Standardized clinical outcome and quality improvementRisk classificationSupporting clinical assessments	Diagnosis and patient management	DiagnosisDetection/Measurement
Risk classificationPatient monitoring	Diagnosis and patient management	DiagnosisDetection/Measurement
Patient monitoring	Patient management	Detection/Measurement
Patient monitoring	Self-assessment of symptoms	Therapeutic
Prevention (disease, test ordering)Therapeutics (prescribing, vaccination)Management of chronic disease(s)	Clinical Decision Support	TherapeuticDetection/Measurement
Patient monitoring	Patient management	Detection/Measurement
Patient monitoring	Remote monitoring of patients	Detection/Measurement
Patient monitoringHealthcare professional support	To capture palliative symptoms and concerns,To assist health professionals in matching the types and levels of need experienced by patients and their caregivers with the most appropriate person or service to address identified needs.To help general practitioners in the identification of patients with CHF, COPD, or cancer in need of palliative care.To assess the needs of patients with CHF on four dimensions: physical, psychological, social and existential and to assist in their care planning and monitor the efficacy and impact of interventions.To measure care-related quality of life in patients diagnosed with chronic heart failure.To inform CHF-specific assessments and care.To identify persons in advanced-terminal disease state and requiring palliative care in health and social services.	Detection/MeasurementDiagnosisClassification
Patient Monitoring	End-of-life communication	Detection/Measurement
Patient monitoringHealthcare professional support	Nursing care	Detection/Measurement

Discussion

Quality improvement of DHIs

The benefits of utilizing complex algorithmic decision-making have been reported within the review, as well as in our introduction and research objectives.¹⁴ Confirmation of manual processing of interventions being improved by these technologies. Quality improvements are noted, in particular, by the recommendation from the Canadian Association of Radiologists—Developers should define the healthcare needs addressed by their software (e.g. quality of radiological care, healthcare network integration, patient perspective, employee experience, efficiency and costs), the applicable clinical domain (e.g. mammography and chest radiograph) and task (e.g. breast density evaluation and pneumothorax detection) and describe the expected positive impact on these needs and patient outcomes.¹⁵ These additional factors reported in the literature provide a compelling argument that DHIs incorporating AI and more innovative technologies improve interventions. The operational aspects of implementing DHIs require consideration of the intended purpose, and a lack of experience or understanding of potential complex CDS systems.¹⁶ Frameworks focusing on the design principles of reusability, reliability, privacy, versatility and testability provide the infrastructure to develop high-quality mobile health applications.¹⁷ These principles underpin safety requirements.

Classification of digital health interventions

Our findings align with and extend previous classification efforts. The WHO Digital Health Intervention Classification¹ provides a broad categorization framework but lacks the detailed use case specificity identified in our review. Similarly, the NICE Evidence Standards Framework¹⁴ focuses on evidence requirements and lacks the taxonomic structure our study develops. Our identification of four primary use cases (Diagnosis, Detection/Measurement, Classification, Therapeutics) aligns with medical device regulatory categories. It provides more granular DHI-specific detail than existing frameworks like GMDN coding systems.

The stakeholder mapping (patients, healthcare professionals, manufacturers, regulators) reflects the multi-stakeholder nature of digital health ecosystems identified in previous work by [relevant citations]. Still, our study uniquely maps these relationships to specific use cases.

From the studies reviewed, we find helpful insight into the classification of DHIs, extended from workflow design and behavioral change intervention of mobile applications.¹⁸ Within this are intervention descriptors, use case definitions and stakeholders. The hierarchy of digital technology is assessed to enhance tuberculosis control,¹⁹ collaborates with the hierarchy of AI-based technologies in the previously discussed behavioral change study. The terminology used to assess DHIs utilized during the COVID-19 pandemic (Digital Health MeSH Descriptor Data 2024) also provides a consolidated yet not fully elaborated view of terms used that synergize with classification structures.²⁰

Terminologies and definitions are critical when defining any form of taxonomy or ontology. Our research has included many different DHIs using the following definitions:

Digital Health—Systematic application of information and communications technologies, computer science, and data to support informed decision-making by individuals, the health workforce and health systems to strengthen resilience to disease and improve health and wellness.¹

Digital Health Intervention—a discrete functionality of digital technology that is applied to achieve health objectives.²

Health System Challenge—A generic need or gap that reduces the optimal implementation of health services.²

Intended Use or Purpose—The use for which the device is intended according to the data supplied by the manufacturer.³

Use Cases have been defined for Diagnosis, Detection/Measurement, Classification and Therapeutics. Stakeholders are categorized as Manufacturers, Healthcare Professionals, Patients and Regulators. Both categories are derived from this study.

The taxonomy mapping of the data available through our study is provided in Figure 7. This highlights the relationship to the broader term of digital health and follows international standards, industry guidelines and best practice.^21–25

Figure 7.

Taxonomy of digital health technologies.

Focusing on use cases, stakeholders and health system challenges, we examine the features in Figure 8, between these terms, to demonstrate the one-to-many and many-to-many relationships.

Figure 8.

Use cases, stakeholders and DHI taxonomy.

Use case and intended purpose

The intended use or purpose of a medical device is a defined term used to describe the use of the product.^26–28 The stakeholder analysis from this study provides four personas of technology users: patient, healthcare professional, manufacturer and regulator. These personas can be used in conjunction with their intended use and use case, as shown in Figure 8. This relationship and taxonomy are a precursor to the classification scheme.

Harmonization of ICF Body Structures and ICD-11 Anatomic Detail improves the quality of classification of DHIs. There is potential to align this classification scheme with the intended use/purpose and instructions for use statements for medical devices.²⁹ Using structured languages and natural language processing allows for automation in device classification and many parts of the regulatory certification process.

Medical device nomenclature and coding schemes

Medical device nomenclatures are key to product placement and post-market analysis. These coding structures can be linked to the taxonomy offered in this study by using the product term name and definition^9,10 to align with the digital technology term used in this study. The quality of medical device coding post-certification (in use) is not consistent in both the terminology and post-market incident reporting.³⁰ Adverse event reporting is expected to be simpler to analyze due to the coding schemas offered to manufacturers. Efforts to consolidate coding schemas and provide improved clarity on intended use versus the type of medical device certified would be beneficial.

Policy and practical implications

As illustrated in Figure 8, the taxonomy developed has the potential to support more consistent DHI classification across regulatory legislation, standards and address the inconsistencies in medical device nomenclature frameworks. This provides the advantages of improving regulatory harmonization. From the software development lifecycle (focus of the scoping review), manufacturers could utilize the framework better to align DHI development with the requirements for regulatory and use case objectives. Healthcare professionals and included stakeholders could benefit from clarity in the classification to support the selection, implementation and adoption of DHIs into clinical workflows.

From the perspective of evidence generation to support regulatory requirements, the framework could guide stakeholders to the appropriate level of evidence for the DHI classification, based on risk and safety classification. Finally, from an interoperability perspective, standardization in classification would facilitate improvements in health system integration with extensions of functionality consistent with information exchange.

Limitations and further work

This scoping review provides a preliminary framework for classifying DHIs based on use cases, stakeholder involvement and regulatory considerations. While the proposed taxonomy offers a structured approach to understanding DHI complexity, several limitations should be acknowledged. This includes considerations on the interpretation of the results. While comprehensive, the search was limited to three databases and may have missed relevant studies in specialized repositories. The inclusion of English-only publications may have excluded relevant international perspectives. The scoping review focused on published literature and may have missed grey literature or unpublished findings. Temporal Limitations: The 5-year search window, while capturing recent developments, may have excluded foundational work in DHI classification. The diverse nature of the included studies (different methodologies, DHI types and outcomes) limited quantitative synthesis. The proposed taxonomy has not been externally validated with stakeholder groups. The classification system, derived from 41 studies primarily from high-income countries, may not fully represent the global DHI landscape. The framework requires validation with broader stakeholder groups and testing across different healthcare contexts before widespread implementation. The identification of four primary use cases (Diagnosis, Detection/Measurement, Classification, Therapeutics) and four stakeholder categories provides a foundation for future classification efforts. Still, the relationships between these categories need further empirical validation. The method of hazard assessment, analysis and use of ontologies shows promise for improving safety claims and evidence articulation in DHIs; additional research would establish the effectiveness across different DHI categories and regulatory contexts. Further research to develop the effectiveness of the proposed taxonomy through external validation, consideration of extending to other DHI areas such as social and community care, integration with regulatory frameworks and the development of assessment/assurance tools based on the classification taxonomy.

Conclusion

The scoping review has enabled a classification and taxonomy for DHIs to be developed. The development has incorporated use cases, stakeholders and regulatory considerations, which to date have not been published in literature, nor has the relationship between these terms been explored. The systematic analysis of 41 studies identified four primary use cases that align with medical device regulatory requirements and terminologies: Diagnosis (n = 12), Detection/Measurement (n = 18), Classification (n = 7) and Therapeutics (n = 6).

The taxonomy offered in this scoping review addresses gaps in the classification of DHIs by the provision of the following:

Standard Use Case Categories: representing DHI functions that align with regulatory terminology and intended use

Stakeholder Mapping: four groups identified (patients, healthcare professionals, manufacturers and regulators), including the relationships between each stakeholder

Regulatory standards alignment: The review provides insight into a framework to connect medical device classification with use cases and health system challenges.

The scoping review shows that the most common category of DHIs is those that are multi-functional (31% of studies), from the literature. This confirms the evolution of health software towards more integrated care solutions and services. It should not be understated that the predominance of mobile health technologies represents an ever-growing emphasis on patient-centred care and accessible DHIs.

The practical implications for stakeholders are many. Manufacturers may use the taxonomy for guiding through often complex regulatory requirements, thereby enabling an efficient, effective and consistent pathway to market beyond safety claims being made. For Healthcare professionals, the increased clarity on DHI selection, implementation and alignment with clinical workflows provides an additional advantage from a quality, healthcare planning and improvement in patient outcomes perspective. Finally, regulators may utilize the taxonomy and classification alignment to support consistency in the classification and evidence-based assessment of DHIs.

The scoping review is not without limitations. External validation with additional stakeholders, consideration of other healthcare systems beyond the initial WHO classification scheme (for example, social and community care), before implementation. Further work will focus on the validation of the classification scheme and extending it into social and community care. This validation will enable the development of a more practical assessment-based tool and process based on the taxonomy. This research contributes to a broader objective of enhancing DHI innovation by providing a structured approach (harmonized and evidence-based) to the complexity of DHI use case development within a challenging regulatory environment.

Footnotes

ORCID iDs

Stuart Harrison

Saif Ul Islam

Ethical approval and consent

Not required. The research did not involve human subjects and/or animals or case reports/case series.

Contributors

The method and results of this research were assessed by all authors listed,leading to a contribution to the final version of the manuscript. All authors contributed to the final version of the manuscript.

Funding

The authors disclosed receipt of the following financial support for the research,authorship,and/or publication of this article: This work is co-funded by the HORIZON.2.1—Health Programme of the European Commission,Grant Agreement number: 101095661—Innovative applications of assessment and assurance of data and synthetic data for regulatory decision support (INSAFEDARE).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Guarantor

Stuart Harrison.

References

World Health Organization . WHO guideline: recommendations on digital interventions for health system strengthening. Geneva, Switzerland: World Health Organization, 2019.

Mehl

Tamrat

Labrique

, et al. Classification of digital health interventions v 10. Geneva, Switzerland: World Health Organization, 2018.

Essential principles of safety and performance of medical devices and IVD medical devices. 2018.

ISO/IEC . ISO/IEC guide 63:2019 guide to the development and inclusion of aspects of safety in International Standards for medical devices. ISO, 2019.

ISO/IEC . ISO/IEC 23643:2020(en) software and systems engineering—capabilities of software safety and security verification tools. ISO, 2020.

IEC I . IEC 62366-1:2015/Amd 1:2020 medical devices part 1: application of usability engineering to medical devices. ISO, 2020.

Antonio

Coronato

. IEC 62304: Medical device software—software life-cycle processes. In: Engineering high quality medical software: regulations, standards, methodologies and tools for certification. Geneva, Switzerland: Institution of Engineering and Technology, 2018, pp.87–93.

Onitiu

Wachter

Mittelstadt

. How AI challenges the medical device regulation: patient safety, benefits, and intended uses. SSRN Electron J 2023; lsae007: 1–38.

Gibello

Ren

Caiani

. Development of an AI-based IT tool to support medical device nomenclature standardization for post-market surveillance by automated mapping from GMDN to EMDN standards. Eur Heart J 2023; 44: 1–6.

10.

GMDN . Medical device nomenclature—what next globally? https://doi.org/10.1093/eurheartj/ehad655.3024 (2024).

11.

Chang

. The role of artificial intelligence in digital health. Switzerland: Springer International Publishing, 20237585.

12.

Harrison

Maple

Epiphaniou

, et al. Improving safety claims in digital health interventions using the digital health assessment method. Digit Health 2024; 10: 20552076241258756.

13.

Bilici Ozyigit

Arvanitis

Despotou

. Generation of realistic synthetic validation healthcare datasets using generative adversarial networks. Stud Health Technol Inform 2020; 272: 322–325.

14.

Nwe

Larsen

Nelissen

, et al. Medical mobile app classification using the national institute for health and care excellence evidence standards framework for digital health technologies: interrater reliability study. J Med Internet Res 2020; 22: e17457.

15.

Tanguay

Acar

Fine

, et al. Assessment of radiology artificial intelligence software: a validation and evaluation framework. Can Assoc Radiol J 2023; 74: 326–333.

16.

Gaspar

Iturricha-Caceres

Macedo

, et al. The use of a medical application improves the diagnosis of acute kidney injury: a pre-post study. Front Med (Lausanne) 2022; 9: 817387.

17.

González-Pérez

Matey-Sanz

Granell

, et al. AwarNS: a framework for developing context-aware reactive mobile applications for health and mental health. J Biomed Inform 2023; 141: 104359.

18.

Lisowska

Wilk

Peleg

. SATO (IDEAS expAnded wiTh BCIO): workflow for designers of patient-centered mobile health behaviour change intervention applications. J Biomed Inform 2023; 138: 104276.

19.

Lee

Raviglione

Flahault

. Use of digital technology to enhance tuberculosis control: scoping review. J Med Internet Res 2020; 22: e15727.

20.

Ganjali

Eslami

Samimi

, et al. Clinical informatics solutions in COVID-19 pandemic: scoping literature review. Inform Med Unlocked 2022; 30: 100929.

21.

International Organization for Standardization . ISO/TR 11147:2023 health informatics—personalized digital health—digital therapeutics health software systems. ISO, 2023.

22.

Myler

. s3 connected health. https://www.s3connectedhealth.com/blog/digital-health-solutions-and-dtx-which-one-is-right-for-you (2022).

23.

Milaster

. https://ingeniumdigitalhealth.com/resources/taxonomy (2019).

24.

Rahul Awati

. What is digital health (digital healthcare)? https://www.techtarget.com/searchhealthit/definition/digital-health-digital-healthcare (2024).

25.

National Institute for Health and Care Excellence . Evidence standards framework for digital health technologies. 2022.

26.

U.S. Department of Health and Human Services . Guidance for industry on general/specific intended use. In: Health FaDACfDaR, editor. 1998.

27.

Gov UK . Guidance—Crafting an intended purpose in the context of software as a medical device (SaMD). In: MHRA, editor. 2023.

28.

Uniion E. Intended Purpose. 2024.

29.

Della Mea

Almborg

Martinuzzi

, et al. Harmonization of ICF body structures and ICD-11 anatomic detail: one foundation for multiple classifications. PLoS One 2023; 18: e0280106.

30.

Yagahara

Yokoi

. Terminology integration and inconsistency identification of adverse event terminology for Japanese medical devices using SPARQL. BMC Med Inform Decis Mak 2022; 22: 16.

31.

Tokede

Brandon

Lee

C-T

, et al. Development and validation of a rule-based algorithm to identify periodontal diagnosis using structured electronic health record data. J Clin Periodontol 2024; 51: 547–557.

32.

Ren

Bertoldi

Fraser

, et al. Validation of CORE-MD PMS support tool: a novel strategy for aggregating information from notices of failures to support medical devices’ post-market surveillance. Ther Innov Regul Sci 2023; 57: 589–602.

33.

Venkatachala Appa Swamy

Periyasamy

Thangavel

, et al. Design and development of IoT and deep ensemble learning based model for disease monitoring and prediction. Diagnostics (Basel) 2023; 13: 1942.

34.

Brandl

van Velsen

Brodbeck

, et al. Developing an eMental health monitoring module for older mourners using fuzzy cognitive maps. Digit Health 2023; 9: 20552076231183549.

35.

Ortiz-Barrios

Garcia-Constantino

Nugent

, et al. A novel integration of IF-DEMATEL and TOPSIS for the classifier selection problem in assistive technology adoption for people with dementia. Int J Environ Res Public Health 2022; 19: 1133.

36.

Muangprathub

Sriwichian

Wanichsombat

, et al. A novel elderly tracking system using machine learning to classify signals from mobile and wearable sensors. Int J Environ Res Public Health 2021; 18: 12652.

37.

Cunha

BCR

Rodrigues

Zaine

, et al. Experience sampling and programmed intervention method and system for planning, authoring, and deploying mobile health interventions: design and case reports. J Med Internet Res 2021; 23: e24278.

38.

Ritchie

Frey

Lamy

J-B

, et al. Automated clinical practice guideline recommendations for hereditary cancer risk using chatbots and ontologies: system description. JMIR Cancer 2022; 8: e29289.

39.

Silva

MGD

Sakata-So

Pereira

, et al. Mobile application of the terminology subset for coping with domestic violence against children. Rev Bras Enferm 2021; 74: e20200287.

40.

Haenssle

Fink

Toberer

, et al. Man against machine reloaded: performance of a market-approved convolutional neural network in classifying a broad spectrum of skin lesions in comparison with 96 dermatologists working under less artificial conditions. Ann Oncol 2020; 31: 137–143.

41.

Muro

Larburu

Torres

, et al., eds. Architecture for a multimodal and domain-independent clinical decision support system software development kit. In: Proceedings of the annual international conference of the IEEE engineering in medicine and biology society, EMBS, 2019: Institute of Electrical and Electronics Engineers Inc.

42.

Schaeffer

Ahmed

Rahman

, et al. Development and evaluation of a mobile application for case management of small and sick newborns in Bangladesh. BMC Med Inform Decis Mak 2019; 19: 116.

43.

Vajravelu

Arslanian

. Mobile health and telehealth interventions to increase physical activity in adolescents with obesity: a promising approach to engaging a hard-to-reach population. Curr Obes Rep 2021; 10: 444–452.

44.

Chaudhary

Kakkar

Jadav

, et al. A taxonomy on smart healthcare technologies: security framework, case study, and future directions. J Sens 2022; 2022: 1–30.

45.

Moreira

Pires

van Sinderen

, et al. SAREF4health: Towards IoT standard-based ontology-driven cardiac e-health systems. Appl Ontol 2020; 15: 385–410.

46.

Hupont

Gomez

, eds. Documenting use cases in the affective computing domain using Unified Modeling Language. In: 10th international conference on affective computing and intelligent interaction, ACII 2022, 18–21 October 2022. Nara, Japan: Institute of Electrical and Electronics Engineers Inc.

47.

Boyce

Nyangara

Kamunyori

. A mixed-methods quasi-experimental evaluation of a mobile health application and quality of care in the integrated community case management program in Malawi. J Glob Health 2019; 9: 010811.

48.

Darley

Coughlan

Furlong

. People with cancer and their family caregivers’ personal experience of using supportive eHealth technology: a narrative review. Eur J Oncol Nurs 2021; 54: 102030.

49.

Faric

Hinder

Williams

, et al. Early experiences of integrating an artificial intelligence-based diagnostic decision support system into radiology settings: a qualitative study. J Am Med Inform Assoc 2023; 31: 24–34.

50.

Lapp

Egan

McCann

, et al. Decision support tools in adult long-term care facilities: scoping review. J Med Internet Res 2022; 24: e39681.

51.

Ziebart

Kfrerer

Stanley

, et al. A digital-first health care approach to managing pandemics: scoping review of pandemic self-triage tools. J Med Internet Res 2023; 25: e40983.

52.

Kuo

Cho

Olaye

, et al. Understanding the technological landscape of home health aides: scoping literature review and a landscape analysis of existing mHealth apps. J Med Internet Res 2022; 24: e39997.

53.

Alhasan

Hasaneen

. Digital imaging, technologies and artificial intelligence applications during COVID-19 pandemic. Comput Med Imaging Graph 2021; 91: 101933.

54.

Villarreal

Berbey-Alvarez

. Evaluation of mHealth applications related to cardiovascular diseases: a systematic review. Acta Inform Med 2020; 28: 130–137.

55.

Amor-García

MÁ

Collado-Borrell

Escudero-Vilaplana

, et al. Assessing apps for patients with genitourinary tumors using the mobile application rating scale (MARS): systematic search in app stores and content analysis. JMIR mHealth uHealth 2020; 8: e17609.

56.

Keutzer

Wicha

Simonsson

. Mobile health apps for improvement of tuberculosis treatment: descriptive review. JMIR mHealth uHealth 2020; 8: e17246.

57.

Maaß

Freye

Pan

, et al. The definitions of health apps and medical apps from the perspective of public health and law: qualitative analysis of an interdisciplinary literature overview. JMIR mHealth uHealth 2022; 10: e37980.

58.

Meunier

Raynaud

Guimaraes

, et al. Barriers and facilitators to the use of clinical decision support systems in primary care: a mixed-methods systematic review. Ann Fam Med 2023; 21: 57–69.

59.

Chatterjee

Prinz

Riegler

, et al. A systematic review and knowledge mapping on ICT-based remote and automatic COVID-19 patient monitoring and care. BMC Health Serv Res 2023; 23: 1047.

60.

Ament

Couwenberg

Boyne

, et al. Tools to help healthcare professionals recognize palliative care needs in patients with advanced heart failure: a systematic review. Palliat Med 2021; 35: 45–58.

61.

Ryan

Connolly

Bradford

, et al. Interventions for interpersonal communication about end of life care between health practitioners and affected people. Cochrane Database Syst Rev 2022; 7: CD013116.

62.

Nieto Agraz

Pfingsthorn

Gliesche

, et al. A survey of robotic systems for nursing care. Front Robot AI 2022; 9: 832248.

Digital health interventions use cases: Classification and taxonomy development,a scoping review

Abstract

Objective

Methods

Results

Conclusions

Keywords

Introduction

Purpose and scope of the document

Structure of this document

Digital health interventions as medical devices

Medical device registration

Artificial intelligence and clinical effectiveness

Method

Search strategy details and study criteria

Screening process

Criteria for study inclusion/exclusion

Data extraction, collection and analysis

Results

Paper identification

PRISMA paper extraction

Publication of articles over time

Country of study

Included studies and research focus (digital health intervention)

Stakeholder mapping

Use case, common functions and distribution analysis

Intended use and patient/clinical outcome

Discussion

Quality improvement of DHIs

Classification of digital health interventions

Use case and intended purpose

Medical device nomenclature and coding schemes

Policy and practical implications

Limitations and further work

Conclusion

Footnotes

ORCID iDs

Ethical approval and consent

Contributors

Funding

Declaration of conflicting interests

Guarantor

References